.svg)
How to Audit Your Marketing Measurement: The Data-Strength Framework
When marketing measurement isn't working, the instinct is to reach for a new attribution model. The experts don't. When marketing-measurement leader Kelly Zeitlow, who has built and rebuilt measurement systems at Booking.com, Gorillas, and HeyJobs, and is now VP of Marketing at Spread Group, walks into a new company, she doesn't start by picking a model at all. She starts with an audit. Here's the exact framework she uses, so you can run the same check on your own setup before changing anything.
📺 This draws on our fireside chat with Kelly Zeitlow. Prefer to watch the full session? See it here →
.jpg)
The short version: Before you choose or change an attribution model, audit your measurement. The approach Kelly Zeitlow uses — a "data strength" audit, runs on two tracks: examine the data itself (are the KPIs accurate, available on time, and consistent between the dashboards marketers use daily and the ones executives use for strategy — and do they cover the whole customer journey?), and interview the people using it (which KPI do they actually act on, do they trust it, and does the data arrive when it's supposed to?). The real finding is usually the gap between what people believe about their measurement and what's actually happening — and that gap, not the model, is what to fix first.
Why you audit before you touch a model
A new attribution model can't fix a foundation that was never checked. If your tracking is unreliable, your teams don't agree on what "revenue" means, or half your customer journey is invisible, then every model you run on top of that inherits the flaw. That's why Kelly's first move is a "data strength" audit — a look at the actual health of the data before any decision about how to attribute it. The audit runs on two tracks, and you need both: the data tells you what's technically true, the people tell you what's actually trusted and used.
Track 1: Examine the data yourself
The first track is hands-on, you go into the data and check its health against a few specific questions:
- Are the KPIs accurate? Do the core numbers hold up when you trace them, or do they fall apart under scrutiny?
- Are they available — on time? Is the data that's supposed to be ready each morning actually there, or is it quietly broken and delayed?
- Are they visible to the right people? Can the people who need a KPI to make a decision actually see it?
- Do the daily and strategic dashboards agree? Compare the dashboards marketers and analysts use for daily decisions against the ones executives use for strategy. Do they complement each other — or quietly conflict? Conflicting dashboards mean two parts of the company are steering by different numbers.
- Does it cover the whole customer journey? Or is a chunk of it structurally missing — an entire stage or channel that simply isn't measured? Solid event tracking is what closes those gaps.
Track 2: Interview the people using the data
The second track is where the real diagnosis often hides. Kelly interviews a broad range of people: marketers, analysts, stakeholders, and asks questions like:
- What KPI do you actually act on every day? Not the one on the official slide — the one that genuinely drives their decisions.
- Do you have explicit rules? "If it's over this number I make a change; if it's under, I do X." Or is it vibes?
- Do you trust the data? A number nobody trusts isn't driving decisions, no matter how prominent it is on the dashboard.
- Is it delayed — in theory vs. in reality? It's supposed to land by 9am; does it actually show up the next day?
- What's your understanding of the attribution model and how measurement works? Gaps here reveal where the setup is a black box.
- What have you tried before that didn't work, and why do you think it failed? Past failures are a map of the real constraints.
The diagnosis is the gap
Here's the part Kelly says she's no longer surprised by: ask these questions and two people at the same company give completely different answers. Marketers and analysts disagree; stakeholders disagree with both. That divergence — between what people believe about their measurement and what's actually happening in the data — is itself the finding. It's also the thing a new model can never fix, because the problem isn't the math; it's that the organization isn't working from one shared, trusted picture.
What to do with what you find
Once you've run both tracks, the issues you surface tend to fall into a few recognizable buckets, unreliable tracking, numbers no one reconciles, a biased inherited model, or a setup that can't support modern methods. The order of fixes matters: start with the foundation — trustworthy tracking and one agreed definition of each metric, a real single source of truth — before you refine the model on top. A sharper attribution model built on data no one trusts just produces confident, well-attributed nonsense.
And then keep doing it
One last principle from Kelly: measurement is never finished. New channels launch, tactics work until they don't, products ship, and privacy rules keep tightening — so a one-time audit goes stale. The best-in-class teams treat measurement as a continuous investment, re-checking data strength as the business changes rather than setting it up once and walking away. The audit isn't a project; it's a habit.
Run the audit: watch the chat, then check your setup
Kelly walks through this framework and much more — the model biases, MMM and incrementality, and why "CPCs are up" is a symptom, not a cause — in the full fireside chat.
▶ Watch "Marketing Measurement: Yesterday, Today, and Tomorrow" →
Running the audit yourself is a great start. If you'd rather have it done rigorously, our Measurement Audit is exactly this framework applied to your setup — a structured read across your tracking, data layer, and attribution, with a clear picture of what you can trust and a prioritized roadmap to fix what you can't.
Speak with the Adasight team →
Frequently asked questions
What is a marketing measurement audit?
It's a structured check of whether your measurement can actually be trusted — before you choose or change an attribution model. Kelly Zeitlow's version, a "data strength" audit, runs on two tracks: examining the data itself (accuracy, availability, dashboard consistency, journey coverage) and interviewing the people who use it (what they act on, whether they trust it, and what's failed before).
Why audit measurement before choosing an attribution model?
Because a model built on an unchecked foundation inherits all its flaws. If tracking is unreliable or teams don't agree on definitions, no attribution model will produce trustworthy numbers — it'll just attribute bad data more precisely. The audit tells you whether the foundation can support any model at all.
What should you check in a measurement audit?
On the data side: are KPIs accurate, available on time, and visible to the right people; do daily and executive dashboards agree; and is the whole customer journey covered. On the people side: which KPI teams actually act on, whether they trust the data, whether it's delayed in reality, and what they've tried before that didn't work.
Why interview people as part of a data audit?
Because the biggest problems often live in the gap between what people believe about the data and what's actually true. When two people at the same company give completely different answers about which number to trust, that divergence is the diagnosis — and you'd never see it by looking at dashboards alone.
How often should you audit your marketing measurement?
Continuously. Measurement isn't a one-time setup — new channels, tactics, products, and tightening privacy rules all degrade it over time. Best-in-class teams treat it as an ongoing investment, re-checking data strength as the business evolves rather than auditing once and leaving it.
