Jacob Kildebogaard - Web Juice (All Things Data 2015)
0 likes531 views
The document discusses attribution modeling and how to analyze a company's digital marketing performance. It addresses how different channels like AdWords can have direct and assisted conversions. It also discusses building custom attribution models that weight channels differently and adjusting bidding based on the analysis, which led to a 22% increase in assisted value. The key lessons are to understand how channels are used in user paths, analyze assisted ratios, use and customize attribution models, and adjust strategies based on data insights.
Jacob Kildebogaard - Web Juice (All Things Data 2015)
- 2. Assisted conversions makes you look… Good? >> The path to conversion << @webanaly)ker #ATDconf
- 3. >> Me… << @webanaly)ker #ATDconf GaTricks.com
- 4. >> Lets set the scene! << ”Its not as simple as it looks!” One aBribution in Google Analytics 1 Another in AdWords with AdWords conversion script 2 A third in Analytics aBribution reports 3 @webanaly)ker #ATDconf
- 5. >> The source might not be what you think<< 2 direct visit Group 1 Group 2 Search, referral, paid traffic, campaigns… All traffic, identified as direct traffic 4 examples: 1 direct visit + 1 Adwords visit 2 visits from Adwords 1 direct visit + 2 visits from AdWords @webanaly)ker #ATDconf
- 6. >> AdWords conversion tracking << ”30 days cookie!” @webanaly)ker #ATDconf
- 7. >> Assisted conversions<< ”Do people click and buy later?” Direct conversions and assisted conversions 1 See total conversions a source has been involved in 2 @webanaly)ker #ATDconf
- 8. >> We know it from sport << @webanaly)ker #ATDconf GOAL
- 9. >> All traffic reports << AdWords generates 27% of revenue @webanaly)ker #ATDconf
- 10. >> Assisted… << Assists and last click @webanaly)ker #ATDconf
- 11. >> Assist ratio << Assist conversions Last-‐‑Click Conversions Assist Ratio @webanaly)ker #ATDconf
- 12. >> Assist touchpoints << Source: @sandramcamacho Assist ratio @webanaly)ker #ATDconf
- 13. >> Closing touchpoints << Source: @sandramcamacho Assist ratio @webanaly)ker #ATDconf
- 14. >> Assisted… << Assists and last click -‐‑> ratio @webanaly)ker #ATDconf
- 15. >> Assist and last click << @webanaly)ker #ATDconf
- 16. >> Total impact… << @webanaly)ker #ATDconf
- 17. >> Total impact… << AdWords is affecting 22% @webanaly)ker #ATDconf
- 18. >> Data so far << Analytics: 2,66 mio 1 Assist: 2,5 mio 2 Last click: 2 mio 3 Total conversion value: 3,8 mio 4 @webanaly)ker #ATDconf
- 20. >> Assisted… With brand << Brand is a huge part of the revenue 1 @webanaly)ker #ATDconf
- 21. >> Total impact… With brand << Nonbrand is 33% of all AdWords @webanaly)ker #ATDconf
- 22. >> Non brand AdWords << All traffic: 553K 1 Assist: 994K 2 Last Click: 445K 3 Total impact: 1.309K 4 @webanaly)ker #ATDconf
- 23. >> Assist vs last click<< 1 2 3 4 Increase cpc @webanaly)ker #ATDconf
- 24. >> Paths << @webanaly)ker #ATDconf
- 25. >> ABribution models << @webanaly)ker #ATDconf
- 26. >> ABribution models << @webanaly)ker #ATDconf
- 27. >> ABribution models << Focus on 1 visit in path Focus on full path @webanaly)ker #ATDconf
- 28. >> ABribution models << @webanaly)ker #ATDconf
- 29. >> Compare models <<
- 30. >> Build your own model << Why do you make your own? 1 What should be treated different from standard? 3 Any assist sources, you want to weight higher? 2 Custom ABribution models @webanaly)ker #ATDconf
- 31. >> Custom ABribution model << @webanaly)ker #ATDconf
- 32. >> Custom ABribution model << @webanaly)ker #ATDconf
- 33. >> Custom ABribution model << More credit to middle because I aim to grow the business! @webanaly)ker #ATDconf
- 34. >> Custom ABribution model << @webanaly)ker #ATDconf
- 35. >> ABribution models << @webanaly)ker #ATDconf
- 36. >> Awesome, but not perfect << @webanaly)ker #ATDconf
- 37. >> The issues << 90 days lookback window 2 No cross device tracking 1 @webanaly)ker #ATDconf
- 38. >> So what did we do? << • AdWords generates revenue but is strong as assist • Especially Non-‐‑brand AdWords is strong as assist • New model where assisted weight more • AdWords bids adjusted based on model • After 5 month, we saw 22% increase in assist value ABribution case @webanaly)ker #ATDconf
- 39. >> Summing up << • Different numbers in different reports • Learn how the channel is used in user path, based on assist ratio – if higher than 1 = strong assist • Use aBribution models to go deeper • Adjust your own model based on data Learn and earn: @webanaly)ker #ATDconf
- 40. Jacob Kildebogaard Jk @webjuice.dk @Webanalytiker GaTricks.com