SuperWeek 2016 - Garbage In Garbage Out: Data Quality in a TMS World
17 likes4,108 views
Simo Ahava discusses data quality and the importance of a data-driven process and culture. He advocates for breaking down silos between teams by implementing a shared data layer and involving stakeholders from all teams in iterative development through a definition of done that incorporates data tracking requirements. Empowering developers to facilitate data collection and analysis, and hiring hybrid profiles with both business and technical skills can also improve data quality.
![1. Set of business
requirements
for tracking
digital assets,
visits, and
visitors.
2. Encoded, global
data structure,
accessed and
modified by
connected
platforms.
2. Data model of a
connected
platform, which
copies or digests
information in the
global structure.
dataLayer.push({
'pageType' : 'home'
});
google_tag_manager['GTM-123']
.dataLayer
.set('pageType', 'home');](https://image.slidesharecdn.com/superweek2016-160206074912/85/SuperWeek-2016-Garbage-In-Garbage-Out-Data-Quality-in-a-TMS-World-42-320.jpg)
SuperWeek 2016 - Garbage In Garbage Out: Data Quality in a TMS World
- 1. Reaktor Mannerheimintie 2 00100, Helsinki Finland tel: +358 9 4152 0200 www.reaktor.com info@reaktor.com Confidential ©2015 Reaktor All rights reserved garbage in, garbage outData quality in a TMS world Simo Ahava Senior Data Advocate
- 2. Simo Ahava Senior Data Advocate, Reaktor Google Developer Expert, Google Analytics Blogger, developer, www.simoahava.com Twitter-er, @SimoAhava Google+:er, +SimoAhava
- 3. Data quality isn’t fixed. Depending on the hypothesis, a single data set can shift from useless to incredibly insightful without a single datum changing shape, size, form, or function. #1 Data is subjective
- 4. Plug-and-play Analytics @SimoAhava from @ReaktorNow | #SPWK | 5 Feb 2016
- 5. Plug-and-play Analytics Data quality isn’t acquired — it’s earned. @SimoAhava from @ReaktorNow | #SPWK | 5 Feb 2016
- 8. Claim 1: Data quality is destroyed by laziness and lack of ambition.
- 9. Claim 2: A TMS empowers developers more than others.
- 10. The root of all evil @SimoAhava from @ReaktorNow | #SPWK | 5 Feb 2016
- 11. The root of all evil The "project" @SimoAhava from @ReaktorNow | #SPWK | 5 Feb 2016
- 12. Your organization is creating absurd amounts of data with every passing second, and it’s very difficult to adapt to the fluctuations without an agile, process-driven mindset. #2 Data is a process
- 13. The project is often a series of handovers, breeding non- involvement.
- 14. Specification Implementation Analysis Results
- 15. Specification Implementation Analysis Results Business owner
- 16. Specification Implementation Analysis Results Business owner Marketing
- 17. Specification Implementation Analysis Results Business owner Marketing Developer
- 18. This leads necessarily to silos, which have entry and exit conditions.
- 25. Silos, so what? @SimoAhava from @ReaktorNow | #SPWK | 5 Feb 2016
- 26. Silos, so what? As long as the work gets done, right? @SimoAhava from @ReaktorNow | #SPWK | 5 Feb 2016
- 27. Data is the lifeblood of the organization. It flows through all departments, across job titles, permeating the very fabric of the organization, reinforcing its foundations for growth. #3 Data abhors silos
- 28. Do these sound familiar:
- 29. Monthly reports which lack relevance, are rife with generic suggestions that lack research in the context of your business, reiteration of previous month’s points, even if there are solid reasons why they weren’t addressed.
- 30. Ridiculously ugly and ineffective JavaScript hacks for measurement points which should be tackled in the Data Layer.
- 31. Hiding behind data, and passing blame to other silos. Could someone fix the Bounce Rate metric on our site?
- 32. Analytics feature requests are deprioritized, and deployed extremely infrequently. Fix transactionRevenue to show revenue, not customer weight.
- 33. Communication is difficult due to the overhead of meeting face-to-face, project plans are set in stone during sales, and it’s difficult to change existing project goals or set new ones due to consultants being hired as "extra pairs of hands" rather than advisors.
- 34. These are symptoms of data being treated as a project outcome.
- 35. Cure I: The Data Layer @SimoAhava from @ReaktorNow | #SPWK | 5 Feb 2016
- 36. Cure I: The Data Layer Using technology to solve communication problems @SimoAhava from @ReaktorNow | #SPWK | 5 Feb 2016
- 37. Typically, there are three definitions of Data Layer that we use in the digital world.
- 39. 1. Set of business requirements for tracking digital assets, visits, and visitors.
- 40. 1. Set of business requirements for tracking digital assets, visits, and visitors. 2. Encoded, global data structure, accessed and modified by connected platforms.
- 41. 1. Set of business requirements for tracking digital assets, visits, and visitors. 2. Encoded, global data structure, accessed and modified by connected platforms. 2. Data model of a connected platform, which copies or digests information in the global structure.
- 42. 1. Set of business requirements for tracking digital assets, visits, and visitors. 2. Encoded, global data structure, accessed and modified by connected platforms. 2. Data model of a connected platform, which copies or digests information in the global structure. dataLayer.push({ 'pageType' : 'home' }); google_tag_manager['GTM-123'] .dataLayer .set('pageType', 'home');
- 43. Across all three definitions, the purpose of a Data Layer is simple:
- 44. DMP / DWH / TMS / etc. X X Actions Presentation Data Layer
- 45. DMP / DWH / TMS / etc. X X Actions Presentation Data Layer The purpose of a Data Layer is to provide a bilateral layer on the digital asset, which decouples, normalises, and uniformly encodes semantic information passed through and stored within.
- 46. The Data Layer is a joint venture, where people and systems communicate across silos.
- 47. DMP / DWH / TMS / etc. let tracker = GANTracker.sharedTracker() tracker.trackEvent("revenue", action:"Q1", value:"15678000") tracker.trackEvent("revenue", action:"Q2", value:"16888000") tracker.trackEvent("revenue", action:"Q3", value:"15991000") tracker.trackEvent("revenue", action:"Q4", value:"19133000") rq12014,rq22014,rq32014,rq42015 15677998,16887988,15990988,19133400 analytics.collect({ 'revenueQ1' : '15677998.00', 'revenueQ2' : '16887988.00', 'revenueQ3' : '15990988.00', 'revenueQ4' : '19133400.00' })
- 48. DMP / DWH / TMS / etc. let dataLayer = new Array() dataLayer.push({ "revenue_Q1_2014" : "15677998.00", "revenue_Q2_2014" : "16887988.00", "revenue_Q3_2014" : "15990988.00", "revenue_Q4_2014" : "19133400.00" })
- 49. Cure II: The Process @SimoAhava from @ReaktorNow | #SPWK | 5 Feb 2016
- 50. Cure II: The Process Involve, involve, involve @SimoAhava from @ReaktorNow | #SPWK | 5 Feb 2016
- 51. An iterative, agile process is necessary for optimal utilization of a TMS.
- 53. Definition of Done Developed features do not impede measurement. Developed features are trackable. Sprint
- 54. Definition of Done Developed features do not impede measurement. Developed features are trackable. Sprint If necessary, feature is encoded with tracking attributes. If necessary, feature is linked to a Data Layer object. Feature
- 55. Definition of Done Developed features do not impede measurement. Developed features are trackable. Sprint If necessary, feature is encoded with tracking attributes. If necessary, feature is linked to a Data Layer object. Feature Attribute syntax is correct for tracking. Data Layer object syntax is correct. Task
- 59. Cure III: Empowerment @SimoAhava from @ReaktorNow | #SPWK | 5 Feb 2016
- 60. Cure III: Empowerment We are all hybrid beings @SimoAhava from @ReaktorNow | #SPWK | 5 Feb 2016
- 61. The entire life cycle of a single data point, from collection to reports, requires knowledge and expertise to manage. #4 Data is difficult
- 62. Developer facilitation is crucial to data quality and optimized data collection.
- 63. 1: Education
- 64. 1. JavaScript: www.codecademy.com, www.codeschool.com, Professional JavaScript for Web Developers, DOM Enlightenment… 2. Digital analytics: www.kaushik.net, www.simoahava.com, Successful Analytics, Practical Google Analytics and Google Tag Manager for Developers… 3. Training, courses, certifications: Digital Analytics Association, Digital Analytics Fundamentals (Google), Market Motive… 4. Conferences: MeasureCamp, SMX, eMetrics, Digital Analytics Hub, ConversionXL, Superweek, All Things Data…
- 65. treat content as a product2: hybrid skills
- 66. "Business owner" - No operational skills + Strategic "Developer" - Uncooperative + Methodical "Marketer" - Bully + Consultative
- 67. + Passionate, actively interested + Understands ever-changing requirements + Good grasp of digital tech + Statistical mindset + Knows the product / service inside and out + Critical about the present, curious about the future
- 68. treat content as a product3: Passionate interest
- 69. + Dedicated sandbox + Website or blog to test new ideas on + Test and debug setups in Google Analytics and Google Tag Manager + Utilization of GTM environments
- 70. Hire to educate, not to delegate PO Developer Analyst
- 71. Hire to educate, not to delegate PO Developer Analyst
- 72. Data is difficult @SimoAhava from @ReaktorNow | #SPWK | 5 Feb 2016
- 73. Data is difficult Data quality is earned, not acquired @SimoAhava from @ReaktorNow | #SPWK | 5 Feb 2016
- 74. Thank you! simo.ahava@reaktor.com www.simoahava.com Twitter: @SimoAhava Google+: +SimoAhava Data is difficult - http://goo.gl/53aFUU The Schema Conspiracy - http://goo.gl/o2Pwys Further reading: 10 Truths About Data - http://goo.gl/EpesEj