Latest / The CMO Podcast with Fexingo: Marketing Leadership, Budgets, and Executive Strategy / Why CMOs Are Abandoning Attribution Models
Transcript
- Lucas: Look, if you are a marketing leader right now, you probably have a dashboard that tells you exactly which ad got the sale. But here is the thing: that dashboard is lying to you about half the time. Luna: That sounds dramatic, but I have seen CMOs panic over a drop in last-click conversions when they clearly just launched a massive brand campaign. Lucas: It is not dramatic, it is structural. We are still running twenty-first century attribution logic on a post-cookie, multi-device reality. The standard model assumes that the last person who touched the screen gets credit. That works fine for buying toothpaste online. It fails completely when you are selling a car or insuring a home. Luna: So you are saying the model itself is broken because it ignores the long tail of influence? Lucas: Exactly. And worse, it actively punishes the channels that build trust. Think about a TV spot or a high-end print feature. A viewer sees it, trusts the brand, then goes to Google three days later to search for the product name. The search engine gets the credit. The TV ad gets zero. So next quarter, the CFO cuts the TV budget because it showed zero return. Luna: And that creates a feedback loop where you only fund short-term performance tactics, killing long-term growth. Lucas: Precisely. You starve the top of the funnel. I looked at data from a mid-sized apparel retailer last month. They had eighty percent of their budget in paid social and search. Their last-click attribution said those channels drove ninety-five percent of revenue. But when they ran an incremental lift test, they found that removing those ads didn't hurt sales much. However, cutting their brand video campaigns dropped overall volume by fifteen percent. Luna: Fifteen percent? That is huge. So the video was doing all the heavy lifting, but the algorithm gave the credit to the clicks. Lucas: Right. The clicks were just the receipt. The video was the reason anyone wanted to buy anything. This is what we call the 'dark funnel.' It is everything that happens before the user ever lands on your site. Word of mouth, social shares, even just seeing a billboard on their commute. Current tools cannot see that. Luna: But isn't that changing with new privacy laws and the death of third-party cookies? Lucas: That is the pressure cooker. Apple’s privacy updates and the GDPR restrictions mean we are losing granular tracking data every single day. You can no longer follow a user across ten different apps to see their path. So, relying on click-based attribution is like trying to navigate a ship with a map that is actively burning away. Luna: So what is the alternative? Do we just guess? Lucas: No, we move to Media Mix Modeling and Incremental Lift Testing. These are older methods, honestly, dating back to the eighties, but they are coming back because they are more robust. Instead of tracking individuals, you look at aggregate data. You correlate your spending across all channels with total sales over time. Luna: That feels less precise though. How do you know which specific ad creative worked? Lucas: You don't need to know which specific ad worked to know that the channel worked. MMM tells you that when you spend one million dollars on broadcast, sales go up by two hundred thousand. It smooths out the noise. It handles the dark funnel perfectly because it looks at the whole picture. Luna: But MMM has a lag. It takes weeks to get enough data to run the model. Isn't that too slow for dynamic markets? Lucas: It is slower, yes. But it is also more accurate for budget allocation. For tactical tweaks, you use incremental lift tests. You show an ad to half your audience and hide it from the other half. Then you compare conversion rates. That gives you real causal proof, not just correlation. Luna: That sounds expensive to run continuously. Small companies can't afford randomized controlled trials on every campaign. Lucas: They shouldn't try to. They should pick their biggest risk areas. If you are spending fifty thousand dollars on Facebook, run one test there. Prove the incrementality. If Facebook adds value, keep it. If it doesn't, cut it. Stop guessing. Luna: I think a lot of marketers resist this because it requires admitting they don't have full visibility. It feels vulnerable. Lucas: It does feel vulnerable. But hiding behind a dashboard that says you are efficient while your actual sales plateau is worse. I spoke to a CMO of a fintech startup who told me they realized their referral program was driving forty percent of signups. Their digital ads were driving zero incremental users. They moved all the budget to incentives and grew twice as fast. Luna: Wow. So the referral program was the hero, but the ads were just getting the credit because they were easier to track. Lucas: Exactly. Ease of tracking is not the same as business impact. This is the core mistake. We optimize for what is easy to measure, not what moves the needle. Luna: If these marketing conversations have sparked something you've actually used, we appreciate your support. It keeps the show independent and free for everyone. Honestly, if today was worth a coffee to you, that's the link — buy me a coffee dot com slash fexingo. Lucas: Thanks for keeping us going. It helps us dig deeper into cases like this without needing corporate sponsors telling us what to say. Luna: Yeah, and it means we can stay critical of bad practices without fear of losing ad revenue. Lucas: Back to the data. One of the biggest hurdles for CMOs adopting this shift is internal politics. Your sales team loves last-click attribution because it makes their closing skills look better than the marketing lead generation. Luna: Right, the classic turf war. Sales wants credit for the close, marketing wants credit for the click. Everyone loses. Lucas: So the solution is unified metrics. First-touch, middle-touch, last-touch. Or better yet, contribution margin per customer. You stop looking at individual channels and start looking at the customer lifetime value generated by each acquisition source. Luna: That requires finance to work closely with marketing. Not many organizations do that well. Lucas: True. But the ones that do, like Salesforce and HubSpot, treat marketing as a revenue center, not a cost center. They use these advanced models to prove that every dollar spent brings in four or five dollars back. That changes the conversation entirely. Luna: It shifts the power dynamic. If you can prove revenue impact, you get more budget. If you can only prove clicks, you get squeezed. Lucas: Exactly. And the market is rewarding that sophistication. Investors are looking at marketing efficiency ratios now. They want to see CAC payback periods and LTV:CAC ratios, not just vanity metrics like impressions. Luna: Impressions are so yesterday. Nobody buys a house based on how many times they saw a banner ad. Lucas: No, but they might search for the company after seeing a documentary sponsorship. That documentary is invisible to last-click. It is visible to MMM. It is visible to common sense. Luna: So the takeaway is: stop trusting the default dashboard. Lucas: Start testing. Run small experiments. Trust the data from lift tests, not the algorithms designed to sell you more ads. Luna: It is a harder path, but it is the only one that leads to actual profit. Lucas: Agreed. The future of marketing is not about tracking every pixel. It is about understanding human behavior in a complex world. Luna: And accepting that some of the most important moments happen off-screen. Lucas: Which is why we need better models. Models that respect the complexity of the journey, not the simplicity of the click. Luna: Let's see if that becomes the standard next year. Lucas: I hope so. Because right now, we are flying blind and calling it navigation. Luna: Here is to finding our way.