Download to read offline
Uploaded byBooster Box
SMX Advanced - ROAS is dead. Welcome FUMOAS.
The document critiques traditional performance marketing metrics, particularly ROAS, claiming it is inadequate and suggests focusing on future contribution margins instead. It highlights the importance of leveraging machine learning and first-party data while shifting from keyword targeting to audience targeting. The document outlines strategies for integrating audience data, building a comprehensive view of business results, and conducting RFM analysis to enhance paid campaign effectiveness.
More Related Content
Similar to SMX Advanced - ROAS is dead. Welcome FUMOAS.
SMX Advanced - ROAS is dead. Welcome FUMOAS.
- 1. ROAS is Dead. WelcomeFUMOAS.
- 2.
- 3. Advisor for Google’sown equity fund Capital G. Who am I… Gianluca Voted 2nd Most Influential PPC Person in 2019. Binelli
- 4. We’re the best*scientific performance marketing agency Who is Booster Box... * we are the best according to our mums and now that we’ve won an award as the Best small PPC agency in Europe, it’s official: mums are always right. That’s quite a mouthful, but it’s just a way to say that we love numbers and the scientific method. We have been running wildly successful international Paid and Brand Media Campaigns since July of 2016
- 5.
- 6.
- 7. Sin 1: Relyingon ROAS ROAS is a broken metric. 7 1
- 8. 8 When we lookin Google Ads, we usually see just a portion of the truth
- 9. 9 We are blindto the other portion sitting in the CRM RETURNS MARGIN MARGIN ON ADV SPEND
- 10. Why Is ROASa Broken Metric? 10 Revenue, not margins Does not include future transactions not Lifetime value of existing customers
- 11. Sin 2: Luddism Googleis becoming a machine learning driven black box, embrace technology. 11 2
- 12. The world isbecoming a black box driven by machine learning
- 13. ● Highly accuratelevel of optimization ● Budgeting ● Manual Bidding ● Ad Copy Testing ● Shopping Optimization ● Keyword Selection Human- driven ● Revenue ● Budgeting (eg CBO ) ● Smart Bidding ● Responsive Ads ● Smart Shopping ● Keyword Selection Machine-driven
- 14. Sin 3: Relyonly on Google’s Algorithms They will soon be a commodity. 14 3
- 15. “When everybody issuper, no one will be.” Syndrome, The Incredibles
- 16. Algorithms are levelling downthe competition
- 17. Sin 4: Forgetabout Audiences We are people, not keywords. 17 4
- 19.
- 20. Platforms are movingtowards audiences
- 22. Sin 5: Don’tchallenge the data Platforms Silos claim the same revenue. 22 5
- 23.
- 24. 10$ 20$ 25$ Platforms aresiloed and claim the same value
- 25. Sin 6: Relyon third party data A meteor’s coming our way: cookie restrictions. 25 6
- 26. Google Warns Against Blocking‘Cookies’ entirely, triggering Criticism The Wall Street Journal April 2019
- 28. A meteor’s comingour way: cookie restrictions
- 29. Houston, we havea problem! 1. ROAS is Broken. Use Future Margin On Ad Spend instead 2. The world is a machine learning driven black-box. Embrace it 3. Algorithms are a commodity. The advantage is in what you feed them with 4. People, not keywords. Platforms are moving towards Audiences 5. Platforms Silos claim the same revenue. Break the silo in a unified view 6. A meteor’s coming our way: cookies restrictions. Move to first party data
- 30.
- 31. 31 3 Steps toSolve the Challenges 1. Define Success Step beyond ROAS and optimize for future contribution margin.
- 32. 32 3 Steps toSolve the Challenges 1. Define Success Step beyond ROAS and optimize for future contribution margin. 2. Connect Audience Data to Platforms Embrace machine-learning and outsmart competition by feeding algorithms with cookigeddon proofed 1st-party data.
- 33. 33 3 Steps toSolve the Challenges 1. Define Success Step beyond ROAS and optimize for future contribution margin. 2. Connect Audience Data to Platforms Embrace machine-learning and outsmart competition by feeding algorithms with cookigeddon proofed 1st-party data. 3. Connect Business Results to Platforms Build a channel agnostic source of truth
- 34.
- 35. Any online business AcquisitionChannels (eg. FB, Organic, Search, etc.) Your Website Money in the Bank Use first party data to make acquisition smarter
- 37. Let’s zoom inon key elements
- 39. 39 Let’s see areal-life example
- 40. Adjustment based on StandardAudiences Advanced CRM Integration Basic CRM Integration STEP 1 STEP 2 STEP 3
- 41. Let’s consider just2 recipes out of our data puzzle 41
- 42. We tested acouple of clustering approaches 42 KMEAN ML
- 43. We tested acouple of clustering approaches. RFM analysis seemed to be the more promising. 43 KMEAN ML RFM RFM Image Source: analyticalmarketer.io/recency-frequency-monetary-analysis
- 44. We noticed thatCore & High Value clients count for 19% of total Transactions 44
- 45. How can youleverage Your 1st Party Data to boost your Paid Campaigns? 45 Use High Value & Core Clients as audience seeds for Google Ads & Facebook Ads to generate Similar Audiences. 1 2 Increase Bid and/or craft customized messages/campaigns for Promising & those who Need Attention 3 Exclude Lost & New Clients from Paid Campaigns. 4 Apply all Clusters in Observation in GAds to Monitor Performance.
- 46. We focused ourattention in building Seeds from our Top Clients Clusters 46 Use High Value & Core Clients as audience seeds for Google Ads & Facebook Ads to generate Similar Audiences. 1 2 Increase Bid and/or craft customized messages/campaigns for Promising & those who Need Attention 3 Exclude Lost & New Clients from Paid Campaigns. 4 Apply all Clusters in Observation in GAds to Monitor Performance.
- 47. And then wetested Pixel LAL Based vs CRM LAL Based 47 ● What to test against The best audience to test our Custom Audience against has been pinpointed as LAL from the Purchaser ● Geo Boundaries In order to avoid skewed results due to the different distribution of the audiences, we have kept the entirety of the territory to proceed with the test PIXEL CRM
- 48. LAL CRM Audiencehas brought +31% in Revenue and +33% higher ROAS than LAL Pixel Purchase Audience 48 *The experiment results show a 0.10 p-value. This might be considered significant but we do not deem it sufficient to proceed with a direct confirmation.
- 49. FREE SAMPLE RFMQUERY :-) www.boosterboxdigital.com/super-secret-scripts
- 50. Let’s consider just2 recipes out of our data puzzle 50 1 2
- 52.
- 53. 53 When we lookin Google Ads, we usually see just a portion of the truth
- 54. 54 We are blindto the other portion sitting in the CRM RETURNS MARGIN MARGIN ON ADV SPEND
- 55. 55 CRM & GoogleAds Offline Conversions import can help us filling the gap
- 56.
- 57. 57 🔥 BURNING YOURMARING
- 58.
- 59. 59 😇 YOUR HAPPYHAPPY PLACE
- 60. 60 🐘 QUANTITY OVERQUALITY
- 61.
- 62.
- 63.
- 64.
- 65. A 7-step Checklist 1.Get the basics right: UTMs, GTM 2. Pick a database: BigQuery would do 3. Connect CRM data into your database (Funnel, Fivetran, …) 4. Connect Ad Platforms data into your database 5. Define your attribution model 6. Define your Clustering method 7. Push back your audience and LTV data into the Advertising Platforms (eg API Connectors) 65
- 66.
- 67. RFM Analysis 67 LOST -These customers have not purchased from you in a long while. Some of them might have been high frequency and big spenders, but stopped buying at some point. NEED ATTENTION - Customer, who once purchased from you with a medium to high frequency but stopped for some reason a while ago. CORE - Your core group of loyal customers. They are still very valuable as they are regular and recent purchasers of your products. HIGH VALUE - Your most valuable customers. They buy frequently, are spending a high amount on each transaction and are still very active PROMISING - They buy fairly often, but haven’t reached the frequency levels of the Loyal Customers or High Value Customers yet. NEW - They recently had their first transaction with you. As such they obviously will have low frequency score. Even though they can have high monetary scores already, if they are high spenders. GENERAL / OTHER
- 68. RFM Analysis -Methodology The whole analysis process is based on 3 steps: 1. Raw Dataset Creation This dataset provides: N. of Transactions, Clean Margin, First and Last Transaction (from which we get the overall period of activity) and Avg. Transactions and Margin per day. 2. Clients Ranking Client are ranked from 1 to N-of-Clients (excluding those w/ only 1 transaction) based on Recency of the last purchase, Avg. Transactions per Day and Avg. Margin per Day. 3. Clients Clustering The ranks are used to divide each category in 5 groups (based on percentiles). The 124 combinations resulting from the three ranks will form our Client Segments. 68
- 69.