Attribution In Marketing

Is ROAS the right metric for RMNs? Retail Media Networks (RMNs) have outgrown their early days when untapped demand meant every dollar spent was both high-ROAS and high-incrementality. Today, focusing solely on ROAS incentivizes behaviors that may appear efficient but harm long-term profitability and growth. Here’s how ROAS can be gamed—and why it’s problematic: 1️⃣ Over-spending on Retargeting or Brand Keywords. These tactics drive high ROAS but focus on customers who were likely to convert anyway, resulting in low incremental growth. 2️⃣ Discount-Driven Sales. Discounting boosts ROAS by generating short-term revenue but lowers margins, attracts low-LTV customers, and conditions buyers to expect promotions. 3️⃣ Cutting Spend on High-Incrementality Campaigns. Investing in new customer acquisition or brand building may have lower ROAS but drives long-term growth and quality customer cohorts. These behaviors lead to: ⛔️ Shrinking new customer cohorts. ⛔️ Increased reliance on discounts, reducing margins. ⛔️ Lower customer lifetime value (LTV) and diminished profitability over time. In essence, chasing ROAS at all costs leads to slower growth and declining margins—a losing combination for any business. Efficiency metrics like ROAS are necessary but must be balanced with an effectiveness metric that focuses on long-term outcomes. For example: ✅ 180-Day Contribution LTV: Measure the total revenue contribution from full-price customers acquired over six months. ✅ Incremental Revenue from Non-Brand Keywords: Track revenue generated from truly new demand sources. ROAS is an excellent efficiency metric but a poor north star. Striking the right balance between efficiency and effectiveness will ensure your business scales sustainably while maintaining margins. Keen to hear what other metrics are used for RMNs #advertising #media #tech

Modeled conversions are a weather forecast. They are not facts. They are probabilities. When the forecast says 70% chance of rain, you do not argue with it. You plan around uncertainty. Digital measurement is moving the same way. As signal loss increases, more of what we see in platforms is inferred, not observed. The number looks precise. The reality is a range. Serious operators do not obsess over the single total on the dashboard. They ask what confidence sits behind it. When performance becomes probabilistic, governance becomes the edge. The question is not whether you trust the forecast. It is whether you understand the risk inside it. #Measurement #Attribution #PerformanceMarketing #DigitalStrategy #MarketingGovernance

Attribution is BS. There, I said it. Despite being a past proponent of attribution, I've come to believe that it’s a lie, that we're using flawed math to make critical business decisions… and it’s hurting us. THE ASSUMPTION PROBLEM Every attribution model — first-touch, last-touch, multi-touch, AI-powered — makes fundamental assumptions about buyer behavior: ⊙ What interactions to count (and which not to) ⊙ How far back in time to look ⊙ How to weight different touchpoints But as you may have heard: when you assume, you ‘make an ass out of u and me’. B2B buying is a complex, nonlinear system. Like weather or stock markets, it has sensitive dependence on initial conditions, emergent behaviors, and feedback loops that make precise prediction impossible. Example: A junior analyst downloads your whitepaper but takes no action. Two years later, she’s a Director at a new company and her team faces the exact problem you solve. Your attribution model will never connect that original download to the eventual seven-figure contract. There are millions of examples like this. Attribution pretends buying is a tidy cause-and-effect machine. It's not. Buyers are two-thirds through their process before they engage with vendors. By then, they've often defined needs, shortlisted options, and chosen favorites. The touches that actually influence the deal — thought leadership consumed anonymously, word-of-mouth recommendations, prior experience with your brand — happen long before we can track anything. Yet we give credit to whatever campaign happens to be running when they finally fill out a form, or whichever SDR happens to call at the right moment. We're high-fiving the wrong tactics and teams entirely. THE REAL DAMAGE When teams focus on attribution credit, four things break: 1️⃣ Over-attributing success to demand starves brand and early-stage programs 2️⃣ Short-termism replaces strategic thinking 3️⃣ Marketers optimize for measurable touches instead of buyer experience 4️⃣ The sales-marketing teamwork required for complex deals breaks down I've watched companies gut brand investments because they "couldn't prove ROI" while doubling down on lead magnets that generate terrible experiences but great attribution scores. THE BETTER WAY 𝐔𝐬𝐞 𝐚𝐭𝐭𝐫𝐢𝐛𝐮𝐭𝐢𝐨𝐧 𝐭𝐨 𝐢𝐦𝐩𝐫𝐨𝐯𝐞, 𝐧𝐨𝐭 𝐩𝐫𝐨𝐯𝐞, 𝐲𝐨𝐮𝐫 𝐦𝐚𝐫𝐤𝐞𝐭𝐢𝐧𝐠. Don't claim your webinar ROI is exactly 114%. But you can use attribution to guide decisions, e.g. perhaps webinars seem to perform better than content syndication (at least given a set of assumptions). Focus on directional insights, not false precision. The most successful teams use shared metrics: ✅ Everyone-sourced pipeline ✅ Account progression ✅ Net revenue retention across the full customer journey Stop grading your marketing with broken math. Start guiding it with better questions. What's your take? Is attribution valuable or BS? #B2BMarketing #Attribution #MarketingOps #GoToMarket #MarketingStrategy

We just published new research on the TikTok Halo Effect and the results are hard to ignore. Most brands still measure TikTok Shop in isolation. Platform-level profitability. Did it 'work' on TikTok or not. That approach is fundamentally broken. We analysed aggregated data across TikTok Shop brands to understand what actually happens after someone discovers a product on TikTok. What we found: • TikTok Shop activity and Amazon sales show a strong correlation of ~0.86–0.87 once customer decision timing is accounted for • Amazon sales consistently rise 2–3 days after TikTok activity increases • On average, every £1 of TikTok Shop GMV is associated with ~£0.50–£0.60 of incremental Amazon revenue • TikTok is acting as a demand creation engine, not a standalone checkout channel In short: People discover on TikTok. They often convert on Amazon. And most attribution models miss this entirely. If you are judging TikTok Shop purely on same-day profitability, you are almost certainly underestimating its true impact. We published the full research here 👇 https://lnkd.in/ezWP3j6y This is exactly why cross-channel measurement matters in discovery-led commerce. Would be curious to hear how others are currently measuring TikTok’s downstream impact.

Meta, Google, TikTok, and other ad channels are misleading you. Third-party attribution tools like Triple Whale and North Beam aren't better—they’re flawed too. Tracking has always relied on estimated models, not hard numbers. After iOS 14, tracking became harder, leading to a surge in third-party solutions. But these also provide conflicting data, making it tough to find the truth. So, what is the truth? The only reliable way to measure your marketing efforts is through incrementality tests. These tests answer the question, "What if this channel or ad never existed?" By showing ads to one group and withholding from another, you can measure the true impact on revenue and profit. For example, if you're running Facebook ads and selling on Shopify and Amazon, incrementality tests reveal how Facebook ads impact Amazon sales. Without the initial Facebook touchpoint, an Amazon purchase might not have happened, even though traditional attribution wouldn’t show this. This is why ROAS and third-party attribution aren’t accurate. They use models that can be thwarted by privacy settings and cross-channel purchases. By running incrementality tests, you discover the true impact of your marketing efforts. We ran a 14-day Meta holdout test and found that zip codes shown ads generated 50% more Amazon revenue than those not shown ads, despite sending traffic to Shopify. Now is the perfect time to run these tests. Q3 is calm, free from major holidays that skew results. This is your chance to optimize before Q4. If your brand generates seven figures annually, this should be a top priority to grow profits in Q4.

On the 10 year journey to Chubbies’ IPO, the realization that changed how we invest marketing resources was this --> Increasing ROAS * decreased * our growth. btw, I was the world’s largest ROAS (AKA Return on Ad Spend) fanboy for embarrassingly too long, but hey, my loss is your gain, so here's: 1. Three counterintuitive things I learned about ROAS 2. Two new ways to think about it 3. Three things you can do about this right now let's do it. ** Three counterintuitive things I learned about ROAS ** 1. “ROAS has been presented as a growth metric, when it’s actually anything but. In fact, ROAS is precision-engineered to keep brands small,” says Tom Roach. Chasing ROAS chases easy sales, not growth. Brand growth comes from light buyers, but focusing on high ROAS can lead to you targeting heavy buyers, therefore limiting growth. 2. ROAS is not actually a measure of *effectiveness* but how *efficiently* you achieved it. As Les Binet says: “Effectiveness first, efficiency second.” 3. Simply put, ROAS is the opposite of incrementality. ** Two new ways to think about it ** 1. It's like hiring an employee to stand just inside the entrance of your shop and tap shoppers on the back as they enter. A week later, the employee demand a raise, claiming credit for all the customers they’ve “enticed” to come in. 2. Imagine a soccer coach believing their forward is entirely responsible for every goal. As a result, in their infinite wisdom, they ditch their defense and midfield, only keeping their center forward. They end up losing every future game, but their “Goals Per Player” (the ROAS of this example) is higher than ever! ** Three things you can do about it right now ** 1. Vanity VS Value: Understand the negative externalities of the metrics we goal our teams on. For example, because many of us are seeing headwinds, brands either cut marketing spend or increase the ‘accountability’ of the dollars spent. The negative externality is that we're over-harvesting our existing customers in order to hit our numbers. ROAS and revenue from returning customers may be up (vanity metrics), but contribution dollars, share of search, and new customer revenue from unpaid sources (real business metrics) are likely down. 2. Party & Ponder: Spend half a day with your team and deeply consider the metrics you want to optimize your team’s efforts around in 2024. The whole team needs to take ownership of the metrics that matter AND have a deep understanding of the negative externalities of vanity metrics like ROAS. This is a super high-leverage use of time 3. Cultivate Creativity Completely (the 3C's of winning): Since marketing works by influencing future buyers, think about developing creative that gets noticed and gets remembered. Give your team permission to be bold, put on a show and have a little fun. As John Dawes of the Ehrenberg-Bass Institute says, “The brand that gets remembered is the brand that gets bought." Enjoy

Most ROAS reports are just good storytelling. They look great on slides. But here’s what they rarely tell you: 1. Attribution is broken. That last click didn’t do all the work. Search brand terms? Often piggybacking on organic demand. The real drivers — brand, content, referrals — get buried. 2. ROAS hides the real costs. Returns, discounts, agency retainers, platform fees — conveniently left out of the equation. You’re not tracking profitability. You’re tracking presentation. 3. High ROAS ≠ scalable growth. A 10x ROAS on a tiny remarketing list feels great — Until you try to scale and it crumbles. Low CAC doesn’t mean you have a growth engine. Good marketing isn’t just about showing numbers. It’s about knowing what they really mean.

Attribution has never been perfect, but for DTC brands, it has become significantly harder in the past few years. Apple’s iOS14 updates, third-party cookie deprecation, and increased privacy regulations have disrupted traditional attribution models. Brands that once relied on last-click attribution, ad platform reporting, or rule-based LTV calculations now face major blind spots in understanding which marketing efforts drive long-term value. Even those investing in first-party data strategies, post-purchase surveys, and media mix modeling (MMM) struggle to fully connect the dots. The reality is that data is still fragmented across multiple platforms such as Shopify, Klaviyo, Google Analytics, ad networks, and third-party analytics tools. Most solutions focus on aggregating data, but aggregation alone doesn’t tell the full story of how customers move through the funnel and what actually drives retention. Rob Markey - In his article, "Are You Undervaluing Your Customers?" published in the Harvard Business Review, Markey emphasizes the significance of measuring and managing the value of a company's customer base. He advocates for creating systems that prioritize customer relationships to drive sustainable growth. Chip Bell - Recognized as a pioneer in customer journey mapping, Bell has contributed significantly to the field of customer experience. In an interview titled "The father of customer journey mapping, Chip Bell, talks driving innovation through customer partnership," he discusses how organizations can co-create with customers to drive innovation and enhance the customer journey. So how do brands solve this? 1. Shift from static LTV models to predictive insights - Traditional LTV calculations are backward-looking, often based on averages that don’t account for future behavior. Predictive analytics, using real-time behavioral and transactional data, can provide a more accurate forecast of customer lifetime value at an individual level. 2. Invest in first-party data strategies that go beyond acquisition - Many brands have adapted to privacy changes by collecting more first-party data, but few are fully leveraging it. Loyalty programs, surveys, and on-site behavioral tracking can provide valuable insights into retention and repeat purchase drivers, helping brands reallocate spend more effectively. 3. Adopt AI-driven segmentation and customer equity scoring - RFM segmentation and standard cohort analysis have limitations. AI-powered models can help identify high-value customers earlier in their lifecycle, predict churn risk, and optimize acquisition based on true long-term value, not just early spend. Markey and Bell have long emphasized that customer loyalty isn’t built on transactions alone, it’s about the entire journey. Brands that can better understand and predict customer value will be the ones that thrive in a world where third-party tracking is no longer a reliable option. #CustomerJourney #Attribution #CustomerEquity

Multi-model AI just got smarter. That’s exactly what makes it more dangerous in healthcare. This week, a major AI platform introduced 19 models working in parallel delegating tasks, cross-checking outputs, synthesizing conclusions into something that feels seamless… even authoritative. And that’s where clinical risk begins. In medicine, authority without traceability is a liability. When one model makes a mistake, we can interrogate it. When nineteen collaborate, error becomes distributed. Layered. Abstracted. Harder to see. I’ve reviewed cases where a single flawed assumption cascaded into three additional decisions. Each step looked “reasonable.” Together, they were wrong. In a multi-model system, who owns the flawed premise? Which model introduced the hallucination? Where did synthesis transform uncertainty into confidence? The more seamless the orchestration, the less visible the seams. And in healthcare, seams matter. Because compounded errors don’t stay theoretical. They become delayed diagnoses. Incorrect treatments. Real harm. Multi-agent AI isn’t just a capability upgrade. It’s a supervision challenge. Before deploying orchestration in clinical environments, we should be asking: • How is model attribution preserved across layers? • How do we audit parallel reasoning chains? • What new failure modes emerge at the synthesis layer? • At what point does confidence outpace verifiability? Capability is accelerating. Oversight architecture is not. That gap is where patient safety risk lives. If multi-model AI becomes standard in healthcare, what safeguards would you require before trusting it with your patients? #ArtificialIntelligence #ClinicalSafety #AIGovernance #HealthTech #PatientSafety

Your ad platform is telling you one number. The truth is somewhere else entirely. I ran the same period of data through Google Ads, Meta Ads and independent measurement tools simultaneously. The gaps were not small. Google Ads claimed 9.3x ROAS. Independent measurement showed 3.8x to 6.2x. Meta claimed 11x. Independent measurement showed 2.6x to 6.5x. And GA4 - which most brands are using as their revenue source of truth - undercounted total revenue by 27% versus Shopify. This is not a coincidence. Ad platforms are incentivised to show high ROAS so you keep spending. Their attribution windows are set to maximise credit for their own channel. The practical implication is straightforward: do not make budget allocation decisions based on in-platform ROAS. In this sample, true performance was overstated by between 50% and 330% versus independent measurement. There is a free setup that fixes this. No expensive tools required. Full breakdown in the comments.