MMM package
Client-owned modeled tables and outputs, model-health documentation, data lineage and reconciliation, and a decision-ready allocation view.
Independent measurement for DTC brands
Independent marketing analytics consultants for DTC & ecommerce
Make the next media-allocation decision with evidence you can inspect.
I’m Eric Brinker. Brinker Analytics is my independent measurement practice for DTC brands.
Marketing measurement consulting, the way I do it, starts with one budget decision. Then I pick the method that fits it and do the work.
So your reports disagree about what’s working. I help you decide what to do next: build a model, run a test, clean up the data, or wait.
This is help with a live budget decision. It isn’t analyst recruiting, a software resale, or a reporting project with no decision at the end.
See if a diagnostic fits your decision70+ incrementality tests designed and run$123M personally launched in digital campaigns~$60M/yr ad spend directed, 2023–202610+ years in DTC marketing analytics
The short answer
Consultant, platform, or agency?
Buy a platform if your gap is collecting the data: server-side events, postback, matching one customer across devices. Let your agency report on running the media day to day. Hire an independent consultant when a budget decision needs a test the agency doesn’t control, or reporting and models built around your own data.
The problem
Every platform claims the sale. So who gets the next dollar?
“Every channel says it’s working. So why aren’t sales up?”
So you pull the platform reports. Each one claims a share of the same orders. Add them up and you may have “sold” more than your store shipped.
You check last-click. It hands the credit to whatever touched the customer last. Usually that’s demand you’d have captured anyway. And the upper-funnel spend? It barely shows up.
Can a pixel tell you which channel caused the sale? Can your agency’s dashboard? Not on their own. They show you what happened near the sale. A test shows what the ad caused.
And before you pick a method, you need to know whether an MMM (marketing mix model: a statistical read of how spend moves sales over time) even makes sense for you yet.
A strong fit
Do you have enough signal to make a decision?
- A DTC brand spending $1M+ a year on ads
- That spend spread across 2–3+ channels, so there’s a real allocation question
- A decision owner ready to act on what a test says
Daily sales history you can break out by region helps a lot. A copy of your web analytics data in your own warehouse is even better. Neither one is a must.
Not enough history for an MMM or a geo test yet? An MTA (multi-touch attribution) built on your own first-party data can still work.
Better to wait
When should you wait?
- Ad spend below that range, or too little history and variation to read
- One paid channel does nearly everything, with no organic baseline or tactical variation to compare against
- Nobody owns the call
Then the next step may be cleaner data, a narrower question, or waiting for a test you can actually run. I’d rather tell you that up front.
The measurement decision diagnostic
What happens before anyone picks a method?
A fixed-scope, paid diagnostic. It finds the smallest useful next move.
It isn’t a pre-sold annual contract. And it isn’t a promise that you need an MMM. Not every team does.
If your gap is ownership, definitions, or reconciliation (getting two systems to agree on the same order), you may need data work before any model.
- 01
Frame the decision
Clarify the allocation question, KPI, conflicting reports, decision owner, and cost of getting it wrong.
- 02
Review evidence & feasibility
Assess the available history, media, outcomes, data gaps, and whether an MMM, testing, foundation work, or waiting is credible.
- 03
Deliver the next move
Receive a feasibility finding, data-gap register, recommendation, assumptions, guardrails, and a 30/60/90-day learning roadmap.
Consultant, platform, or agency
When to hire a marketing measurement consultant
Three kinds of help. They do different jobs. Start with the decision on your desk. Then pick.
When should you buy a measurement platform?
When your gap is collection. Server-side events (tracking sent from your server instead of the shopper’s browser). Postback (sending conversions back to the ad platforms). Cross-device matching (knowing the phone visit and the laptop order are the same person).
That’s infrastructure. Buy a platform built for it. It’s work I’d pass on. But everything after the data lands, I build. Custom, around your data sources, instead of a standard template.
Is your agency’s reporting good enough?
For running the media day to day? Sure. Pacing, creative, bids. That’s what it’s for.
But it isn’t enough to judge whether the media works. The agency is grading its own work with platform and pixel attribution. That’s correlation, not cause.
So for a budget decision, check it with a test the agency doesn’t control. Ideally a geo test (you change spend in some regions, hold others steady, and compare sales).
Already paying for a measurement platform?
I don’t need to replace it. I can check its numbers with a geo test, so you know which of its readings you can trust, and build reporting that puts its output next to your other sources.
When do you need an independent consultant?
When one allocation decision needs a method choice, a test design, or a model review. And nobody on the vendor or media side is paid to question the answer.
No ad platform pays me, so I have no stake in which channel looks best. If I ever recommend a preferred partner, I’ll tell you up front if I’m paid for the referral.
It’s just me. No big-company wrapper: no account team, no billing department. I work from published methods, use AI to move faster, and bring in a statistician when the question needs one. That’s also why I take on a small number of engagements at a time.
I’ve designed and run 70+ incrementality tests over about three years, at the geo and user level, across several channels. Before that, I built an in-house measurement practice at 4Patriots, a $200M+ DTC brand.
What if none of us is the answer yet?
If you’re below the fit above, cleaner data is often the right next step before a consultant or a new tool.
Compare MTA, an MMM, and testing · Check whether a geo test is feasible
The approach
Why build more than one model?
I think of models like weather forecasts. They may disagree. That doesn’t mean I ignore either one.
So I run several reasonable versions of the model on your data. I look for the findings that hold up across all of them, and I flag the ones leaning on weak signal or big assumptions. Response and carryover assumptions (how an ad’s effect builds and fades over time) get tested, not taken on faith.
And when a test covers the question, the test anchors it. An attribution model’s accuracy can only be checked by pairing it with a geo test.
If a call is still shaky, you get an explanation and a next-test plan, not false precision for every channel.
ActEvidence supports a bounded change.
WatchDirection is plausible; use a guardrail.
TestDisagreement matters and can be resolved.
Do not score yetThe data cannot support a useful read.
Engagement paths
A package follows the question, not a fixed software plan.
Experiment package
A test-design brief, feasibility or power assessment, measurement plan, execution guardrails, and a clear decision rule.
Combined path
A deliberate sequence that uses experimentation to calibrate a high-stakes claim and an MMM to guide the broader media portfolio.
Scope follows the readiness finding and the decision at stake. Consultations are used to determine whether the diagnostic makes sense; no fixed prices are published.
What you receive
Decision assets your team can keep using.
Fixed decision report
Findings, uncertainty, recommended actions and guardrails, built for a decision, not an open-ended custom dashboard project.
Channel classification & roadmap
Act, Watch, Test, or Do not score yet, plus a 30/60/90-day action and learning roadmap.
Live working session
A direct readout with the decision owner(s), focused on what changes now and what needs validation next.
Open-issues register
Unresolved questions, unscored channels, assumptions to revisit, and the reason each one remains open.
Client-owned handoff
The report is fixed. The learning system is yours.
Core outputs can live in your own data warehouse. The handoff includes suggested prompts, so your team can ask an AI assistant questions about your own data between readouts.
An optional AI context file and refresh runbook can document definitions, where each table lives, and how to reuse it responsibly. AI can cut some compute and repetitive build cost. But judgment, validation, and getting the team to use it are still the work.
Explore custom multi-touch attributionExplore marketing data foundations
Explore marketing reporting systems
Read why disagreement can be useful
A better decision shift
Which channel deserves the next dollar?
A lower-funnel channel that looks great in last-click reporting may mostly be capturing demand that already existed. There may not be much room to scale it. So the move could be a stricter guardrail on last-click ROAS (return on ad spend) or CPA (cost per acquisition), instead of a big budget bump.
An upper-funnel channel with real incremental effect (sales it caused, not just sales it was near) may deserve a controlled increase, a scenario plan, and a validation test. Even when its click credit looks weak.
Questions to ask before committing
Should you hire a measurement consultant, an agency, or buy a tool?
It depends on the gap. Buy a tool when the gap is collecting the data, like server-side events (tracking sent from your server, not the shopper’s browser). Someone on your team still has to run it and check what it says. Use an agency to run the media day to day. They also grade their own work. Hire an independent consultant when a budget decision needs a test or a model that nobody on the vendor or media side is paid to question. I design the test and build the model around your data. Then I check the result.
Can our agency measure its own media?
For running the media day to day, yes: pacing, creative, bids. But it isn’t enough to judge whether the media works, because the agency grades its own work with platform and pixel attribution. That’s correlation, not cause. For a budget decision, check it with a test the agency doesn’t control, ideally a geo test.
We already pay for a measurement platform. Do we have to replace it?
No. I don’t need to replace it. I can check its numbers with a geo test, so you know which of its readings you can trust, and build reporting that puts its output next to your other sources.
Are you independent of the ad platforms?
No ad platform pays me, so I have no stake in which channel looks best. If I ever recommend a preferred partner, I’ll tell you up front if I’m paid for the referral.
How do you check whether an attribution model is accurate?
Pair it with a geo test. Attribution shows the paths people took. On its own, it can’t prove what caused the sale. So be careful with any vendor that calls pixel-based attribution incremental. And when a geo test and a user-level test are both possible, I lean geo-level.
What does the first engagement cover?
The Measurement Decision Diagnostic. It starts with one allocation question, the reports that disagree, the data you have, and what you could realistically learn. It ends with a bounded recommendation: data foundation work, MTA (multi-touch attribution), an MMM, a test, a narrower question, or waiting.
What do you need from us to start?
Bring a report you don’t trust. Or an area where you want more visibility. Rough revenue and ad-spend ranges, the channels in the mix. You don’t need a perfect dataset.
Why use more than one model?
To see which findings survive reasonable assumptions. Think of it as a model-risk review, not a vote. Where the versions differ in a way that matters, that’s where the next test goes.
What if the models disagree?
That can be useful evidence. A relevant, well-designed incrementality test anchors the causal claim for the conditions it tested. Anywhere else, a disagreement that matters becomes an explained limitation or a prioritized test plan. Not a made-up score.
What if we are not ready?
Then the diagnostic says so. It can recommend data work, a narrower question, an experiment package, or waiting for more variation. A full model build shouldn’t start when it can’t support a useful decision.
A practical first conversation
Bring the next media-allocation decision, not a perfect dataset.
Answer eight quick questions, then pick a time with me. We’ll figure out whether a Measurement Decision Diagnostic fits the decision.
Bring the decision and the reports that disagree about it. The messy version is fine!
By continuing, you agree that Brinker Analytics may use the details you provide to prepare for a conversation. Read the privacy policy.