Updated October 23, 2026

How do you judge a measurement partner?

Look for business context first and stats skill second. A strong partner bends to your data and your KPIs. They don’t lock you into an annual contract or a percent-of-sales deal. They talk on your schedule, reply within 24 hours and let you reach the people doing the work. And in the first 30 days, they hand you a test design you can push back on, plus a written list of what the method can’t measure.

Full disclosure: I run Brinker Analytics, and I sell this kind of work. So weigh my take with that in mind.

What should you look for?

Business context. A stats background is nice. But AI tools can now double-check a lot of the math. What they can’t do is know your business.

I’ve worked with stats experts at big companies who were sharp. Many still lacked a nose for the exceptions that make a business tick.

One vendor couldn’t see why I’d compare revenue across test cells. They wanted to assume an AOV (average order value) and compare order counts. That works for a high-volume brand with a steady AOV. It breaks for a direct-response brand whose hero products sell at very different prices. It also breaks for a brand with loss leaders and big upsells. There, an order is not a profit.

Flexibility. Plug-and-play tools are nice. They become a problem when they’re rigid and built for an average business. If an offer leads with a beautiful dashboard, ask yourself: was it built for what most businesses need, or for what you need?

No handcuffs. Watch for annual contracts and percent-of-sales pricing. Measurement uses the past to guess the future. So a partner who stakes a claim on your future sales, based on a great rearview mirror, is putting a lien on your business.

Flat-fee, one-off or short-term deals don’t mean lower quality. They can lower the risk of a bad ROI.

What are the red flags?

Missed deadlines. Picture this: a geo test is set to start. Then it slips a surprise week while they wait on data. Then another week for the review meeting. Now you’re bending your plans around their calendar. It should be the other way around.

No room for your own data. Your best supporting metrics often live outside the tool. For example:

None of these replace the test read. They make good supporting metrics and help tell the story. If a partner can’t use them, that’s a flag.

Email and scheduled Zoom only. I like a lot of communication. If there’s no Slack (or something like it), small questions wait days.

No answers on the method’s limits. One partner used a zip-to-DMA lookup table instead of GA4’s IP-based metro (DMA) field. GA4’s field is also approximate, but it’s closer to where the visit happened. Where the partner’s table came from, and how fresh it was, wasn’t clear. Some zip codes straddle two DMAs. And a billing or shipping zip may not be where the ad was served.

MDE (minimum detectable effect, the smallest lift a test can reliably pick up) is a game of inches. That kind of error matters. The partner wouldn’t consider it.

How often should you hear from them?

As often as you need. That might be a weekly call, on-demand check-ins, or lots of light async notes in Slack.

At minimum, replies should come within 24 hours.

How often they review depends on what’s being reviewed. A geo test in flight gets checked every couple of days at first. Then weekly.

And you should get direct access to the people doing the work. Not a customer success rep passing notes back and forth.

What should the first 30 days produce?

  1. A data audit. What you have, what’s missing and what can be trusted.
  2. A test design you can push back on. A draft, not a decree.
  3. A written list of what the method can’t measure. If it’s not written down, it tends to get forgotten.

What makes a partner outstanding?

They co-design the program with you. It starts from your expectations and goes back and forth from there. The goal is a program that fits your business and that you’ll actually run. A partner can set the pace and lead. But you’re in charge.

They only pitch tests leadership would act on. Say leadership can’t run a true brand strategy for 8 to 12 weeks but wants a 4-week read. Don’t push that test. Or say they’d never put a 100% budget increase on a channel. Then don’t test one. A result nobody will act on is just an expensive report.

They’ve had their hands on the keyboard. Look for real DTC marketing experience. Also look for someone who reads marketing and measurement research and listens to marketing podcasts.

They know the industry’s case studies. Published work is worth knowing. Tecovas’s CMO has described tracking customers over an eight-week window to see their full value (“How Tecovas Is Aligning Performance Marketing With Long-Term Brand Storytelling,” The CMO Wire, May 25, 2026, accessed 10/23/2026). Les Binet and Will Davis’s IPA report Go Big or Go Home calls budget-setting the single most important marketing decision (IPA, “‘Small thinking’ is killing advertising warn Les Binet and Will Davis,” 2026, accessed 10/23/2026). I’ve also noticed brands like HexClad and Ridge Wallet regularly share their test results on marketing podcasts. Over and over, I’ve found measurement “experts” who didn’t know this kind of work.

Here’s why it matters. One partner built its product around raising and dropping spend in short bursts (hill and valley) to map response curves. They applied it to YouTube Demand Gen, which is still somewhat upper funnel.

Google reps pushed back. Short spend spikes limit how often any one person sees the ad. YouTube needs time in market to build up. The partner was thinking purely in math, and they were stunned. The product they’d built their business around didn’t fit the strategy. It was one more thing the brand had to teach them.

It’s the same reason last click undervalues upper-funnel channels. The effect builds over time, and short reads miss it.

If you want a partner who works this way, here’s how I work.

Next, what to ask before buying an attribution tool to evaluate the stack beyond the demo, how to evaluate an ecommerce attribution tool to buy a clearer decision instead of another dashboard, the incrementality testing framework to plan a decision-useful DTC test, or what tools measure DTC marketing, by method to pick the method first.