Datalicious data driven media planning
8 likes2,622 views
The document discusses media planning and optimization using data-driven marketing techniques. It describes how media mix modeling and media attribution analysis can be used to predict future media performance and analyze past performance to inform optimal channel strategies and mix. Agent-based modeling is recommended to simulate consumer behavior under different media scenarios to test and optimize media spending. Case studies demonstrate how these techniques can boost marketing ROI.
Datalicious data driven media planning
- 1. > Media Planning < Media mix modelling and media a-ribu1on to boost media ROI
- 2. > Smart data driven marke3ng Media A5ribu3on & Modeling Op3mise channel mix, predict sales Tes3ng & Op3misa3on Remove barriers, drive sales Boos3ng ROMI Targe3ng & Merchandising Increase relevance, reduce churn “Using data to widen the funnel” May 2013 © Datalicious Pty Ltd 2
- 3. > Wide range of data services Data PlaHorms Data collec3on and processing Adobe, Google Analy3cs, etc Web and mobile analy3cs Tag-‐less online data capture Retail and call center analy3cs Big data & data warehousing Single customer view Insights Analy3cs Data mining and modelling Tableau, Splunk, SPSS, R, etc Customised dashboards Media a5ribu3on analysis Marke3ng mix modelling Social media monitoring Customer segmenta3on Ac3on Campaigns Data usage and applica3on SiteCore, ExactTarget, etc Targe3ng and merchandising Marke3ng automa3on CRM strategy and execu3on Data driven websites Tes3ng programs May 2013 © Datalicious Pty Ltd 3
- 4. > Best of breed technologies May 2013 © Datalicious Pty Ltd 4
- 5. > Clients across all industries May 2013 © Datalicious Pty Ltd 5
- 6. > Data driven media planning § Media mix modelling – Predic1ng future media performance – Macro insights to inform channel strategy § Media a-ribu1on – Analysis of historic media performance – Micro insights to inform channel op1misa1on May 2013 © Datalicious Pty Ltd 6
- 7. > Media mix modelling 101011010010010010101111010010010101010100001011111001010101 010100101011001100010100101001101101001101001010100111001010 010010101001001010010100100101001111101010100101001001001010 May 2013 © Datalicious Pty Ltd 7
- 8. > Media mix modelling § Predict the future based on the past, i.e. predict sales based on media investment using a model that is based on historic data – Old school approach: Regression-‐based § Looks at correla1on between summary data, i.e. media investment and total sales per week – Recommended approach: Agent-‐based § Simulates market condi1ons through a set of consumers (agents) and then exposes those agents to different media scenarios to test them (predict sales) before they are even implemented May 2013 © Datalicious Pty Ltd 8
- 9. > ThinkVine agent-‐based modelling May 2013 © Datalicious Pty Ltd 9 ThinkVine uses agent-‐based modeling (ABM) to simulate the ac1ons of consumers in the market based on a combina1on of internal client data as well as external market data. This enables adver1sers to predict and test the impact of different budget mix configura1ons before they have been implemented thus improving overall media effec1veness.
- 10. > Data requirements and process May 2013 © Datalicious Pty Ltd 10 1. Training the system: Datalicious analysts develop and calibrate a custom model based on agents that recreates past sales. 2. Proving the system: Once the system is trained, we validate it by comparing predicted sales to actual sales. 3. Using the system: Once the model has been calibrated & validated, it can be used to predict sales for different media scenarios thus tes1ng them before they are executed
- 11. > Modelling data requirements § Sales data – As granular as possible, i.e. by week by store § Media data – Data on the media ac1vity that lead to the above sales § Customer profile data – Any quan1ta1ve and qualita1ve insights on exis1ng customers that help configure the ThinkVine agents (i.e. simulated consumers) § Consumer profile data – Any quan1ta1ve and qualita1ve insights on Australian consumers in general that help configure the agents (i.e. simulated consumers) May 2013 © Datalicious Pty Ltd 11
- 12. > Media mix modelling outputs May 2013 © Datalicious Pty Ltd 12 The so^ware enables fast, objec1ve comparisons of marke1ng plan alterna1ves, i.e. you can vary spending levels, 1ming, mix of tac1cs and consumer groups targeted. The so^ware enables your team to easily run “what if?” scenarios, i.e. you can find the best investment levels to meet various strategic and tac1cal objec1ves by predic1ng poten1al sales for various different media budget scenarios.
- 13. > ThinkVine food services May 2013 © Datalicious Pty Ltd 13 In this case study, the company achieved greater media efficiencies and ROI and also quan1fied the halo effect of cross brand marke1ng. More specifically, more than $10 million of media spend was over-‐saturated due to a mismatch between the target consumers’ media consump1on habits and the deployed marke1ng tac1cs.
- 14. > Media a5ribu3on 101011010010010010101111010010010101010100001011111001010101 010100101011001100010100101001101101001101001010100111001010 010010101001001010010100100101001111101010100101001001001010 May 2013 © Datalicious Pty Ltd 14
- 15. > The ideal media dashboard Channel Investment ROMI Return Brand equity Baseline ($100) n/a $40 Offline TV, print, outdoor, etc $7 330% $30 Direct Direct mail, email, etc $1 400% $5 Online Search, display, social, etc $2 1150% $25 May 2013 © Datalicious Pty Ltd 15
- 16. > ROMI as compe33ve advantage May 2013 © Datalicious Pty Ltd 16 74% of marketers do not engage in any form of media a-ribu1on aside from the last click leaving 26% of marketers with a serious compe11ve advantage as their media investment is likely to generate a much higher ROMI.
- 17. Closer Paid search Display ad views TV/print responses > Full purchase path tracking May 2013 © Datalicious Pty Ltd 17 Influencer Influencer $ Display ad clicks Online sales Affiliate clicks Social referrals Offline sales Organic search Social buzz Retail visits Life3me profit Organic search Emails, direct mail Direct site visits Introducer
- 18. > Custom models most effec3ve May 2013 © Datalicious Pty Ltd 18 56% of marketers consider a unique or custom (weighted) media a-ribu1on approach that does not use a standard out-‐of-‐the-‐box methodology as most effec1ve.
- 19. Touch point 1 > Analy3cs to pick the best model May 2013 © Datalicious Pty Ltd 19 Touch point 2 Touch point 3 Touch point N Closer Influencer Influencer $ Introducer Touch point 1 Touch point 2 Touch point 3 Touch point N Touch point 1 Touch point 2 Touch point 3 Touch point N ✖ ✔ ✖
- 20. > A5ribu3on models compared May 2013 © Datalicious Pty Ltd 20 COST PER CONVERSION Last click a-ribu1on Custom (weighted) a-ribu1on
- 21. > Media a5ribu3on May 2013 © Datalicious Pty Ltd 21 Aussie purchase path tracking and media a-ribu1on modelling in close coopera1on with Amnesia designed to op1mise the overall Aussie budget mix across paid and earned media resul1ng in an overall project ROI of 910%.
- 22. > Media a5ribu3on May 2013 © Datalicious Pty Ltd 22 Suncorp purchase path tracking and media a-ribu1on modelling in order to op1mise the overall Suncorp insurance budget mix across paid and earned media resul1ng in an overall project ROI of 2,078%.
- 23. May 2013 © Datalicious Pty Ltd 23 Contact me cbartens@datalicious.com Learn more blog.datalicious.com Follow us twi5er.com/datalicious
- 24. Smart data driven marke3ng May 2013 © Datalicious Pty Ltd 24