How to evaluate a B2B copywriting agency before you sign

Most B2B copywriting agencies have the same website. A hero section with the word 'results.' Three client logos you half-recognise. A case study that says revenue went up without telling you by how much or why. If you have spent time evaluating agencies recently, you already know this. The pitch calls blur together.

The problem is not that agencies are dishonest. It is that the buying process rewards polish over proof. A founder who spends 30 minutes on a Zoom call has almost no signal about whether the writing will actually be any good. So they default to proxies: the deck looked clean, the account manager seemed sharp, the price felt reasonable. Then the first draft lands and it reads like a LinkedIn post written by a committee.

This post is about getting real signal before you sign anything.

Ask to see work in your category, not their portfolio highlights

Every agency shows you their best work. That is not useful. What you want is three or four pieces from a client who sells something structurally similar to what you sell. Not the same industry necessarily, but the same sales dynamic. Long sales cycle. Technical buyer. High-consideration purchase. If you sell a $40,000-a-year data platform, a portfolio full of e-commerce product descriptions tells you almost nothing.

When you ask for relevant samples, pay attention to two things. First, does the copy name a specific problem, or does it gesture at a vague category of pain? Good B2B copy says 'your data team is rebuilding the same dashboard every quarter because the source definitions keep changing.' Weak copy says 'streamline your data operations for better business outcomes.' Second, is there a clear point of view in the writing, or does it hedge every claim into meaninglessness? An agency that cannot take a position in a blog post will not take one in your sales page either.

If they push back on showing you category-relevant work because of NDAs, that is fine. Ask them to write a short spec piece instead. Which brings us to the next test.

The paid test project is the only real interview

A test project is not a free trial. Pay for it. $200 to $500 for a 600-word piece is reasonable and it changes the dynamic entirely. When money is involved, the agency assigns a real writer rather than whoever is available. You see how they handle a brief. You see whether they ask clarifying questions or just start writing. You see the quality of the first draft, which is the single most predictive signal you have.

Here is what to look for in that draft. Does the opening sentence earn your attention, or does it start with a definition? ('Account-based marketing, or ABM, is a strategy in which...' is a failing grade.) Does the writer understand the difference between a feature and a mechanism? A feature is what the product does. A mechanism is the specific reason it works. The best B2B copy names the mechanism. Does the piece have a point of view, or is it a balanced overview that could have been written by anyone?

Also notice what happens when you give feedback. An agency that argues with every note, or conversely accepts every note without comment, is a problem. You want a writer who pushes back on feedback that would make the piece worse, and incorporates feedback that would make it better. That requires judgment. Judgment is what you are actually buying.

How they handle a brief tells you everything about the relationship

Before you even get to a test piece, send them a brief. Not a perfect brief. A realistic one. The kind you will actually send when you are busy and the deadline is tomorrow. A paragraph of context, a rough sense of the audience, a link to a competitor piece you admire. Then see what they do with it.

A good agency asks three to five specific questions. Not 'can you tell us more about your target audience' but 'you mentioned enterprise buyers. Are we writing for the economic buyer who signs the contract, or the technical evaluator who builds the shortlist? They care about different things.' That kind of question tells you the writer has already started thinking about the problem.

A weak agency either asks no questions and just produces something generic, or sends you a 20-question intake form that makes you do all the thinking for them. Neither is what you want. You want a collaborator who does the intellectual work of figuring out what the piece needs to do, not a vendor who executes instructions.

For more on what a good brief looks like from your side, see how to brief a B2B copywriter.

Pricing structure reveals how they think about quality

Most agencies price by word count or by deliverable. Both models create the wrong incentives. Per-word pricing rewards length. Per-deliverable pricing rewards speed. Neither rewards the thing you actually want, which is copy that makes a reader do something.

Ask how they price revisions. If unlimited revisions are included, that sounds generous until you realise it means the first draft is treated as a starting point rather than a finished product. Agencies that take first drafts seriously tend to charge for rounds beyond two, because they have skin in the game on quality from the start.

Also ask what happens when a piece is not working. Not when you want minor edits, but when the whole angle is wrong. A good agency will have a clear answer. A bad one will get vague. The answer you want is something like: 'we rewrite it. That is on us if we missed the brief.' The answer you do not want is a reference to your approval of the outline.

For a clearer picture of what different price points actually buy you, affordable B2B copywriting breaks down what you get at $89, $300, and $1,500 a month.

The references question nobody asks

Everyone asks for references. Almost nobody asks the right question. 'Were you happy with the work?' produces a useless answer. The client said yes to a reference call, so of course they say yes.

Ask this instead: 'What was the hardest piece of feedback you gave them, and how did they respond?' Or: 'Was there a piece that did not work? What happened?' These questions require a specific memory. Vague praise is easy. Specific recollection requires that the relationship was real.

If the reference pauses, thinks, and then tells you a real story with some texture to it, that is a good sign. If they give you a polished two-sentence endorsement that sounds like a testimonial they already wrote, it probably is.

The goal of all of this is not to make the evaluation process adversarial. It is to compress the timeline to truth. A bad agency match costs you three to six months of mediocre content, a transition period, and the opportunity cost of not having good copy during that time. The tests above take maybe four hours total. That is a reasonable investment before committing to a retainer.

The one thing that predicts a good agency relationship more than anything else: they care about whether the writing works, not just whether you approved it.