Dominic Bonaker started building websites for small businesses in the UK before the pandemic pushed him toward the data behind those campaigns. Today he runs Lunar, a data partner for marketing agencies, and he joined Deconstructing Data to make the case that most attribution models are built to flatter ad platforms, not to tell you the truth.
Here’s the thing about last-click attribution: it’s not broken by accident. Dominic Bonaker, founder and CEO of Lunar, put it plainly on this week’s episode. Google Ads gives credit to the last click. Meta gives credit to the last click. That’s not a coincidence, and it’s not really measurement either.
“The reason that they do that is cuz it incentivizes them for you to incentivizes you to spend more money on the platform cuz it looks like those are the ones that are driving results.”
— Dominic Bonaker, Founder and CEO, Lunar
That’s a big deal once you sit with it. The platforms grading your campaigns are the same platforms getting paid based on the grade. And most businesses never question it, because last-click is the number that shows up in the dashboard by default.
What’s wrong with last-click and first-click attribution models?
Bonaker’s issue isn’t that last-click is useless, it’s that it ignores everything that happened before the final tap. Brand awareness campaigns, the ones that build the demand a last-click ad later “captures,” get zero credit under that model. He points to New Balance as the case study he keeps coming back to: a previous CEO ran the brand at roughly 70% direct-response performance marketing and 30% brand, and results kept sliding. A new CEO flipped the ratio to 70% brand and 30% direct response, and the brand’s visibility took off.
First-click attribution is a step in the right direction because it at least credits top-of-funnel activity, but Bonaker isn’t a fan of that model either. His actual answer is what he calls scientific attribution: ranking every touchpoint in a customer’s journey by how much it actually mattered, based on real behavioral data like time spent on a page or whether someone watched a video before converting.
“We’re looking across the entire um entire conversions in the journey that a customer goes through, and then we’re ranking those individual touchpoints based on the importance that we deem those touchpoints.”
— Dominic Bonaker, Founder and CEO, Lunar
The practical payoff is that once you know which touchpoint actually correlates with your highest average order value, you can engineer more customers through that same path. Bonaker’s example: if people who watch a two-minute video before buying consistently spend more, you don’t just note that. You build more videos into the funnel.
Why don’t more businesses build their own attribution data?
Bonaker’s answer here is refreshingly honest: most businesses don’t build data infrastructure because they’re busy, and because it’s easier to hide from bad results than confront them.
“I think they’re also scared of the results because it’s easier just to kind of hide away from the actual results of of how business is performing. And if the data is complicated, you can step away from it.”
— Dominic Bonaker, Founder and CEO, Lunar
He ran a LinkedIn poll asking marketers how they’d decide what to cut if their budget was slashed by 50%. A large chunk answered “gut feeling” or “previous experience.” That’s the whole problem in one data point. If you’re not measuring, you’re not deciding, you’re guessing with confidence.
How do you build marketing reports that actually change decisions?
This is where Bonaker’s advice gets tactical. His litmus test for any metric on a report: if this number changed tomorrow, would you actually do anything differently? If the answer is no, it doesn’t belong in the weekly report.
“If your if your CAC decided to go at your customer acquisition cost decided to double overnight, well, you best believe that you’re going to do everything in your power to understand why because that is fundamental to the business. That should be in reporting.”
— Dominic Bonaker, Founder and CEO, Lunar
Vanity metrics like likes and impressions don’t clear that bar. LTV, CAC, and the ratio between them do. And when it comes to actually presenting the numbers, Bonaker has a rule I hadn’t heard framed this cleanly before: a number with a dollar sign in front of it lands harder than a multiplier.
“A number with a a dollar sign in front of it is always going to be more impactful than a you know, an a 4X kind of number because there’s more context to it.”
— Dominic Bonaker, Founder and CEO, Lunar
Telling a client “we turned your $100,000 into half a million” beats “we delivered a 5x ROAS,” even though they’re the same math. That’s not spin, it’s translating marketing numbers into language a CFO already speaks, since finance people and marketers are, as Bonaker put it, “talking about two different types of minds.”
How should marketers use AI in reporting without losing accuracy?
Lunar puts a disclaimer on every report it ships: “Prepared by AI, reviewed by a real human.” Bonaker said clients respond well to the transparency, and the AI has genuinely sped up the work, cutting reports that used to take two hours down to twenty minutes of proofreading instead of writing from scratch.
But he’s clear-eyed about the limits. Ask any LLM why a metric moved and it will hand you ten or fifteen plausible reasons. Only someone who actually understands the business can connect a CAC spike to the specific offer change that caused it.
“An AI model is going to find it really hard to, you know, reverse engineer what has happened unless it is, you know, plugged into the business 24/7.”
— Dominic Bonaker, Founder and CEO, Lunar
His broader experimentation framework is worth stealing too: put 70% of budget behind what’s already proven to work, 20% into adjacent variations of that (a new hook, a different headline), and 10% into genuinely wild ideas you don’t expect to work. That last 10% is where the breakthroughs tend to hide.
The bigger picture
What struck me most in this conversation is how much of “good attribution” comes down to a willingness to look at uncomfortable numbers. Bonaker isn’t selling a magic model, he’s selling discipline: track everything, including the changes you make along the way, tie every metric to a decision someone would actually make, and use AI to speed up the grunt work without letting it replace the judgment call.
That’s the same philosophy behind identity resolution and clean data more broadly. The platforms have every incentive to make their own numbers look good. The businesses that win are the ones building their own source of truth instead of renting someone else’s scoreboard.
To learn more about Dominic Bonaker and Lunar, visit asklunar.com or connect with him on LinkedIn.
For companies that need clean identity data to power smarter decisions, visit bdex.com and click “Talk to an Expert.”
Watch the Episode
Data Attribution, Reporting, and AI-Powered Insights
This article was adapted from an episode of Deconstructing Data, BDEX’s weekly podcast on data-driven marketing. Tune in live every Thursday at 4:15 PM Eastern on LinkedIn.
About BDEX: For companies that need clean identity data to power their products, BDEX offers unmatched quality and execution. Visit bdex.com and click “Talk to an Expert” to get started.
