Data driven is the least discriminating phrase in agency marketing, because no firm describes itself as intuition driven. The claim is worth something only when it survives specific questions, and the useful ones are not about tools. Any agency can connect a dashboard. The tests that matter are whether they can tell you what their data cannot see, whether they have ever recommended stopping something profitable-looking on the evidence, and whether their measurement survives contact with your finance team's numbers. This page gives you those tests in an order you can run during a single call.
Ask what their data cannot see
This is the fastest discriminator in the category and almost nobody asks it. Every measurement setup has blind spots: consent banners that suppress a share of analytics collection, browsers that block third party identifiers, offline conversions that never return to the platform, phone calls that were never tracked, long sales cycles that outlast attribution windows, and modelled conversions that the platform estimates rather than observes. An agency that genuinely works with data will answer this fluently and tell you roughly how large each gap is in accounts like yours. An agency using the phrase as decoration will insist they track everything, which is the wrong answer in every account that has ever existed. Follow it with a second question: when the ad platform and your own analytics report different conversion counts, which do you report to the client and why. The honest answer shows both and explains the gap. The revealing answer is whichever number is larger.
Ask for a decision, not a dashboard
Request a specific example where analysis changed what the agency did, including what it changed from, what it changed to, and what happened next. Good firms have several and tell them with the awkward parts intact, usually including something that did not work. Then ask for the harder version: a time the data said to stop something the client liked, and how that conversation went. Agencies that have never had that conversation are reporting, not analysing. Also ask who does the analysis. In many firms the dashboard is built by a junior and interpreted by an account manager reading the same charts you can read, which is a service you are paying a premium for and receiving nothing extra from. Ask to meet the person who will actually look at your numbers each month, and ask what they will look at first.
Test whether their measurement survives your finance team
The end state of a data driven engagement is that marketing numbers and finance numbers can be reconciled, or at least that the difference is understood. Ask each candidate how they would connect reported conversions to booked revenue in your business, given how you actually record sales. If you sell through a CRM with a long pipeline, the answer must involve importing outcomes back into the marketing systems, not just counting form fills. If you sell offline, the answer must involve call tracking and some accepted approximation. Ask what they do about incrementality: how they would tell whether the sales credited to a campaign would have happened anyway. Nobody solves this perfectly, but the willingness to discuss holdouts, geographic tests and simple before and after comparisons separates practitioners from dashboard operators. This is also where digital marketing and SEO services stop being a channel purchase and become a measurement discipline, since organic contribution is the hardest thing in the mix to attribute honestly.
Ownership, tooling and what happens when you leave
Confirm three ownership facts in writing before signing. First, your business owns the analytics property, the tag manager container, the advertising accounts and any data warehouse, with the agency granted access rather than the reverse. Second, dashboards built during the engagement are handed over in a form you can keep running, or you accept openly that they are a service that ends. Third, any proprietary tool the agency uses is either exportable or explicitly a rental, so you are not surprised later. Then ask what the fee assumes in hours, what is subcontracted, and what the disclosed minimum engagement is. Our index records published pricing and disclosed minimums precisely because firms that publish them can be compared on the same axis, and in this category the agencies willing to be specific about their own commercial terms tend to be the ones willing to be specific about your results.
Questions people ask about data driven marketing agencies
Does a data driven agency need in house analysts?
Not necessarily, but someone must be accountable for interpretation rather than reporting. Small firms often have one strong generalist who does this well. What does not work is a setup where nobody can explain why a number moved, which you can detect by asking about any surprising figure in their sample report and seeing whether the explanation is causal or descriptive.
How much of the budget should measurement consume?
Enough to be trusted and no more. Setup is a real project at the start; ongoing measurement should be a modest share of the retainer. Be wary when a large part of the fee funds dashboards while little funds the work being measured, and equally wary when measurement is treated as free, because then it was not done.
What if our data is a mess right now?
Say so during selection rather than after. A competent firm will scope a short remediation phase, tell you what it costs and what it will not fix, and set expectations that early reporting is directional. A firm that promises to work around it will produce confident reports built on broken tracking, which is worse than no reports.
Which single question exposes the most?
Ask what they would recommend you stop doing in your current marketing, before they have your data. A firm with a point of view will name something and explain what evidence would confirm it. A firm selling a bundle will say it needs more information, which is true but is also the answer that commits to nothing.