Is My Marketing Agency Doing A Good Job

Updated September 2026 · Written and maintained by the Progression Agency strategy team

Evaluating an agency you already work with is a different problem from choosing one. What is knowable changes by stage, results depend partly on the channel rather than the agency, and the most expensive failure mode is not collapse but slow drift. This page sets out what is fair to judge at each stage, what a report should contain, how to run a review that produces action, and how to decide between fixing the relationship and replacing it.

Judge the relationship by stage, not by a single number

The most common mistake in evaluating an agency is applying a twelve-month standard in month three, or a month-three standard in month twelve. Different things are knowable at different points, and holding an engagement to the wrong one produces either premature termination or indefinite patience.

Early on, almost nothing about results is knowable. What is knowable is whether the work is being produced at the promised volume, whether the agency understood your business, and whether questions get answered. Those are process signals, and they predict outcomes better than any early metric.

Later, process stops being the point and outcomes become fair game. An agency that is still reporting activity in month twelve — posts published, hours spent, keywords tracked — without connecting it to anything commercial is telling you what it did, not whether it worked.

Reporting that proposes nothing is not reporting

A monthly report exists to support a decision. If reading it changes nothing about what happens next month, it is a status update wearing a report’s clothes.

The test is simple: does the document say what the agency intends to do differently, and why? A report that presents rising numbers with no recommendation is as unhelpful as one presenting falling numbers with no explanation. Both leave you unable to act.

The second test is whether anything in it ever went wrong. Real programs produce failures — concepts that did not land, channels that underperformed, tests that returned nothing. A report containing only wins is not being used to make decisions, and may not be being read internally either.

Look for a recommendation, not a summary

The section that matters says what changes next month and why. Everything above it is evidence for that recommendation.

Check whether the metrics ever change

An agency reporting the same six numbers for a year is not learning. As a program matures, what is worth measuring should shift.

Ask what they stopped doing

Programs accumulate activity. An agency that has never discontinued anything is adding without pruning, which quietly consumes the budget.

Watch for metric substitution

When a headline number turns down, weaker agencies begin reporting a different number. Note which metrics disappear from the deck.

Separate the agency’s performance from the channel’s

Some channels underperform for reasons no agency controls, and some agencies hide behind that fact. Distinguishing the two is the core difficulty in this assessment.

The fair question is not whether results are good but whether the decisions were sound given what was knowable at the time. An agency that recommended a channel, tested it properly, measured it honestly and recommended stopping when it did not work has performed well even though the channel failed.

The unfair version is judging a considered-purchase B2B program on ninety-day revenue, or a brand campaign on direct response metrics. If the objective was never achievable in the timeframe, the agency may still be at fault — for agreeing to it — but the work itself may be sound.

The three questions that reveal most, fastest

If you only have one meeting, these three produce more signal than any report review.

First: what did you learn about our customers this quarter that we did not already know? Twelve months of work should produce insight, not only output. An agency with no answer has been executing rather than examining.

Second: what would you stop doing if it were your money? Programs accumulate activity, and an agency unable to name its own weakest component is not evaluating the work.

Third: what do you need from us that you are not getting? A meaningful share of underperformance originates on the client side, and an agency that has been raising this quietly will say so plainly when asked directly.

Signs the relationship is drifting rather than failing

Outright failure is easy to identify. The more common and more expensive pattern is slow drift, where nothing is wrong enough to act on and the value quietly declines.

The signals are relational rather than numerical. The senior people who pitched are no longer on calls. Response times lengthen. Recommendations become safer and less specific. You find yourself chasing deliverables that used to arrive. Meetings become status recitals rather than decisions. Nobody has proposed anything new in two quarters.

Drift is worth naming early because it is usually recoverable. Most agencies respond to a direct conversation about it, and the alternative — waiting until it is bad enough to justify leaving — means paying for a declining service for several more months.

Run a structured review rather than forming an impression

Impressions accumulate and then surface all at once, usually as a decision to leave. A short written review twice a year converts that into something both sides can act on.

Cover four things: what was delivered against what was agreed, what the agency learned that you did not know before, what they recommend changing, and what they need from you that they are not getting. That last item matters — a meaningful share of underperforming engagements are starved of approvals, access or information on the client side.

Write it down and share it. An agency given a specific written assessment can respond to it. An agency given a vague sense of dissatisfaction will guess, and usually guess wrong.

Ask what they would do with double the budget

The answer reveals whether they have a plan or are executing a template. Vague answers about more of the same are informative.

Ask what they would cut first

Equally revealing, and rarely asked. An agency that cannot name the weakest part of its own program is not evaluating it.

Ask what they need from you

Approvals, access, subject-matter time, faster feedback. Being starved of these is a common and fixable cause of underperformance.

Ask what they have learned about your customers

Twelve months of work should produce insight, not just output. If nothing has been learned, nothing was being examined.

When the honest answer is that you are the problem

This is uncomfortable and frequently true. Agencies underperform for client-side reasons more often than either party acknowledges, and switching agencies does not fix any of them.

The recurring causes are slow approvals that make timely work impossible, no access to the people who understand the product, changing objectives every quarter, a budget too small for the objective agreed, and decisions overturned by someone who was not in the room. An agency that raised these and was ignored has done its job.

The test is whether the same problems followed you from a previous agency. If they did, the next agency will encounter them too, and the money spent switching will buy a repeat of the same experience.

Deciding whether to fix it or leave

Once the assessment is done, the decision usually comes down to two questions rather than a balance of many.

First: is the problem capability or effort? An agency that cannot do the work will not learn to quickly, and no amount of pressure changes it. An agency that can but has drifted usually responds to a direct conversation, and replacing it costs three to six months of ramp-up you would not otherwise spend.

Second: has the relationship survived a difficult conversation before? Partnerships that can absorb direct criticism tend to recover. Ones where raising a problem produces defensiveness rather than a plan rarely improve, because the mechanism for improving is the thing that is broken.

What good actually looks like

It is worth stating the positive case, because a list of warning signs makes every relationship look suspect.

A good agency tells you things you did not want to hear, early. It recommends against work that would earn it money. It reports failures alongside wins without being asked. It knows your commercial numbers, not just its channel metrics. Its recommendations change as evidence accumulates. And the people who pitched are still recognizably involved.

None of that guarantees results, because results depend on the market as well as the work. But an agency doing all of it is giving you the best available chance, and one doing none of it is not, regardless of what this quarter’s numbers happen to show.

Reference videos

Measurement and reporting fundamentals relevant to the assessment above.

Frequently asked questions