SlideShare a Scribd company logo

Incibeta ignite google-measurement stategy (1)

•
•
Incubeta NMPi
Incubeta NMPi

Google UK speaker Alex Maximov, explains Measurement strategy in the industry, predicting upcoming shifts and trends in the current retail space that resides in an ever-changing digital marketing sphere.

1 of 20
Download to read offline
Alex Maksimov Measurement & Analytics Lead Google UK Measurement Strategy ©2020 INCUBETA. ALL RIGHTS RESERVED.
Measurement strategy Measurement & Analytics Lead Alex Maksimov
Why?
Privacy and personalisation Consumers User expectations are rising as people demand more transparency, choice, and control over how their online data is used Advertisers Companies want to provide more personalised messaging and retargeting through customer data vs
The industry is shifting Regulatory Changes New regulations such as GDPR are impacting how data can be collected and used User Privacy Controls Users are demanding more control and transparency over data collected and used for ads personalization Browser Updates Heightened controls are impacting traditional data collection (e.g. third party cookies and device identifiers)
The predictive era Broadcast Era Target an audience in an environment with the understanding not everyone is relevant. Reaching the majority of your target audience. Precision Era Target the right customers at the right time with the right media. Aspiration to 1:1 marketing. Predictive Era Striving for the best of both worlds in a privacy safe way. Doing more with less. Sophisticated algorithms. You are here
Implications for advertisers How and what brands can measure Crediting marketing activity with customer or business outcomes Getting to the right number Marketers need to do more with less
The right number? Econometrics/MMM A regression model estimates the effect of all advertising on weekly sales, alongside all other factors. Digital Attribution Credit is assigned to digital media touchpoints based on their involvement in conversions Controlled Experiments Users or regions are matched to differ only by their media exposure. Uplift is inferred to have been caused by media. Three common tools
Lloyd Shapley 1923 - 2016
(in the precision era) Digital Attribution Television Radio Social Paid Search Organic Search Email Direct Mail Display OLV Content Affiliate Conversion
(in the prediction era?) Digital Attribution More data Less data Deterministic Probabilistic 3P 1P
Strengths and weaknesses (not exhaustive) Digital Attribution Strengths ● Managing budgets within a channel ● Getting fast insight - can be close to real time ● On the fly bidding decisions ● Valuing beyond the last click Weaknesses ● Showing the incremental impact of the advertising ● It can’t tell you anything about something that is not in the system ● It assumes that the media in the tool is what caused the outcome
Francis Galton 1822 - 1911
Simplified version Econometrics for marketing Sales ~ Base Sales + Advertising + Other Factors + Error What would have happened anyway Paid EarnedOwned Under Advertiser Control Outside Advertiser Control Because no model is perfect Traditional Digital Products Distribution Promotions Price Trends Weather Competitor Activity EconomyService Season ● TV ● Radio ● Outdoor ● Print ● Sponsorship ● Door Drops ● Paid Search ● Display ● Online Video ● Paid Social ● Website ● Owned Social ● Direct Mail ● PR ● Organic search ● Social engagement ● Word of mouth
Strengths and weaknesses (not exhaustive) Econometrics for marketing Strengths ● Estimating the contribution of marketing and media to the overall business ● Explaining what (is likely to have) caused short term sales ● Making budget recommendations for advertising channels ● Showing how channels work together ● [Advanced] Looking at brand advertising’s impact on sales Weaknesses ● Recommending what to do when you change price, product, strategy ● Making quick decisions - they require >2 years of historical data - and take a fair time to do (most of this is collecting and cleaning data) ● Explaining things that follow sales (Example: always on media - often there won’t be enough variation in the data for the model to pick it up)
Louis Pasteur 1822 - 1895
The scientific method in the general form Controlled Experiments Make an observation Ask a question Form a hypothesis Conduct an experiment Accept the hypothesis Reject the hypothesis
Strengths and weaknesses (not exhaustive) Controlled Experiments Strengths ● Showing the incremental impact of a change on any metric ● Proving the value of a specific channel ● Testing new approaches ● Moderating the findings from other tools Weaknesses ● Dealing with many changes at once ● Allowing marketers to carry on with the same level of spend ● Getting real-time results - normally experiments take 4-12 weeks to run
Multiple solutions for a diverse set of insights The future? Incremental marketing mix attribution: A blend of MMM, Experiments and Attribution, that works across all channels, and controls for external factors. Maybe it’ll do the creative too :-)
Thank you

More Related Content

Incibeta ignite google-measurement stategy (1)

  • 1. Alex Maksimov Measurement & Analytics Lead Google UK Measurement Strategy ©2020 INCUBETA. ALL RIGHTS RESERVED.
  • 2. Measurement strategy Measurement & Analytics Lead Alex Maksimov
  • 4. Privacy and personalisation Consumers User expectations are rising as people demand more transparency, choice, and control over how their online data is used Advertisers Companies want to provide more personalised messaging and retargeting through customer data vs
  • 5. The industry is shifting Regulatory Changes New regulations such as GDPR are impacting how data can be collected and used User Privacy Controls Users are demanding more control and transparency over data collected and used for ads personalization Browser Updates Heightened controls are impacting traditional data collection (e.g. third party cookies and device identifiers)
  • 6. The predictive era Broadcast Era Target an audience in an environment with the understanding not everyone is relevant. Reaching the majority of your target audience. Precision Era Target the right customers at the right time with the right media. Aspiration to 1:1 marketing. Predictive Era Striving for the best of both worlds in a privacy safe way. Doing more with less. Sophisticated algorithms. You are here
  • 7. Implications for advertisers How and what brands can measure Crediting marketing activity with customer or business outcomes Getting to the right number Marketers need to do more with less
  • 8. The right number? Econometrics/MMM A regression model estimates the effect of all advertising on weekly sales, alongside all other factors. Digital Attribution Credit is assigned to digital media touchpoints based on their involvement in conversions Controlled Experiments Users or regions are matched to differ only by their media exposure. Uplift is inferred to have been caused by media. Three common tools
  • 10. (in the precision era) Digital Attribution Television Radio Social Paid Search Organic Search Email Direct Mail Display OLV Content Affiliate Conversion
  • 11. (in the prediction era?) Digital Attribution More data Less data Deterministic Probabilistic 3P 1P
  • 12. Strengths and weaknesses (not exhaustive) Digital Attribution Strengths ● Managing budgets within a channel ● Getting fast insight - can be close to real time ● On the fly bidding decisions ● Valuing beyond the last click Weaknesses ● Showing the incremental impact of the advertising ● It can’t tell you anything about something that is not in the system ● It assumes that the media in the tool is what caused the outcome
  • 14. Simplified version Econometrics for marketing Sales ~ Base Sales + Advertising + Other Factors + Error What would have happened anyway Paid EarnedOwned Under Advertiser Control Outside Advertiser Control Because no model is perfect Traditional Digital Products Distribution Promotions Price Trends Weather Competitor Activity EconomyService Season ● TV ● Radio ● Outdoor ● Print ● Sponsorship ● Door Drops ● Paid Search ● Display ● Online Video ● Paid Social ● Website ● Owned Social ● Direct Mail ● PR ● Organic search ● Social engagement ● Word of mouth
  • 15. Strengths and weaknesses (not exhaustive) Econometrics for marketing Strengths ● Estimating the contribution of marketing and media to the overall business ● Explaining what (is likely to have) caused short term sales ● Making budget recommendations for advertising channels ● Showing how channels work together ● [Advanced] Looking at brand advertising’s impact on sales Weaknesses ● Recommending what to do when you change price, product, strategy ● Making quick decisions - they require >2 years of historical data - and take a fair time to do (most of this is collecting and cleaning data) ● Explaining things that follow sales (Example: always on media - often there won’t be enough variation in the data for the model to pick it up)
  • 17. The scientific method in the general form Controlled Experiments Make an observation Ask a question Form a hypothesis Conduct an experiment Accept the hypothesis Reject the hypothesis
  • 18. Strengths and weaknesses (not exhaustive) Controlled Experiments Strengths ● Showing the incremental impact of a change on any metric ● Proving the value of a specific channel ● Testing new approaches ● Moderating the findings from other tools Weaknesses ● Dealing with many changes at once ● Allowing marketers to carry on with the same level of spend ● Getting real-time results - normally experiments take 4-12 weeks to run
  • 19. Multiple solutions for a diverse set of insights The future? Incremental marketing mix attribution: A blend of MMM, Experiments and Attribution, that works across all channels, and controls for external factors. Maybe it’ll do the creative too :-)