How do you choose an edtech marketing agency?

Most agency evaluations compare deliverable lists and client logos, both of which are close to worthless. What separates a good fit from an expensive quarter is whether the agency understands how people decide to enrol in something, and whether it will tell you things you would rather not hear.

Start with whether they understand the buying behaviour

Education products have a buying pattern that most markets do not share. Consideration windows run for months. Decisions get triggered by events in someone's life rather than by campaigns. The dominant objection is self-doubt rather than price or features. Enrolment is often gated by a start date. And the proof that matters is outcomes for people who resemble the buyer.

A generalist agency will learn this. The question is whether you want to pay for the quarter in which they learn it. Ask directly what they think the main conversion barrier is for your program. If the answer is about friction, headlines or form length, they are describing a software funnel.

Ask who does the work

The person in the pitch is frequently not the person who will do the work. This is standard at larger agencies and it is not automatically dishonest, but you should know before you sign.

Three questions that settle it:

• Who specifically will be working on this account?
• How many hours a month does the fee include, and at what seniority?
• Who do I talk to when something is wrong?

If the third answer is an account manager rather than someone doing the work, you are buying a layer of translation. On a large budget that layer is useful. On a small one it is a share of your hours going to meetings about your hours.

Ask what they would tell you not to do

This is the most informative question you can ask, and most agencies handle it badly because every incentive points toward agreeing with you.

An agency worth hiring can name something you are currently spending on that they would cut, and a condition under which they would recommend reducing their own scope. If they cannot produce an example of a client they advised to spend less, you are talking to a firm whose recommendations will always be more.

Judge the proposal on its diagnosis, not its plan

A strategy written before anyone has looked at your analytics, your enrolment data or your past cohorts is a template. It will be competent, it will be generic, and it will describe the channels the agency likes running.

What you want to see instead is a diagnosis: here is where we think you are losing people, here is why we think that, here is what we would need to look at to confirm it. A proposal that names its own uncertainty is a better signal than one that names twelve deliverables.

Check the case studies for the part everyone omits

Read case studies for the starting point rather than the result. A large percentage improvement from a very low base is a different achievement from a modest improvement on a mature funnel, and case studies are written to blur that.

Worth asking about any result you are shown:

• What was the number before, in absolute terms rather than percentages?
• Over what period, and across how many cohorts?
• What else changed at the same time, including things you did not control?
• What did not work during that engagement?

The last question is the one that tells you whether you are speaking to someone who ran the work or someone who wrote it up afterwards.

Warning signs

• Guaranteed results. Nobody can guarantee an outcome that depends on your product, your price and your market.
• Twelve month lock-ins. A minimum term protects the agency from the consequences of underperforming, which is precisely when you need to be able to leave.
• A proposal with no questions in it. If they did not need to ask you anything, they did not write it for you.
• Vagueness about hours. An agency that will not tell you how much time you are buying is a reasonable thing to walk away from.
• Pressure to decide quickly. You are the one with a deadline problem. They are not.

Agree what 30, 60 and 90 days look like

Most engagements sour because nobody defined what progress would look like, so both sides are quietly measuring different things by month two.

Set the markers before starting. At 30 days, the honest measure is work shipped and questions answered, not results, because in a market with a months-long consideration window nothing has had time to happen yet. At 60 days you should be able to see whether the first hypothesis held. At 90, or at the end of a cohort cycle, you should have a clear enough picture to continue, change direction, or stop.

Write those down. An agency that resists writing them down is telling you something.

The shortest version

Hire the agency that diagnosed your problem instead of describing its services, told you what it would not do, was specific about who does the work and how many hours you get, and did not need you to sign this week. Those four things correlate with the outcome far better than the client list does.

Common questions