Aiim Industry Watch: Content Analytics: automating processes and extracting knowledge

AIIM Market Intelligence
Delivering the priorities and opinions of AIIM’s 80,000 community
Content Analytics:
automating processes and extracting
knowledge
aiim.org l 301.587.8202
Industry
Watch
Underwritten in part by:

Industry
Watch
©2015 AIIM - The Global Community of Information Professionals 1
ContentAnalytics:automatingprocessesand
extractingknowledge
About the Research
As the non-profit association dedicated to nurturing, growing and supporting the information management
community, AIIM is proud to provide this research at no charge. In this way, the entire community can
leverage the education, thought leadership and direction provided by our work. We would like these
research findings to be as widely distributed as possible. Feel free to use individual elements of this
research in presentations and publications with the attribution – “© AIIM 2015, www.aiim.org”. Permission is
not given for other aggregators to host this report on their own website.
Rather than redistribute a copy of this report to your colleagues or clients, we would prefer that you direct
them to www.aiim.org/research for a download of their own.
Our ability to deliver such high-quality research is partially made possible by our underwriting companies,
without whom we would have to return to a paid subscription model. For that, we hope you will join us in
thanking our underwriters, who are:
Process Used and Survey Demographics
While we appreciate the support of these sponsors, we also greatly value our objectivity and independence
as a non-profit industry association. The results of the survey and the market commentary made in this
report are independent of any bias from the vendor community.
The survey was taken using a web-based tool by 238 individual members of the AIIM community between
April 17 2015, and May 08, 2015. Invitations to take the survey were sent via e-mail to a selection of the
80,000 AIIM community members.
Survey demographics can be found in Appendix 1. Graphs throughout the report exclude responses from
organizations with less than 10 employees taking the number of respondents to 222.
Swiss Post Solutions AG
Pfingstweidstrasse 60b
8080 Zürich
Switzerland
Email: global.sps@swisspost.com
Web: www.swisspostsolutions.com

Industry
Watch
extractingknowledge
About AIIM
AIIM has been an advocate and supporter of information professionals for 70 years. The association
mission is to ensure that information professionals understand the current and future challenges of
managing information assets in an era of social, mobile, cloud and big data. AIIM builds on a strong
heritage of research and member service. Today, AIIM is a global, non-profit organization that provides
independent research, education and certification programs to information professionals. AIIM represents
the entire information management community: practitioners, technology suppliers, integrators and
consultants.
About the Author
Doug Miles is head of the AIIM Market Intelligence Division. He has over 30 years’ experience of working
with users and vendors across a broad spectrum of IT applications. He was an early pioneer of document
management systems for business and engineering applications, and has produced many AIIM survey
reports on issues and drivers for Capture, ECM, Information Governance, SharePoint, Mobile, Cloud,
Social Business and Big Data. Doug has also worked closely with other enterprise-level IT systems such
as ERP, BI and CRM. Doug has an MSc in Communications Engineering and is a member of the IET in
the UK.
© 2015
AIIM The Global Community of Information Professionals
1100 Wayne Avenue, Suite 1100
Silver Spring, MD 20910
+1.301.587.8202
www.aiim.org

Industry
Watch
extractingknowledge
Table of Contents
About the Research
About the Research . . . . . . . . . . . . . . . . . . . . . . 1
Process Used and Survey Demographics . . . . . . . 1
About AIIM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
About the Author . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Introduction
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Key Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Drivers and Adoption
Drivers and Adoption . . . . . . . . . . . . . . . . . . . . . 5
Drivers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Importance and Leadership . . . . . . . . . . . . . . . . . . 7
Adoption and Applications . . . . . . . . . . . . . . . . . . . 7
Progress and Issues . . . . . . . . . . . . . . . . . . . . . . . 8
Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Process Automation and
Inbound Routing
Process Automation and Inbound Routing . . . 9
Automating Email Classification . . . . . . . . . . . . . 10
Project Success . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Information Governance and Metadata
Generation / Correction
Information Governance and Metadata
Generation / Correction . . . . . . . . . . . . . . . . . . 12
Project Success . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Legal Judgment . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Contextual Search, Curation, and
E-discovery
Contextual Search, Curation,
and E-discovery . . . . . . . . . . . . . . . . . . . . . . . . . 15
Metadata Creation/Correction . . . . . . . . . . . . . . . 15
E-discovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Curation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Analysis / Business Insight /
Customer Input
Analysis / Business Insight /
Customer Input . . . . . . . . . . . . . . . . . . . . . . . . . 17
Real-Time or Near-Time . . . . . . . . . . . . . . . . . . . 18
Social Media Monitoring . . . . . . . . . . . . . . . . . . . 19
Business Advantage . . . . . . . . . . . . . . . . . . . . . . 19
Progress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Big Content Projects
Big Content Projects . . . . . . . . . . . . . . . . . . . . . 20
ROI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Opinions
Opinions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Spend
Spend . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Conclusion and Recommendations
Conclusion and Recommendations . . . . . . . . 23
Recommendations . . . . . . . . . . . . . . . . . . . . . . . 23
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Appendix 1: Survey Demographics
Appendix 1: Survey Demographics . . . . . . . . 24
Survey Background . . . . . . . . . . . . . . . . . . . . . . . 24
Organizational Size . . . . . . . . . . . . . . . . . . . . . . . 24
Industry Sector . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Job Roles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Appendix 2: General Comments
Appendix 2: General Comments . . . . . . . . . . . 26
Do you have any general comments to
make about your content analytics projects?
(Selective) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
UNDERWRITTEN IN PART BY
UNDERWRITTEN IN PART BY . . . . . . . . . . . . . 27
Swiss Post Solutions AG . . . . . . . . . . . . . . . . . . . 27
AIIM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

Industry
Watch
extractingknowledge
Introduction
The capacity of computers to recognize meaning in text, sound or images has progressed slowly and
steadily over many years, but with the arrival of multi-processor cores, and the continual refinement
of software algorithms, we are in a position where both the speed and the accuracy of recognition can
support a wide range of applications. In particular, when we add analysis to recognition, we can match
up content with rules and policies, detect unusual behavior, spot patterns and trends, and infer emotions
and sentiments. Content analytics is a key part of “big data” business intelligence, but it is also driving
auto-classification, content remediation, security correction, adaptive case management, and operations
monitoring.
The first step for many analytic processes is capture and recognition – from paper, from emails, and from
other inbound channels. This in itself involves validation and some “intelligent guesswork” based on word
matching and sentence construct. Similar principles can be applied to search and knowledge extraction,
moving beyond simple keywords to contextual analysis, taking into account the significance and use of
the search terms.
Humans hate filing. Even more, they hate sifting content for deletion - and they are generally bad at
it. Computers are much more consistent in their application of rules, and given suitable criteria for
classification, or for deletion, can hugely reduce unwanted content. This improves the searchability and
business value of what remains, and also make-safe any sensitive content. Beyond this, we can use
meaningful extraction of comments, opinions, diagnoses, reports, claims, social chat, and so on, to gain
business insight, improve competitive advantage, or achieve fast response.
In this report we will look at the take-up of analytics applications for inbound routing and text recognition,
for content classification and metadata correction, for improved search and knowledge extraction, and to
provide business insight. We look at the success factors and outcomes, and the choices being made for
deployment.
Key Findings
n 73% of respondents agree that enhancing the value of legacy content is better than wholesale
deletion. 53% agree that auto-classification using content analytics is the only way to get content
chaos under control.
n 54% feel that their organization is exposed to considerable risk due to stored content that is not
correctly identified.
n 73% consider that there is real business insight to be gained if they can get the analytics right.
63% are being held back by a lack of analytic skills and an absence of allocated responsibilities.
n 34% of responding organizations are using content analytics for process automation,
information governance, contextual search or business insight. A further 44% have plans in
place.
n 17% consider content analytics to be “essential” now for their organization, growing to 59% in 5
years’ time. Plus 28% feeling it “is something we definitely need”.
n The biggest issues for adoption are lack of expertise (36%), and a need to set information
governance policies first (36%). 43% admit that their current capability in enterprise search is poor,
33% have problems with BI, and 19% have poor ECM.
Process Automation
n 15% are using OCR data capture of inbound content for process input, 14% are auto-classifying
content for archive, and 12% are auto-routing to specific processes or to case-files. 10% are
triggering processes from inbound content, including 5% from mobile device input.
n 5% have fully automated filing or archiving of inbound emails, and 11% user-prompted filing.
24% have plans in the next 12-18 months.

Industry
Watch
extractingknowledge
n Benefits from inbound analytics include faster flowing processes (50%), happier staff (32%) and
improved governance (20%). 18% are seeing high levels of “hands-off” processing.
Information Governance
n 20% are already using auto-classification to assist staff with filing, metadata tagging, or records
declaration, and 17% have immediate plans. 18% are using automated or batch agents to correct
metadata for improved searchability, to better align metadata between repositories, or to detect security
and compliance risks.
n Improved search is the biggest benefit of auto-classification (reported by 52%) along with better
staff productivity (40%), and improved compliance and governance (31%). Defensible deletion and
recovered storage space are also reported (19%).
Contextual Search and Curation
n Only 35% have contextual search, including 11% across multiple internal sources and 7% across
external sources. 8% rely heavily on their contextual e-discovery tools, although a further 10% have
them but don’t use them.
n 19% have some automated curation tools to create custom libraries and alerts, although 9% are
from internal sources only. 6% have manual curation processes. 59% have neither, but feel it would
be useful.
Business Insight
n 24% have at least one “big content” project for business insight, with 10% having several.
Improved product or service quality is the strongest objective, followed by core investigations and
research, and then detection of non-compliance.
n Nearly half have used in-house development, and 17% external custom. 27% have used cloud or
SaaS products and 27% products from their ECM vendor.
n 34% have achieved ROI in 12 months or less, and 68% in 18 months or less.
Spend
n Most of our respondents expect to spend more on content analytics in the next 12 months.
Strongest growth is in enhanced or contextual search, analytics for business insight, and automated
classification tools or modules.
Content analytics by its nature places demands on how content is stored and managed within the
business. Poorly cataloged content spread out across multiple repositories and file-shares, immature
information governance policies, and only basic search and BI tools will make knowledge extraction
difficult. This is an area where many of the content correction and re-classification tools that we discuss
later can help to improve these situations.
As we can see in Figure 1, 18% of our respondents rate their ECM capability as poor, although only 40%
consider it to be good or excellent. When it comes to records management and content retention, 30%
admit it is poor and only a third rate it as good or excellent. Business Intelligence (BI) and reporting is
a frequent cause for complaint from line-of-business managers in most organizations, and 33% of our
respondents would consider it to be poor. But the biggest shortcomings are in enterprise-wide search with
43% having poor capabilities, and only 20% in good shape.

Industry
Watch
extractingknowledge
Figure 1: How would you best characterize the following capabilities across
your organization? (N=222)
Against this background, it is understandable that many organizations may feel that content management
comes first, with content analytics further down the track. However, it may well be that these low ratings
come from poorly deployed or poorly used ECM and RM systems. This can be particularly true of many
SharePoint implementations1
. Automated classification, and content correction across existing content,
would be a good way to re-vitalize these failed or stalled projects.
Drivers
Process productivity, business insight, and adding value to legacy content take the top places when it comes
to key drivers. This is followed by improving the benefits and compliance of ECM/RM - by more consistent
declaration and classification of records. Reducing unidentified risk in what is termed “dark data” is important
for 25%, and this rises to 32% for the largest organizations. This refers to content which may contain
sensitive or personally identifiable information about customers or staff, or may have business sensitivity.
In a more general sense, 25% are keen to use content analytics to help them reduce overall storage
requirements, or to clean up content before migrating it to newer systems or consolidated repositories.
Figure 2: What would be the THREE biggest drivers for content analytics in
your organization? (N=217)
0% 10% 20% 30% 40% 50% 60% 70% 80% 90%100%
Document and content management (ECM)
Records management (RM) / content
retenƟon
Business intelligence (BI) / reporƟng
Enterprise-wide search
Excellent Good Fair Poor
0% 10% 20% 30% 40% 50% 60% 70%
Improving process producƟvity by removing
manual steps
Providing business insight
Adding value to our legacy content,
improving search
Improving the benefits/compliance of our
ECM/RM - staff are poor at classificaƟon
Freeing up process boƩlenecks and
overloads
Reducing unidenƟfied risk in our “dark data”
Reducing our storage/migraƟon
requirements in a defensible way
DetecƟng fraud, crime, policy infringement,
unacceptable use, etc
0% 20% 40% 60% 80% 100%
Process automaƟon/inbound rouƟng
InformaƟon governance and metadata
generaƟon / correcƟon
Contextual search, curaƟon, e-discovery
Analysis / business insight / customer input
Yes Plans in Place No plans
0% 5% 10% 15% 20% 25%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90%100%
Document and content management (ECM)
Records management (RM) / content
retenƟon
Business intelligence (BI) / reporƟng
Enterprise-wide search
Excellent Good Fair Poor
0% 10% 20% 30% 40% 50% 60% 70%
manual steps
improving search
overloads
0% 20% 40% 60% 80% 100%

Industry
Watch
extractingknowledge
Importance and Leadership
Looked at today, 17% of our respondents consider content analytics to be “essential”, with 48% feeling it
is “something we definitely need”, but projecting that to five years’ time, this grows to 59% feeling it will be
essential, and 28% a definite need, with only 13% seeing it simply as “useful”.
There has been much talk about the need for a CDO – variously described as a Chief Data Officer or Chief
Digital Officer – to raise awareness and realize the potential of analytics or big data projects, but when we
asked, only 4% of our sample have such a position, with 1% having a CAO or Chief Analytics officer. 10%
said they have plans in place, and 6% felt their organization has such a job role, but not with that job title
(CIO is given as the most likely alternative). By implication, therefore, 80% of our responding organizations
have yet to allocate a senior role to initiate and coordinate analytics applications.
Adoption and Applications
Taking a broad look at adoption across the four areas that we have identified (and remembering that this is
a self-selected survey and will over-read the general population) 38% are using content analytics for one
or more types, with around 20% using any one of the types, and 20-30% with plans in place. Contextual
search and e-discovery is the most popular overall, but information governance and metadata correction
shows the most potential growth. Looking at usage across business sizes, mid-sized organizations (500-
5,000 employees) are lagging somewhat, especially in analysis and business insight applications, where
14% have applications in use, compared to 28% of the largest organizations (5,000+ employees). Smaller
organizations at 21% are surprisingly active here.
Figure 3: Are you using content analytics for any of the following? (N=219)
Looking in a little more detail at specific applications, 21% are extracting data from emails, forms or
invoices – most likely invoices - and 19% are using free-text search, although it is likely that many of these
applications do not use a high degree of text analysis, relying mostly on keyword extraction.
16% are generating or correcting metadata for content classification or tagging, and 13% are applying this
to email management and archiving. 9% are using content analytics as part of a big data project across
multiple data sources.
0% 10% 20% 30% 40% 50% 60% 70%
manual steps
improving search
overloads
0% 20% 40% 60% 80% 100%
Process automaƟon/inbound rouƟng
InformaƟon governance and metadata
generaƟon / correcƟon
Contextual search, curaƟon, e-discovery
0% 5% 10% 15% 20% 25%
To extract data from emails, correspondence,
forms or invoices
For free-text search/indexing
To generate or correct metadata for content
classificaƟon/tagging
To manage/archive emails
To route inbound content or mail to the
appropriate processes / people / archive
To check or correct for security or privacy issues
As part of a big data project involving mulƟple
internal data sources
For analysis or curaƟon of internal/external
content/knowledge bases
To monitor and/or extract knowledge from social
streams
For fraud/crime detecƟon or intelligence
To build business insight or formal knowledge
extracƟon
To filter or re-classify unwanted content,
pre-migraƟon or ongoing
For sound, image or video files

Industry
Watch
extractingknowledge
Figure 4: Are you currently using content analytics on unstructured content
in any of the following ways? (N=212)
Progress and Issues
As with any relatively new software application, interest is high, but progress is mixed. A quarter of our
respondents feel it is either not applicable, or that they are stuck in a world of paper processes. 37% either
have no one tasked to investigate, no mandate from above, or no budget to proceed (or a combination of
these). For 23%, a start has been made, but progress is slow, or of mixed success. 11% are underway and
encouraged by the results, and 4% are already showing a return on their investment.
Figure 5: How would you best describe current progress in your organization towards
the use of content analytics? (N=220)
0% 5% 10% 15% 20% 25%
To extract data from emails, correspondence,
forms or invoices
For free-text search/indexing
To generate or correct metadata for content
classificaƟon/tagging
To manage/archive emails
To route inbound content or mail to the
appropriate processes / people / archive
To check or correct for security or privacy issues
As part of a big data project involving mulƟple
internal data sources
For analysis or curaƟon of internal/external
content/knowledge bases
To monitor and/or extract knowledge from social
streams
For fraud/crime detecƟon or intelligence
To build business insight or formal knowledge
extracƟon
To filter or re-classify unwanted content,
pre-migraƟon or ongoing
For sound, image or video files
0% 5% 10% 15% 20%
It’s not really applicable
We are stuck in a world of manual processes
It could be useful but no one is tasked to
invesƟgate
It has not been set as a priority from above
There is genuine interest but no budget to
move forward
We are invesƟgaƟng possibiliƟes but progress
is slow
We have tried a few projects but with mixed
success
We are convinced this is the way to go and
are working on it
It has already proved its ROI and we are
proceeding apace
0% 10% 20% 30% 40%

Industry
Watch
extractingknowledge
Issues
Again, as we might expect for a new technology, lack of expertise is a big issue, reported by 36%. As we
suggested before, not having firm and agreed information governance and content retention policies is also
an issue that needs to be solved before rules-based classification can be implemented. Our respondents
are also reporting some technical issues around connecting repositories and setting up the rules. Compared
to big data projects in general, “over-hyped management expectations” does not seem to be a significant
issue for our early adopters.
Figure 6: What are the biggest issues for you with content analytics projects? (N=207)
60% of our respondents feel that content analytics will become an essential capability for their organization
within the next five years, and while initial efforts are a little varied in outcome, users are applying the
technology across a range of application areas.
Process Automation and Inbound Routing
More recently tagged as “smart business processes”, automated and adaptive processing based on
analysis of inbound content has been growing steadily in recent years. As the volume, variety and urgency
of multi-channel inbound content has grown, users have been looking at ways to reduce handling loads,
speed up response, and embed compliance into their customer or supplier-facing processes. The most
popular application has been invoice processing (accounts payable) where invoices are recognized out
of the inbound mail, examined for layout of key fields and OCR’d to capture the actual data. This is then
validated against the original purchase order data from the finance system.
Varying degrees of analytic capability can be built into this application, and it can, of course, be extended
to any number of inbound forms. As the inbound capture extends across more and more types of content,
especially where the digital mailroom concept is employed (centrally or distributed), recognition of content
type and automated routing to specific processes becomes very useful. In many cases, the arrival of a
specific form or piece of customer correspondence (paper or email) can kick off a downstream process such
as on-boarding, a support ticket, or a claim.
It has not been set as a priority from above
There is genuine interest but no budget to
move forward
We are invesƟgaƟng possibiliƟes but progress
is slow
We have tried a few projects but with mixed
success
We are convinced this is the way to go and
are working on it
It has already proved its ROI and we are
proceeding apace
0% 10% 20% 30% 40%
We lack experƟse in this area
We need to set InformaƟon Governance (IG)
policies first before we can set the rules
It needs a considerable investment in tools
and resources
We haven't really looked at it recently
ConnecƟng to and between repositories and
systems can be difficult
Seƫng the analysis rules can be difficult and
Ɵme-consuming
It’s hard to predict that the outcome will be
successful
We need to comply with data privacy laws
The tools are immature and hard to use
Management expectaƟons are over-hyped
0% 2% 4% 6% 8% 10% 12% 14% 16% 18%
OCR data capture to process, with validaƟon
Auto-classificaƟon/tagging for archive, ECM,
or RM
CollecƟon of documents/emails into case
folders
Automated rouƟng of inbound mail to
specific acƟve processes
SeparaƟon of content types in the mail-
stream (e.g. forms, invoices, etc.)
Process triggered from inbound mail item
(scanned from paper)
Process triggered from inbound email
In-process workflow adjustment, e.g.
adapƟve case management
Fraud detecƟon
Process triggered from mobile device input

Industry
Watch
extractingknowledge
It then becomes particularly useful if a case-folder is created, and subsequent inbound items, such as proof
of identities, assessment reports, income statements, etc., can be automatically routed to the case folder.
This is also where intelligent case management can use information derived from the inbound content to
adapt the required processes within the case, ensuring that procedures are followed in a compliant way. The
most advanced organizations (5%) are even able to trigger processes from mobile device apps.
Figure 7: Are you using content analytics for any of these inbound content functions? (N=196)
Automating Email Classification
It has been one of the longest running dilemmas of electronic records management systems as to whether
to declare important emails as records into the system, and if so, how to rely on staff to do so reliably and
responsibly, and how to avoid overloading the system with irrelevant records. As emails now carry full
evidential weight in litigation cases, many organizations have implemented bulk email archiving systems, or
long-term stored back-ups, in order to cover off potential legal discovery or freedom of information requests.
Unfortunately, many of these archives are of the “store and forget” variety with little in the way of applied
metadata, and no legal hold and e-discovery tools for contextual searches. They are certainly not optimized
for surfacing knowledge or being part of the “corporate memory”.
Given that humans will never become consistent in filing and classification, and that the volume of emails
continues to grow rapidly, automation is likely to be the only solution that can provide a usable and
defensible way to archive emails. This may be fully automated, or may be a prompting system, asking users
to confirm the suggested classification. As we will see later, there will be those who question the accuracy of
machine classification, but email is particularly interesting in this context as most of us already rely on (and
trust) a degree of spam filtering on our inbound emails, and the latest email clients are making their own
judgments as to what emails to prioritize.
Only 5% of responding organizations are currently using fully automated classification of emails, with 11%
using user-prompted techniques. However, a further 24% have plans in the next 12-18 months to do so, a
sign that this long-running problem may finally be reaching an accepted solution.
Ɵme-consuming
It’s hard to predict that the outcome will be
successful
We need to comply with data privacy laws
The tools are immature and hard to use
Management expectaƟons are over-hyped
0% 2% 4% 6% 8% 10% 12% 14% 16% 18%
OCR data capture to process, with validaƟon
Auto-classificaƟon/tagging for archive, ECM,
or RM
CollecƟon of documents/emails into case
folders
Automated rouƟng of inbound mail to
specific acƟve processes
SeparaƟon of content types in the mail-
stream (e.g. forms, invoices, etc.)
Process triggered from inbound mail item
(scanned from paper)
Process triggered from inbound email
In-process workflow adjustment, e.g.
adapƟve case management
Fraud detecƟon
Process triggered from mobile device input
Yes, fully
automated, 5%
Yes, user
prompted, 11%
As batch
correcƟon or
enhancement,
2%
Plans in next 12-
18 months, 24%
No immediate
plans, 52%
Unlikely we ever
will, 5%

Industry
Watch
extractingknowledge
Figure 8: Are you using auto-classification for filing or archiving inbound emails?
(N=168, excl. 34 Don’t Know)
Project Success
The benefits of content analytics for users of inbound processing seem to be well defined. We can see
in Figure 8 that processes are flowing more smoothly, staff are happy to avoid the tedious task of filing,
and governance and compliance are much improved. As far as productivity improvements, 18% report
that they are achieving high levels of “hands-off” processing, where large chunks of the process are
handled by the computer.
There have been some issues, particularly accuracy and miss-hits, and to overcome those has involved
a higher degree of set-up and tuning than some users were expecting. However, 27% report a positive
ROI already.
Figure 9: How would you describe the success of your inbound analytics projects?
(Check all that apply) (N=44, excl. 102 “Not applicable”, 50 “Too early to say”)
Only 5% of respondents have fully automated classification for filing or archiving emails, with another
11% having user-prompted filing. According to forward plans, this is set to more than double in the next
12 to 18 months.
Yes, fully
automated, 5%
Yes, user
prompted, 11%
As batch
correcƟon or
enhancement,
2%
Plans in next 12-
18 months, 24%
No immediate
plans, 52%
Unlikely we ever
will, 5%
0% 10% 20% 30% 40% 50% 60%
Processes are flowing faster and more
smoothly
Staff are pleased to avoid otherwise
tedious tasks
Governance and compliance are much
improved
We are achieving high levels of “hands-off”
processing
Fraud discovery rates have gone up
considerably
We have some issues with accuracy and
miss-hits
It has involved more set-up and tuning than
we expected
The overall ROI has been very posiƟve
Yes, across a
number of
content types,
10%
Yes, across one
or two content
types, 10%
No plans, 41%

Industry
Watch
extractingknowledge
Information Governance and Metadata Generation /
Correction
We have seen a very rapid acceptance of the idea of auto-classification2
for the purposes of improving
compliance over the last three years, although as we will see, improving searchability is also a prime
driver. In this survey 20% are already actively using it, with a further 9% just getting started. An additional
31% have plans to do so, including 8% in the short term. Overall, this represents nearly two-thirds of our
respondents.
Figure 10: Are you using auto-classification to assist staff with content filing /
metadata allocation / records declaration? (N=190)
Although what we might call the classic view of auto-classification is that content is classified based on
analysis of its text (or sound, or imagery) at the point of creation or ingestion, there is a strong application
area that uses batch agents to crawl over existing content in whatever repository it exists, and to apply
or correct its metadata based on a set of rules aligned to the information governance policy, and/or to the
current taxonomy.
Once the metadata has been sorted out, many useful management controls can be applied. Searchability
is improved, particularly in terms of accuracy and completeness. This can hugely benefit knowledge
sharing and maximizes the value of stored information for research, reuse and audit, as well as speeding
up the legal discovery process. Aligning metadata and taxonomies between repositories will also facilitate
enterprise-search or content federation. If content is to be migrated between systems, aligned metadata is
essential, and, of course redundant, obsolete and trivial content (ROT) can be left behind and deleted.
Governance and compliance are much
improved
We are achieving high levels of “hands-off”
processing
Fraud discovery rates have gone up
considerably
We have some issues with accuracy and
miss-hits
It has involved more set-up and tuning than
we expected
The overall ROI has been very posiƟve
Yes, across a
number of
content types,
10%
Yes, across one
or two content
types, 10%
Just geƫng
started, 9%
Keen to
automate as
soon as we can,
8%
We have plans
to do so in the
future, 23%
No plans, 41%
0% 2% 4% 6% 8% 10%12%14%16%18%20%
Add or correct metadata to improve
searchability
Add or correct metadata prior to migraƟon
alignment between repositories
Detect duplicate files (by content)
Add or correct metadata and flag for
deleƟon/retenƟon
Detect security risks and misallocated
access rights
Detect sensiƟve or privacy-related content
Encrypt or redact sensiƟve content
Detect offensive content (text)
Detect infringing or offensive images/video
0% 10% 20% 30% 40% 50% 60%
Our content search is much more accurate
and useful

Industry
Watch
extractingknowledge
Figure 11: Do you use automated or batch agents to perform any of the following functions?
(N=189, 59% “None of these”)
This removal of ROT, and also detection of duplicate content (even if filenames are different), can recover
considerable amounts of storage space, which in itself speeds up and improves search. Content type-
classification and correctly set metadata will be an essential step in determining retention periods, with the
knock-on effect that potentially risky or non-compliant content can be defensibly deleted. If sensitive content
is detected, it can be tagged for a higher access level and even encrypted or redacted for enhanced security.
Finally, offensive or unacceptable content can be detected, and dealt with immediately. For some
organizations, this capability alone is sufficient to justify the purchase of a content remediation tool.
Project Success
52% of those using auto-classification report much improved content search, 40% have seen an
improvement in staff productivity, and 31% feel that their general compliance and governance is much
improved - a strong endorsement across a number of important goals within the business. The benefits
continue: defensible deletion, recovered storage space and better optimized systems are all cited. On the
issues side, some experienced difficulties with rules-setting to align with IG policies, and it is taking time for
some to see the expected results.
Figure 12: How would you describe the success of your auto-classification / metadata correction
projects? (Select all that apply) (N=48, excl. 99 “Not applicable”, 43 “Too early to say”)
to do so in the
future, 23%
0% 2% 4% 6% 8% 10%12%14%16%18%20%
searchability
deleƟon/retenƟon
access rights
0% 10% 20% 30% 40% 50% 60%
and useful
Staff producƟvity is much improved
Our general compliance and governance is
much improved
It has helped us to beƩer standardize our
metadata across mulƟple repositories
We are defensibly deleƟng considerable
amounts of redundant content
We have recovered significant storage space
Our merged/migrated/upgraded system is
much more effecƟve and IG compliant
We are sƟll struggling with rules-seƫng and
IG alignment
We have yet to achieve the promised/
expected results
We have a achieved a considerable ROI
Yes, across one
or two content
types, 10%
Just geƫng
started, 9%
Keen to
automate as
soon as we can,
8%
We have plans
to do so in the
future, 23%
No plans, 41%
0% 2% 4% 6% 8% 10%12%14%16%18%20%
searchability
deleƟon/retenƟon
access rights
0% 10% 20% 30% 40% 50% 60%
and useful
Staff producƟvity is much improved
Our general compliance and governance is
much improved
It has helped us to beƩer standardize our
metadata across mulƟple repositories
We are defensibly deleƟng considerable
amounts of redundant content
We have recovered significant storage space
Our merged/migrated/upgraded system is
much more effecƟve and IG compliant
We are sƟll struggling with rules-seƫng and
IG alignment
We have yet to achieve the promised/
expected results
We have a achieved a considerable ROI

Industry
Watch
extractingknowledge
Legal Judgment
Knowing that some legal advisors might take a view that automated classification is not sufficiently
accurate to rely on, particularly as regards deletion of emails, we asked if our respondents had
encountered any legal resistance. 34% indicated wide acceptance within their organization, including 2%
who withstood a challenge in court. Of the remainder, 42% are not in full operation, and only 15% report
that this issue is holding up adoption.
Figure 13: Have you encountered any legal resistance or compliance questions regarding auto-
classifying emails or other records pre-deletion? (N=52, excl. 136 Don’t Know, N/A)
As a follow up question, we asked what degree of accuracy of classification, both for emails, and for
general content, might be deemed acceptable in their organization. We also suggested that this should
apply to human classification as well as automated. More than a third (36%) are OK with an 85%
accuracy or less, another third (38%) with 95% or less. Only 26% feel that greater than 95% accuracy is
needed, including 9% who are seeking 99% accuracy. It would be interesting to audit the content systems
in these companies to see if human accuracy can actually achieve these levels!.
Figure 14: For emails and general content, what would you consider to be an acceptable accuracy
of classification within your organization (human or automated)? (N=138, excl. 47 Don’t know)
37% are using or just getting started with auto-classification, and are seeing the benefits of corrected
metadata in searchability, productivity and compliance. 74% are looking for an accuracy of 95% to avoid
any legal resistance.
We withstood a
court challenge,
2% It’s accepted as a
consistent, rules-
based procedure,
15%
There is concern,
but humans are
no beƩer at this
than computers,
17%
We are not yet
100% reliant on
full automaƟon,
42%
This is something
that is holding up
adopƟon, 15%
60-70% accurate,
11%
70-80% accurate,
14%
80-85% accurate,
11%
85-90% accurate,
18%
90-95% accurate,
20%
95-98% accurate,
17%
99% accurate, 9%
Yes – across
mulƟple internal
and external
repositories/
libraries, 7%
Yes – across
mulƟple internal
repositories, 11%
Yes – within a
No – simple
search, single
We don’t have
any searchable
ECM/DM/RM
systems, 11%
We withstood a
court challenge,
2% It’s accepted as a
consistent, rules-
based procedure,
15%
There is concern,
but humans are
no beƩer at this
than computers,
17%
We are not yet
100% reliant on
full automaƟon,
42%
This is something
that is holding up
adopƟon, 15%
60-70% accurate,
11%
70-80% accurate,
14%
80-85% accurate,
11%
85-90% accurate,
18%
90-95% accurate,
20%
95-98% accurate,
17%
99% accurate, 9%
Yes – across
mulƟple internal
and external
repositories/
libraries, 7%
We don’t have
any searchable
ECM/DM/RM
systems, 11%

Industry
Watch
extractingknowledge
Contextual Search, Curation, and E-discovery
As we mentioned earlier, many content search engines rely on simple keyword searches, perhaps
extended with some Boolean capabilities. Users are increasingly frustrated that these search methods fall
so short of what is available with Google search on the web. Of course, indexing web pages, with their
links and popularity, is somewhat less demanding than searching across multiple corporate repositories
for important but little-referenced documents.
Users expect the indexing to include the significance of the keywords, as set by their position in
headlines, body text, and so on. They are looking for differentiation between authoritative documents (and
authors) and others. They only want the final version of a contract, or the customer letters that threaten
legal action. They may like captions and annotations on drawings, or even photos, to show up in the
keyword index.
Only 35% of our respondents have any form of contextual search, and this includes 17% who are
restricted to a single repository. 7% have sophisticated search across multiple internal and external
repositories or libraries. A third are restricted to simple search across a single repository, or do not even
have a searchable ECM/DM/RM system.
Figure 15: Do you have a search capability that includes contextual analysis (as opposed to
simple free text or keywords)? (N=175, excl. 16 Don’t Know)
Metadata Creation/Correction
We talked earlier of adding value to the dark data that exists in most organizations, and the way to do
this is to use content remediation or correction tools to trawl through the content, and intelligently add
metadata, or fix metadata that is wrong or doesn’t match the current classification scheme. In this way,
even less sophisticated search tools can be made much more effective. 39% have improved their search
capability this way, with 8% feeling that it made a “huge difference”.
60-70% accurate,
11%
70-80% accurate,
14%
80-85% accurate,
11%
85-90% accurate,
18%
90-95% accurate,
20%
95-98% accurate,
17%
99% accurate, 9%
Yes – across
mulƟple internal
and external
repositories/
libraries, 7%
Yes – across
mulƟple internal
repositories, 11%
Yes – within a
single repository,
17%
No – just simple
search, across
mulƟple
repositories, 23%
No – simple
search, single
repository, 30%
We don’t have
any searchable
ECM/DM/RM
systems, 11%
Yes – it made a
huge diﬀerence,
8%
Yes – it was a
useful
improvement,
16%
Yes – improved
some speciﬁc
areas, 15%
No, our content is
well-enough
tagged already,
3%
No, but we
certainly should
do, 58%

Industry
Watch
extractingknowledge
Figure 16: Have you used metadata creation/correction on existing content to improve
searchability? (N=191)
E-discovery
Contextual analysis can be particularly useful for pre-trial e-discovery work, picking up on contract terms,
intellectual property, survey reports, complaints, etc. Internally, it can also be used for compliance audits.
For example, price-fixing, tax avoidance, money laundering, fraud, etc., will all have a likely vocabulary and
context that can be detected using much the same techniques as external fraud detection.
Having said that, it would seem from our results that half of those who have such a tool (10%) do not use
it very much. 22% have e-discovery tools that are not contextual, 59% have no tools, including 29% of the
largest organizations.
Figure 17: Do you have e-discovery tool(s) with contextual analysis capability?
Curation
In many industry sectors such as medical, pharmaceutical, legal, aeronautical, it is important to stay abreast
of published content from elsewhere, and in the past the curation of this content would be the role of the
company librarian, often with a physical library of books, research reports and periodicals. Today, that sifting
or curation role can be assigned to computers, collecting electronic content, and feeding specific references
on defined topics to those that need them. However, to truly replace the previous role, the content needs to
be collected from outside the business, and include websites, blogs and news feeds.
Yes – it made a
huge diﬀerence,
8%
Yes – it was a
useful
improvement,
16%
Yes – improved
some speciﬁc
areas, 15%
No, our content is
well-enough
tagged already,
3%
No, but we
certainly should
do, 58%
Yes, and we are
very reliant on
this, 8%
Yes, but this
capability is not
much used, 10%
We have
e-discovery tools,
but the search is
not contextual,
22%
We do not have
any e-discovery
tools, 59%
Yes, including
websites, blogs
and news feeds,
3%
Yes, but only
from subscribed
libraries, 7%
Yes, internal only,
9%
This is largely a
We don’t have a
need for this,
16%

Industry
Watch
extractingknowledge
19% of our respondents have some automated curation, although half of those are internal only. 6% have
the traditional manual process. Of the rest, 59% feel it would be very useful to have such a service for
their key knowledge workers.
Figure 18: Do you use content curation to automatically create custom libraries and alerts from
multiple external and internal sources? (N=187)
Only a third of organizations have contextual search, but half of those are restricted to one repository.
39% have improved their search with some form of automated metadata creation or correction.
Analysis / Business Insight / Customer Input
AIIM first reported on content analytics 5 years ago. Our subsequent reports picked up on the big data
theme, or “big content” as we prefer to call it. The problem then, as it is now, is to come up with a pick-
list of the most common applications. Then it was mostly based on blue-sky thinking: what would be the
most useful thing for your business to know? Now we have a much more established set of applications,
although that is not to say that there aren’t plenty of innovative uses yet to come.
Now, as then, help-desk logs and CRM reports are the most popular source for analysis, picking up
on customer experience and marketing insights, and a little further down, the free-form comment fields
from feedback forms. Next come HR applications, particularly screening résumés for match with job
specifications. Web accessible databases figure highly for plans-in-place, and this is often a curated feed,
or might be a check of publicly available data, e.g., FBI records for previous convictions as part of a loan
application. Similarly, incident reports, claims and witness statements are all part of fraud detection or due
diligence.
We have
e-discovery tools,
but the search is
not contextual,
22%
We do not have
any e-discovery
tools, 59%
Yes, including
websites, blogs
and news feeds,
3%
Yes, but only
from subscribed
libraries, 7%
Yes, internal only,
9%
This is largely a
manual process,
6%
We don’t do this,
but it would be
very useful, 59%
We don’t have a
need for this,
16%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90%100%
Help desk logs, CRM reports
Resumés, HR records
Comment form ﬁelds for suggesƟons/ feedback
Web-accessible databases
Incident reports, claims, witness statements
Print-streams and electronic statements
Case notes, prof. assessments, medical notes
Web forums, blogs, raƟngs/reviews
Lab notes, trials, surveys
Picture, video or audio records
External libraries public or subscripƟon
Patents, scienƟﬁc journals, court proceedings
Already do Plans in place Would like to Unlikely
0% 20% 40% 60% 80% 100%
Incoming customer communicaƟon streams
Helpdesk/service-desk conversaƟons
Media channels, news feeds

Industry
Watch
extractingknowledge
Figure 19: Have you considered analyzing any of the following document or content types to
extract business intelligence or solve problems? (N=178. Line-length indicates “N/A”)
Real-Time or Near-Time
Incoming customer communications and help-desk streams also top the list for live or near-time alerting,
along with an increasing interest in media channels and news feeds. There is, quite rightly, as much
interest in what customers are saying on the organization’s own community pages as on external social
streams, and the former is set to grow more. CCTV and audio monitoring obviously have their place, but
this is a more difficult technology.
Figure 20: Have you considered automated analysis of any of the following to extract live or near-
time business intelligence? (N=178. Line-length indicates “N/A”)
We don’t do this,
but it would be
very useful, 59%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90%100%
0% 20% 40% 60% 80% 100%
Customer communiƟes / comments on your blogs
Facebook pages and other social sites
External social streams (eg, TwiƩer, LinkedIn)
Internal chat/Skype
Internal social streams (e.g. Yammer, Jive)
CCTV/audio
This is largely a
manual process,
6%
We don’t do this,
but it would be
very useful, 59%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90%100%
0% 20% 40% 60% 80% 100%
Customer communiƟes / comments on your blogs
Facebook pages and other social sites
External social streams (eg, TwiƩer, LinkedIn)
Internal chat/Skype
Internal social streams (e.g. Yammer, Jive)
CCTV/audio

Industry
Watch
extractingknowledge
Social Media Monitoring
Looking in more detail at social media, the importance of monitoring these fast-moving streams has soared
in the past few years, and as a result most organizations have implemented a monitoring mechanism
(64%) but only 14% have an automated system . Relying on (designated) staff to alert the marketing or
customer service department when complaints (or praise!) show up can be somewhat hit-and-miss, and
the speed of response can be crucial in these situations. Automated monitoring using sentiment analysis is
a much more reliable way to alert the appropriate people to make a response.
Figure 21: How are you monitoring external social streams (e.g. Twitter, LinkedIn, Facebook)?
Business Advantage
Improved products or services comes out as the top benefit from business intelligence derived from
content analytics, followed by core investigations and knowledge research. Detection of non-compliance
rates highly, as do general customer sentiment monitoring, and individual customer complaint handling.
Figure 22: Which of the following business advantages would be the most useful to you based on
intelligence derived from content analytics? (Max 4) (N=176)
We have
automated
monitoring, and it
is successful, 5% We have some
automated
monitoring in
place – mostly
defensive, 11%
We have a project
underway, 4%
We do monitor,
but it is largely
manual, 44%
We aren’t, but it is
something we
probably should
do, 15%
It’s not really
relevant to our
business, 21%
0% 10% 20% 30% 40% 50% 60% 70%
Improved product or service quality
Knowledge research/core invesƟgaƟons
DetecƟon of non-compliance
CompeƟƟve advantage
Customer senƟment monitoring (general)
Rapid response to external events
Customer complaint handling/brand
protecƟon (individuals)
Incident predicƟon
Reduced losses from fraud
Staﬀ senƟment monitoring
0% 10% 20% 30%
We have
automated
monitoring, and it
is successful, 5% We have some
automated
monitoring in
place – mostly
defensive, 11%
We have a project
underway, 4%
We do monitor,
but it is largely
manual, 44%
We aren’t, but it is
something we
probably should
do, 15%
It’s not really
relevant to our
business, 21%
0% 10% 20% 30% 40% 50% 60% 70%
Improved product or service quality
Knowledge research/core invesƟgaƟons
DetecƟon of non-compliance
CompeƟƟve advantage
Customer senƟment monitoring (general)
Rapid response to external events
Customer complaint handling/brand
protecƟon (individuals)
Incident predicƟon

Industry
Watch
extractingknowledge
Progress
As we indicated early on, around 25% of our respondents have active projects in the “business insight”
category, with 10% having several. Across company sizes, the mid-sized businesses are lagging with only
9% active as yet, compared with 40% of the largest, and an encouraging 24% of the smallest, indicating
a readiness to jump in with competitive advantage where possible, or in some cases, build a business on
this.
Figure 23: Do you currently have one or more active “big content” or “content analytics”
applications making use of unstructured or textual data for business insight? (N=180)
Mid-sized companies are falling behind in the take up of business insight projects involving content
analytics, with only 1 in 10 having any active projects, compared with 1 in 4 of smaller organizations, and
nearly half of larger ones.
Big Content Projects
In seeking to characterize the projects being worked on, we asked which of the “three Vs” they involved
– volume, velocity, variety. There is a fairly even split, with 11% involving volume and velocity, 36% high
volume, 15% high velocity, 23% high variety, and 17% neither, but using complex techniques.
We also asked if the big content project involves a link to transactional or structured data, such as CRM
systems, financial systems, data logs, etc. 53% are linked to one or more internal systems, and 5% are
linked to external data sets.
When it comes to how the projects have been deployed, or what tools are being used, nearly half have
used in-house development and 17% external custom (rising to 27% for the largest organizations). 27%
are using cloud products, and 17% products from their ECM vendor, with 13% using analytics products
from a pure-play vendor. 21% are using open source in some form, which is quite prevalent in this area.
Incident predicƟon
Lots, 2% Several,
8%
One or
two,
14%
Planned,
13%
Not as
yet, 52%
Unlikely,
12%
0% 10% 20% 30%
Lots
Several
One or two
Planned
10-500
emps
500-5,000
emps
5,000+
emps
0% 10% 20% 30% 40% 50% 60%
In-house developed tools
Cloud/SaaS services
AnalyƟcs products from your ECM
vendor(s)
Open Source soluƟons
External custom development
Pure-play analyƟcs products

Industry
Watch
extractingknowledge
Figure 24: Are you using any of the following for your big content project(s)?
(N=48 with projects)
ROI
With any new technology, there are likely to be those who have latched on to it to solve a very specific
problem, or to gain a big business advantage, and there will be others with over-ambitious plans, or who are
hampered by lack of analytical skills. 34% of our respondents achieved a return on their investment in 12
months or less, and 68% in 18 months or less. This is a solid expectation of success, although from the 22%
taking 2 years or more to show a return, we can infer that some projects will need a little longer to bed down
and show a return.
Figure 25: How would you rate the ROI from your big content project(s)?
(N=32, excl. 13 “Not Measured” and 12 “Too Early to Say”)
Opinions
Our “opinions” question is intended as a way to take the pulse of active practitioners, and those who are
aware of the possibilities but may have more pragmatic issues to solve.
n 53% agree that auto-classification is the only way to get chaos under control.
n 75% agree that enhancing the value of legacy content is better than wholesale deletion.
n 73% know there are real business insights to be gained.
n 54% feel they are exposed to risk from non-identified content.
n 63% being held back by lack of skills and allocated authority.
0% 10% 20% 30% 40% 50% 60%
In-house developed tools
Cloud/SaaS services
AnalyƟcs products from your ECM
vendor(s)
Open Source soluƟons
External custom development
Pure-play analyƟcs products
6 months or less,
6%
6-12 months,
28%
12-18 months,
34%
18 months to
2 years, 9%
2-3 years, 9%
More than 3
years, 13%
40% 20% 0% 20% 40% 60% 80%
Automated classiﬁcaƟon using content analyƟcs
is the only way to get our content chaos under
control.
Enhancing the value of our legacy content
through analyƟcs is a beƩer strategy than
whole-scale deleƟon.
Content-based automaƟon is the only way to
cope with increasing volumes of mulƟ-channel
inbound content.
We are exposed to considerable risk in the
business due to content that is not correctly
idenƟﬁed.
There are real business insights in our content if
we can get the analyƟcs right.
Monitoring social media for customer senƟment

Industry
Watch
extractingknowledge
Figure 28: How do you feel about the following statements? (N=171)
In summary, content analytics is generally considered to be a promising and useful technology, particularly
as a way to increase content value and deal with increasing volumes of inbound content. For most, a lack
of designated leadership and a shortfall of analytics skills is holding back exploitation of these new tools.
Spend
The indications are for growth in all areas, particularly enhanced/contextual search, analytics for business
insight, and automated classification tools or modules. Inbound workflow automation shows demand as
organizations build up their multi-channel inbound capabilities. Content migration tools have been buoyed
by SharePoint 2007 to 2010 migrations, but are still showing strong growth for the 2010 to 2013 upgrade.
Figure 29: How do you think your organization’s spending on the following areas and applications in
the next 12 months will compare with what was actually spent in the last 12 months?
(N=168, excl. Same ~ 40%)
As we might expect with a new technology, growth forecasts are strong, as early adopters make way for
more mainstream users, driven partly by the need to control content chaos, but also by the refinement of
analytics tools and their ability to provide actionable business insight.
12-18 months,
34%
40% 20% 0% 20% 40% 60% 80%
control.
inbound content.
idenƟfied.
and brand protecƟon is a must these days.
We are being held back by the absence of
allocated responsibiliƟes and a lack of analyƟcs
skills.
Strongly Disagree Disagree Neither Agree nor Disagree Agree Strongly Agree
-20% -10% 0% 10% 20% 30% 40% 50%
Enhanced/contextual search
Content analyƟcs for business insight
Automated classificaƟon tools or modules
Inbound workflow automaƟon
Content migraƟon
Metadata correcƟon /content remediaƟon tools
Dedicated e-discovery tools
OCR and data capture
Social monitoring tools
Less More
12-18 months,
34%
40% 20% 0% 20% 40% 60% 80%
control.
inbound content.
idenƟfied.
and brand protecƟon is a must these days.
We are being held back by the absence of
allocated responsibiliƟes and a lack of analyƟcs
skills.
Strongly Disagree Disagree Neither Agree nor Disagree Agree Strongly Agree
-20% -10% 0% 10% 20% 30% 40% 50%
Enhanced/contextual search
Content analyƟcs for business insight
Automated classificaƟon tools or modules
Inbound workflow automaƟon
Content migraƟon
Metadata correcƟon /content remediaƟon tools
Dedicated e-discovery tools
OCR and data capture
Social monitoring tools
Less More

Industry
Watch
extractingknowledge
Conclusion and Recommendations
Content analytics is rightly taking its place amongst the corporate toolset, but while the business insight (or
big data / big content) projects are still in something of an early-adopter phase, there are a number of other
applications based on content analysis techniques that are already showing strong benefits in smoother
workflows, improved search, and better compliance. We have seen increasing interest and adoption
in recognition and routing of inbound content, automated classification of records and email, metadata
addition and correction, and all of the improvements in access, security, de-duplication and retention that
flow from this.
Staying on top of high volume, multi-channel, inbound content is increasingly difficult if relying on manual
processes, and users are coming to accept that automated handling is as accurate but more consistent
than humans. Email archiving in particular presents a dilemma, and content analytics offers a way to carry
out defensible deletion in line with information governance polices. Dealing with dark data elsewhere in the
business, and adding value to content rather than deleting it is a common objective.
Projects to derive business insight from content analytics are proceeding ahead, with 20% of our survey
respondents already active, and a further 30% with plans. With some of these early projects coming on
stream, 68% are reporting ROI within 18 months or less. Improving products or services is the top-rated
benefit, followed by knowledge research or core investigations, and then improved compliance.
Recommendations
n If your content or records management deployment is stalled due to poor decisions early on regarding
classification, metadata and taxonomies, or if you are migrating content from multiple repositories to a
single system, take a look at metadata correction agents that can sort ROT from valuable content, and
align content types and metadata.
n If you have access to contextual search, ensure that it is properly tuned, and that staff know how to use
it. If you are reliant on more basic search, consider improving the searchability, and therefore the value
of your content, by correcting and enhancing the metadata using analytic agents.
n Unless your staff are diligent and consistent at declaring, classifying and tagging records, consider
providing auto-classification assistance or full auto-classification. Be aware that your information
governance policies need to be updated and consistent as they will provide the rules for automated
agents.
n Take control of your emails. If you have no archive, or the archive is “file and forget” you are losing
potential corporate knowledge, but are also exposing the business to risk, and creating a potential
e-discovery nightmare.
n Look at your retention policies as a way to control increasing storage requirements. Accurate metadata
and enforced retention policies are the only way to limit storage, but will also improve your compliance
and risk exposure.
n Inbound content handling can rapidly overload process staff, and reduce speed of response to
customers. Implement a digital mailroom philosophy, and use automated recognition, routing and data
extraction.
n Look across the range of your business activities to see where content analytics could provide business
insight to understand customer needs, improve competitive advantage, help to solve cases and
investigations, or prevent non-compliance and fraud.
References
1 “Connecting and Optimizing SharePoint” AIIM Industry Watch, January 2015. www.aiim.org/research
2 “Automating Information Governance – assuring compliance” AIIM Industry Watch, May 2014.
www.aiim.org/research

Industry
Watch
extractingknowledge
Appendix 1: Survey Demographics
Survey Background
The survey was taken by 238 individual members of the AIIM community between April 17 2015, and May
08, 2015 using a Web-based tool. Invitations to take the survey were sent via email to a selection of the
80,000 AIIM community members.
Organizational Size
Survey respondents represent organizations of all sizes. Larger organizations over 5,000 employees
represent 31%, with mid-sized organizations of 500 to 5,000 employees at 31%. Small-to-mid sized
organizations with 10 to 500 employees constitute 38%. Respondents from organizations with less than
10 employees have been eliminated from the results, taking the total to 222 respondents.
Geography
72% of the participants are based in North America, with 14% from Europe and 14% rest-of-world.
11-100 emps,
14%
101-500
emps, 24%
501-1,000
emps, 8%
1,001-5,000
emps, 23%
5,001-10,000
emps, 7%
over 10,000
emps, 24%
US, 57%
Canada, 15%
UK, Ireland, 4%
Western
Europe, 7%
Eastern
Europe, Russia,
3%
Australia, NZ,
5%
Middle East,
Africa, S.Africa,
3%
Asia, Far East,
2% Central/
S.America,
Caribbean, 4%
Government &
Public Services -
Local/ State, 13%
Government &
Public Agencies -
NaƟonal/
InternaƟonal, 7%
Finance, Banking,
Insurance, 9%EducaƟon, 4%
Healthcare, 4%
Media,
Entertainment,
Publishing, 4%
Life Science,
PharmaceuƟcal, 3%
Non-Proﬁt, Charity,
3%
Manufacturing,
Aerospace, Food,
Process, 3%
Retail, Transport,
Real Estate, 2%
Legal and Prof.
Services, 1%
Other, 2%
11-100 emps,
14%
101-500
emps, 24%
501-1,000
emps, 8%
1,001-5,000
emps, 23%
5,001-10,000
emps, 7%
over 10,000
emps, 24%
US, 57%
Canada, 15%
UK, Ireland, 4%
Western
Europe, 7%
Eastern
Europe, Russia,
3%
Australia, NZ,
5%
Middle East,
Africa, S.Africa,
3%
Asia, Far East,
2% Central/
S.America,
Caribbean, 4%
Government &
Public Services -
Local/ State, 13%
Government &
Public Agencies -
NaƟonal/
InternaƟonal, 7%
Media,
Entertainment,
Life Science,
PharmaceuƟcal, 3%
3%
Manufacturing,
Aerospace, Food,
Process, 3%
Retail, Transport,
Real Estate, 2%
Legal and Prof.
Services, 1%
Other, 2%

Industry
Watch
extractingknowledge
Industry Sector
Local and National Government together make up 20%, Finance and Insurance 9%, and Energy 9%.
Suppliers of ECM services have been included as their responses are in alignment with other IT and High
Tech. Other sectors are evenly split.
Job Roles
27% of respondents are from IT, 44% have a records management or information management role, and
21% are line-of-business managers or consultants.
US, 57%
Canada, 15%
UK, Ireland, 4%
Government &
Public Services -
Local/ State, 13%
Government &
Public Agencies -
NaƟonal/
InternaƟonal, 7%
Finance, Banking,
Insurance, 9%
Energy, Oil & Gas,
Mining, 9%
Consultants, 9%
IT & High Tech —
ECM supplier, 9%
IT & High Tech —
not ECM, 7%
Document Services
Provider, 5%
Engineering &
ConstrucƟon, 5%
Telecoms, Water,
UƟliƟes, 5%
EducaƟon, 4%
Healthcare, 4%
Media,
Entertainment,
Publishing, 4%
Life Science,
PharmaceuƟcal, 3%
3%
Manufacturing,
Aerospace, Food,
Process, 3%
Retail, Transport,
Real Estate, 2%
Legal and Prof.
Services, 1%
Other, 2%
IT staff, 9%
Head of IT, 3%
IT Consultant or
Project Manager,
15%
Records or
document
management
staff, 24%
Head of records/
informaƟon
management,
20%
Line-of-business
execuƟve,
department
head or process
owner, 8%
Business
Consultant, 10%
Legal or
Compliance, 2%
Chief Data
Officer,
Knowledge
Officer, Analyst,
4%
President, CEO,
Managing
Director, 3%
Other, 1%
US, 57%
Canada, 15%
UK, Ireland, 4%
Europe, 7%
Government &
Public Services -
Local/ State, 13%
Government &
Public Agencies -
NaƟonal/
InternaƟonal, 7%
Finance, Banking,
Insurance, 9%
Energy, Oil & Gas,
Mining, 9%
Consultants, 9%
IT & High Tech —
ECM supplier, 9%
IT & High Tech —
not ECM, 7%
Document Services
Provider, 5%
Engineering &
ConstrucƟon, 5%
Telecoms, Water,
UƟliƟes, 5%
EducaƟon, 4%
Healthcare, 4%
Media,
Entertainment,
Publishing, 4%
Life Science,
PharmaceuƟcal, 3%
3%
Manufacturing,
Aerospace, Food,
Process, 3%
Retail, Transport,
Real Estate, 2%
Legal and Prof.
Services, 1%
Other, 2%
IT staff, 9%
Head of IT, 3%
IT Consultant or
Project Manager,
15%
Records or
document
management
staff, 24%
Head of records/
informaƟon
management,
20%
Line-of-business
execuƟve,
department
head or process
owner, 8%
Business
Consultant, 10%
Legal or
Compliance, 2%
Chief Data
Officer,
Knowledge
Officer, Analyst,
4%
President, CEO,
Managing
Director, 3%
Other, 1%

Industry
Watch
extractingknowledge
Appendix 2: General Comments
Do you have any general comments to make about your content analytics
projects? (Selective)
n This survey has shown how much I do not know about content analytics.
n Our organization will definitely benefit from content analytics, but we need to show some success to
get management support.
n Our organization does not understand what that is. So when I bring it up they do not know how to
respond other than “that would be nice”.
n We have only just started the two projects related to this, so although we may find that we have
enhanced capability (e.g. content analytics for business insight) this is not one of the drivers, and don’t
yet really know how much more we can achieve once the tools are in place.
n Remove the “human element” to establish consistency.
n Unfortunately it’s not applicable for small companies. But some thoughts brought by this survey are
quite useful.
n Not enough attention is paid to unlocking unstructured content. Even a “simple” word doc can be
very hard to understand/contextualize.... analysis is almost the ‘easy’ part... its the preparation /
organization that is tricky.
n Hadoop is great and worth the cost.

Industry
Watch
extractingknowledge
UNDERWRITTEN IN PART BY
Swiss Post Solutions, a division of Swiss Post, offers a comprehensive range of document and business process
outsourcing services. With 7400 people working across Europe, North America and Asia and with access to an
extensive partner network, we are able to support our clients across the globe.
Private and Public sector organizations have chosen to outsource their physical and digital document
processing needs to us, utilizing our extensive knowledge of people-based outsourcing and our capability to
deliver document processing services on, near or offshore. Our corporate information management system is a
unified delivery platform that provides organizations with the ability to cost-effectively on-board and distributes
documents throughout the organization. It provides our clients with the capability to:
• Simultaneously improve productivity and reduce operational costs
• Take an enterprise-wide approach to automating business processes
• Enable improved decision making and customer satisfaction by accelerating business transactions
• Reduce the risk of non-compliance and achieving legislative and regulatory requirements
Regardless of document type, physical or electronic medium, format, language or geographic location, Swiss
Post Solutions offers an end-to-end solution from document creation to content management, production,
distribution and business intelligence.
www.swisspostsolutions.com
Swiss Post Solutions AG

Industry
Watch
extractingknowledge
Learn how to combine content analytics, collaboration, governance and processes
with anywhere, anytime access to deliver value to your customers, partners, and
employees. That’s what ECM -- and these best practices resources -- are all about.
AIIM Content Analytics Resource Centre
www.aiim.org/Resource-Centers/Content-Analytics

Industry
Watch
extractingknowledge
AIIM (www.aiim.org) AIIM is the global community of information professionals. We provide the education,
research and certification that information professionals need to manage and share information assets in an
era of mobile, social, cloud and big data.
© 2015
AIIM AIIM Europe
1100 Wayne Avenue, Suite 1100 The IT Centre, Lowesmoor Wharf
Silver Spring, MD 20910 Worcester, WR1 2RR, UK
+1 301.587.8202 +44 (0)1905 727600
www.aiim.org www.aiim.eu

Aiim Industry Watch: Content Analytics: automating processes and extracting knowledge

More Related Content

What's hot (7)

Viewers also liked (16)

Similar to Aiim Industry Watch: Content Analytics: automating processes and extracting knowledge (20)

More from Swiss Post Solutions (15)

Recently uploaded (20)

Aiim Industry Watch: Content Analytics: automating processes and extracting knowledge