Latest / Marketing Automation with Fexingo: HubSpot, Marketo, and Automated Customer Journeys / Why Your Marketing Automation Needs an Attribution Model Overhaul
Transcript
- Lucas: Luna, I want to talk about something that I think quietly undermines a lot of marketing automation setups, and it's not data quality or deliverability for once. Luna: Okay, I'm curious. What's the culprit? Lucas: Attribution models. Specifically, the fact that most marketing automation platforms default to last-touch attribution, and that distorts pretty much every decision you make about campaigns, content, and sequencing. Luna: Right, so the classic problem — the last click gets all the credit, even though the buyer probably saw a blog post, a webinar, and three emails before that click. Lucas: Exactly. And it's not just unfair — it actively misleads you. A few months ago I looked at a B2B SaaS company that was running a pretty standard HubSpot automation workflow. They had a top of funnel blog content series, a mid-funnel email nurture, and then a bottom-funnel demo request sequence. Lucas: When they looked at their default last-touch report, the demo request emails were showing a 12 percent conversion rate to opportunity. The blog content showed basically zero. So naturally, they were about to double down on optimizing those demo emails — more send frequency, more variation. Luna: Let me guess — when they actually built a custom attribution model, the story flipped. Lucas: Completely. They set up a weighted multi-touch model in their CRM, and it turned out that the blog content was actually the primary influence on 60 percent of closed-won revenue. People would read two or three posts, then maybe a week later click a demo email. Last-touch gave the email all the credit. Luna: So they were basically optimizing the wrong thing. That's a huge waste of time and budget. Lucas: It is. And the fix isn't that complicated, but it does require some thinking upfront. You need to decide on an attribution framework that matches your actual buying cycle, and then configure your automation platform to pass the right touchpoint data. Luna: What are the main frameworks people typically consider? Lucas: There are a handful. First-touch gives all credit to the first interaction, which is useful if you care about top of funnel sources. Last-touch, as we said, credits the last click. Then you have linear, which splits credit evenly across all touches. Time decay gives more weight to touches closer to conversion. And position-based, also called U-shaped, gives 40 percent to first touch, 40 percent to last touch, and spreads the remaining 20 percent across middle touches. Luna: And which one is best for an automation-heavy setup? Lucas: Honestly, none of the off-the-shelf models are perfect. For B2B especially, I think a custom weighted model works best — where you assign percentage weights based on your actual sales cycle data. The company I mentioned ended up with a model that gave 30 percent to first touch, 20 percent to any form fill or event attendance, 30 percent to the last touch before a demo, and 20 percent to the demo itself. Luna: That makes sense — it reflects that the first piece of content sets the stage, but the demo is also crucial. How do you implement that in a platform like HubSpot or Marketo? Lucas: It's not native in either platform, but you can hack it. In HubSpot, you can use custom report builder with attribution enabled, but you need to make sure you're tracking all your touchpoints properly — not just form submissions, but page views, email clicks, and ad interactions. In Marketo, you'd typically use the Revenue Cycle Analytics module, but that's an add-on and requires a lot of configuration. Luna: So it's doable, but it's not a check-box feature. Lucas: Right. And that's why most teams don't do it. They just accept the default last-touch report and make decisions based on that. But the consequences can be significant. For example, if you're optimizing for last-touch, you'll naturally bias toward bottom of funnel offers that close quickly — but you'll starve the top of funnel content that actually fills the pipeline. Luna: I've seen that happen. Teams kill their blog budget because 'it doesn't convert,' and then six months later their pipeline dries up. Lucas: Exactly. And here's another pitfall — double-counting. If you have multiple attribution systems running, like your marketing automation platform and your CRM both trying to assign credit, you can end up with conflicting numbers and a lot of confusion. Luna: Yeah, I've heard horror stories where the CRM says a deal was influenced by a trade show, but the automation platform says it was an email sequence. And neither is wrong, but they both claim 100 percent credit. Lucas: That's a classic sign you need a single source of truth for attribution. Ideally, you'd want your CRM to be the system of record, and your marketing automation platform to pass all touchpoint data to it. Then you run your attribution model in the CRM. Luna: And that requires a clean sync between the two systems, which is easier said than done. Lucas: Absolutely. You need to make sure every meaningful interaction is logged as a campaign member or activity record, and that your CRM's attribution settings are configured to use that data. It's an investment, but it pays off when you start seeing which campaigns truly drive revenue. Lucas: One more thing — time decay matters more than most people realize. In a six-month sales cycle, an email sent in month one is probably not as influential as one sent in month six. But linear or first-touch models don't capture that. Luna: So you'd want to apply a time-decay weighting, where touches closer to the close get more credit. Lucas: Right. And you can do that with a simple formula: assign a decay factor, like 0.7 per touch, so the most recent touch gets full weight and each prior touch gets 70 percent of the previous weight. It's not perfect, but it's better than ignoring time entirely. Luna: I think a lot of marketers avoid this because it feels like moving from 'reporting' to 'analytics' — it's more complex and requires buy-in from sales and leadership. Lucas: It does. But the alternative is flying blind. If your automation platform is telling you that your demo emails are the hero, but your pipeline is actually driven by your blog, you're making resource allocation mistakes every quarter. Luna: So the takeaway is: don't trust the default attribution. Build a model that matches your cycle, and sync the data properly. Lucas: And if that sounds like a lot of work — it is. But the ROI on getting attribution right is enormous. You stop wasting budget on campaigns that look good in-platform but don't actually drive revenue. Luna: On that note, if these marketing conversations have sparked something you've actually used or improved in your own setup, we'd appreciate if you considered supporting the show. Lucas: Yeah, it's a small thing, but listener support is what keeps this podcast ad-free and independent. If today's episode was worth a coffee to you, the link is buy me a coffee dot com slash fexingo. Luna: Exactly. No pressure, just a way to keep these conversations going. So back to attribution — any final thought on how to start? Lucas: I'd say start small. Pick one campaign or one product line, map out the typical buying journey touches, and build a simple spreadsheet model. Then see how the results differ from your platform's default report. That one exercise will probably convince you to invest in a proper setup. Luna: Good advice. Thanks, Lucas. Lucas: Thanks, Luna.