Marketing Attribution Models

Just because "google" shows up in attribution doesn't mean it's what driving buyers to buy. Ask "how did you hear about us" in a free-text required field upon conversion and you'll get the real stuff: -Social media (LinkedIn, Tik Tok, Reddit, Instagram, etc.) -Podcasts (owned, earned, paid) -Communities (Slack, discord, private groups, etc.) -Referrals / Word of Mouth (colleagues, friends, investors, etc.) -3rd party events (e.g. I saw your CEO speak in Belgium last summer) ^^These insights will RARELY or NEVER show up in attribution software. Most B2B companies never ask this question. And most B2B companies don't actually know what's creating their demand. #attribution #revenue #sales #marketing #b2b p.s. This is not meant to be a replacement to digital touchpoint based attribution. It's a different measurement strategy used for a different purpose - to know what buyers report as the most *impactful* touches. p.p.s. Self reported attribution is a *directional* insight that you get directly from customers. Many marketing activities will not get measured by touchpoint based digital attribution and we need another strategy to measure these - podcast, social media, connected TV, Out-of-Home (OOH), referrals, influencer marketing, word of mouth, etc. p.p.p.s. Most companies don't get value from self-reported attribution because they don't use it properly. Require it for all declared intent submissions. Copy it from the lead/contact to opportunity object. Track the results against qualified pipeline and revenue, not just "leads". p.p.p.p.s. Self-reported attribution is 1 of 6 different measurement strategies we use at Passetto to analyze the impact of all GTM Investments. A one-size fits all approach of using touchpoint based digital attribution to measure all Marketing, Sales, and SDR investments is a losing strategy.

Marketers are skeptical of attribution models. And honestly, they should be. Most are built on shaky assumptions, like giving all the credit to the last ad someone clicked before buying. Many are black box. I'm always in search of research on better ways to measure marketing's impact. So this week on The Marketing Architects Podcast we covered a study titled, "Bayesian Modeling of Marketing Attribution" by Ritwik Sinha and David Arbour from Adobe Research and Aahlad Manas Puli from NYU. The researchers modeled customer journeys probabilistically, looking at things like ad decay, exposures across different channels, and purchase probability. All of that came together to change the chance of a sale over time. One finding: When users saw more than 20 ads in a short window, the chance of a sale went down. Another takeaway: Search and display ads had extremely short half-lives. Their influence faded fast. The model also assigned strong credit to owned and offline channels, which traditional digital attribution methods often ignore. (❤️📺) The Bayesian model doesn't just assign credit, it gives us a sense of how much a channel mattered, how long its effect lasted, and how confident you should be in the results. Even if your brand isn’t ready to adopt a model like this, it's interesting to learn about. And backs up why it's important to invest in multiple models and perspectives. Links in the comments to listen to the podcast + read the study.

I’ve mapped out the framework I use to build a custom multi-touch attribution system in HubSpot. 1. The Data Layer (Custom Objects) HubSpot's native Campaigns tool isn't robust enough to support the level of granularity most companies need from multi-touch. By using 3 Custom Objects for Marketing Campaigns, Channel Campaigns, and individual Touchpoints, you create a structured hierarchy. This allows you to track spend and ROI at both the macro and granular levels. 2. The Operational Layer (Workflows) Automation handles the heavy lifting so you can focus on launching campaigns. Campaign Initialization: Select your channels and have a workflow generate Channel Campaign records and unique UTM tracking links automatically. Touchpoint Capture: Real-time triggers (like form fills) that create Touchpoint records and pull in UTM data (or anything else you might want to capture). Association: Automate the essential link between those Touchpoints and the Deals they influenced. 3. The Results Layer (Reporting) By associating touchpoints with deals, you can calculate: Influenced Revenue: Total deal value touched by a campaign. Attributed Revenue: A slice of the pie based on the number of touchpoints associated. ROI: How campaign spend compares to attributed revenue Stop guessing which channels are driving growth and start measuring the actual influence of every touchpoint. Want to master this framework? I cover it and several others in Attribution Academy's Mastery Certification Course. Sign up here: https://lnkd.in/erY2QhA3

Incrementality Factors - one of the easiest concepts to grok in theory, and most difficult to implement in practice. Some challenges we've encountered over the past few years at Haus: 1. Calibrating on ad platform reporting gives many marketers the ick due to strong conditioning to not believe it. While I think this is misguided, customers often lose interest in incrementality-adjusted attribution when we mention calibrating on platform metrics. Many want to calibrate their daily source of truth which might be MTA or last click, not platform. 2. Modeling on top of platform reporting means that changes in platform attribution definitions (e.g., Meta's recent attribution changes) create too much inconsistency. 3. Incrementality Factors are too blunt and go stale as you move through different moments of seasonality and make changes to their campaigns, media mix and even product/offering. 4. KPI Incompatibility: Customers measure incrementality on Shopify KPIs like New Customer Orders/Revenue but many teams do not instrument their Meta events to track new vs. returning orders, making it difficult to link test results to platform metrics. At Haus, we’ve taken all the learnings and have been heads down reimagining Causal Attribution. What we’re solving for: 1. Our new Haus pixel collects conversion events and marketing touchpoints directly so you do not have to rely on platform reporting. 2. Incrementality Index: ML model predictions fill in the gaps for channels/campaigns without experiments. The ML model is trained on thousands of GeoLift experiments to predict incrementality based on vertical, spend level, platform, tactic, and funnel position. 3. Regular auto-refresh and constant calibration: factors auto-refresh based on the ML model which prevents calibration from going stale as the media mix changes and more causal evidence is generated. 4. Time-Varying Factors: To solve for factors changing over time, Causal Attribution will build in time-varying factors (similar to how we solved this in MMM) rather than using 1 blunt, static factor from a single test. 5. Granular Reporting: This system allows for a cleaner linking of your Geo-Lift KPIs and your pixel conversion events. Want to learn more? We'll be unveiling our Causal Attribution product at our Haus Growth Lab Roadshow events in May.

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.

If you are still measuring your sales attribution solely based on last touch, you need to get with the times. Last touch attribution is a flawed measurement that fails to consider the entire customer journey, and its usage is outdated given the vast amount of data you have access to through AMC. Using AMC, you can see, for example, that maybe 100 sales had the same conversion path - a sponsored brand ad, a DSP ad, and a sponsored product. That's one example of a pattern you can begin to pick up on. Then you can see what the rate of purchase is based on how many ad types people interacted with and then which particular ad type they interacted with to buy. You can start to change the attribution model to give credit to some of those ad types that the customer interacted with that were not the last touch attribution. You start to see how different ad types work together. This model is especially helpful particularly when you're driving brand awareness or upper funnel activity. Last touch attribution was the only model before AMC, but the ability to see the entire customer journey is one of the things that makes AMC so powerful. #Amazon #AMC #AmazonSellers

New Metric on LinkedIn Ads: “Conversions (Data-Driven Attribution)” If you’re running LinkedIn campaigns, you may have seen this option show up in the performance view: - Conversions (Data-Driven Attribution) (and similarly for Leads). This is still new, and I had to look up some info on it, but here’s what to know: What it is: This metric uses a machine-learning model to allocate credit across multiple ad touchpoints in the buyer’s journey - not just the last click or last view. What’s different / what to watch: It only appears in the performance chart view, so you may not (yet) see full campaign- or ad-level breakdowns. Because it’s giving “true contribution” rather than full touch counts, you might see lower numbers than you’re used to with “Each” or “Last-Touch” models. For longer B2B journeys (multiple ads, impressions, clicks) this gives a more realistic assessment of which ads actually drove conversion. Why you should care: You can now lean in on the metric that better reflects impact, not just activity. It lets your audit/strategy discussions shift from “how many touches” to “which touches were meaningful”. Suggested action: Continue tracking your standard conversion metrics (Each, Last Touch) for context. Add the DDA metric in your reporting and call out the difference: more realistic contribution vs broad exposure. Use this as part of the conversation: “Here’s what the model shows as true performance, here’s what our touch-path exposure looked like, so here’s where optimizations go.” Note the limitation: if you can’t break the DDA metric down by campaign or ad yet, flag that in your audit and plan for when deeper granularity becomes available. - - This is one of several new features I’ve seen rolling out lately, but they also seem to be immature or not ready for full rollout yet. I want to share this in case anyone else has any additional insights they can contribute here, or so you can check it out for yourself if it’s available! #linkedinads #newfeature #b2bmarketing

Marketers, are you still measuring email the old way? We get told email is dead, but everyone reading this has most likely read an email, logged in using it & made a purchase with it. So it's not dead, but how we judge its effectiveness hasn’t evolved fast. We’ve relied on open rates & click-through rates (CTR) — metrics that, frankly, are no longer fit for purpose. Why open rates are no longer reliable Open tracking depends on image loading, which Outlook often blocks, & Apple & Gmail preload by default. As a result, you might see machines open, not human ones. And proper visibility is vanishing with more “text-only” creatives or image-blocked environments. And CTR? It’s got its own problems Think about user intent. If a customer reads “50% off this weekend” in your subject line, they may just go straight to your site—no click needed. Even Gmail’s AI summarising content & extracting voucher codes means users engage without clicks. Email is quickly becoming a powerhouse for brand awareness, but it doesn't have the metrics to prove this. So, what should we look at? As the rest of adtech races toward incrementality, attention, and post-impression attribution, email needs to catch up. Here’s how: 1. Conversion Attribution (Beyond Last Click) Don't stop at click-based conversions. Track who received the email, & assign influence weightings to openers, clickers, & even non-clickers who later convert. This mirrors how display and social now assess "view-through" impact. 2. Frequency & Multi-Touch Engagement Did the recipient open on mobile in the morning, revisit via desktop, & convert on payday? That’s a multi-touch journey. Look at repeat site visits, device switching, & re-engagement post-send. 3. Pay Day or Trigger-Based Lift Create holdout groups and measure uplift around high-conversion moments (e.g., end-of-month). This mirrors the incrementality testing often used in paid social or programmatic, proving that email drives behaviour, not just volume. 4. Attention Metrics Use tools to estimate dwell time on emails or the time between opening& clicking. These are soft proxies for intent, similar to how platforms measure scroll depth, hover rate, and ad exposure time in other channels. 5. Site Quality Metrics Did email recipients spend longer on site, view more pages, or have higher AOVs? Your session quality tells you if email delivers high-intent traffic, something brands already monitor from Google Ads or affiliates. 6. Ask them! Simple, but powerful: survey your audience. What emails did they find valuable? Did it change their behaviour? Self-reported attribution, done well, can give you what click-tracking can’t. Email deserves more credit than. If adtech is shifting toward attention, incrementality, & deeper behaviour analysis, email should, too. Let's measure actual impact, not just opens & clicks. I bet you will discover that email isn't just for conversion but also a branding-building superpower.

I see this mistake constantly, and it's costing brands money. Teams have built entire strategies around attribution models that don't tell the whole story. Last-click attribution says your paid search is working. Your ROAS looks incredible. The budget flows toward search because the data says it's the top performer. But here's what last-click attribution doesn't show you: the awareness campaigns that put your brand in consideration. The organic content that built trust. The email that reminded someone to convert. Attribution is useful. But it is not true. Brands that rely solely on platform attribution often make three expensive mistakes. First, they underinvest in upper-funnel work. Awareness, consideration, brand building – these don't show up in last-click models. So they get starved of budget. Revenue growth slows. Cost per acquisition rises. And nobody understands why. Second, they overinvest in hyper-targeted short-term activity. Lower-funnel channels look amazing in attribution reports. So the budget concentrates there. But eventually customer acquisition costs rise because you've removed the awareness foundation that feeds the funnel. Third, they miss the long-term compounding effects. Brand signals operate on different timelines than clicks. Mental availability builds over months. Consideration compounds over time. But if you're only measuring last-click, you're optimising away the work that compounds. Here's what actually works: measure both layers. Measure your short-term performance metrics. Conversions, revenue, customer acquisition cost. These matter. But also measure the leading indicators of long-term success. Brand awareness, consideration, customer retention. Measure them separately. Track them over time. Build strategy around both. Stop treating attribution models as complete pictures. Start measuring what they miss. What's your biggest blind spot in how you measure marketing?

We tested Meta’s new incremental attribution model. The results broke a few assumptions. Out of 4 campaigns last week, 1 was optimized for incremental attribution. The other 3 were ASC and ABO campaigns: And yet: - 1-day click CPA was 48.2% lower than the ASC - Incremental CPA was 49.0% lower than the ASC - New audience ROAS was atleast 55% higher than the ASC Why does this matter? Because 1-day click is the most conservative attribution model, it only counts conversions that happen fast and post-click. It’s historically been the cleanest proxy for incrementality. So lower 1DC CPA is exactly what short consideration DTC products need - faster conversions, lower CAC, and better cash efficiency 🚨 How It Works: Meta now runs always-on holdout tests in the background. It splits users into treatment and control groups and continuously asks: “Did this person convert because they saw the ad, or would they have purchased anyway?” The difference between those groups is considered incremental lift and that becomes the basis for attribution, optimization, and reporting. It’s not about what happened within a time window anymore. It’s about what happened because of the ad. 🚨 Why This Matters - Eliminates the guesswork of choosing between attribution windows - Shifts focus from tweaking settings to scaling what actually works - Enables budget consolidation and simpler account structures - Reduces dependency on exclusions and segments, Meta already accounts for it - Moves us toward causal measurement inside the algorithm itself We’ll continue to validate this. But so far, incremental attribution has outperformed our default benchmarks. Including ASC.