Should a Startup Hire a Full-Time Marketer, an Agency, or Neither?

An agency can help your startup when the story is already settled and the work that remains is distribution. It cannot help when the story is still being written, because an agency is built to amplify a message, not to settle one. The test is cheap, and it takes one call.

Ask them what they think your buyer compares you against. Not who your competitors are. What that buyer and their team would do next month if you shut down tonight. There are two kinds of answer. One guesses, confidently, and names two vendors from your own deck. The other says it does not know yet, names how it would find out, and asks who your best customers are. Only the second is holding the right end of the problem.

This matters more in committee sales than anywhere else. Forrester reports that 86 percent of B2B purchases stall mid process, and the main driver is that the buyer cannot picture the state after the purchase in enough detail to justify it internally. An agency optimising cost per click has no instrument for that failure: more of the right people reach a page that still does not tell them what they would otherwise do, which is a busier version of the same stall.

There is a second trap, and it looks like good practice. Persona by persona content is the standard agency deliverable, and what I keep seeing is that without one shared argument underneath it, it hands each function its own reason to hesitate. The channel and the position are not two projects; I argued the whole case for that in how to connect positioning to pipeline.

You choose between marketing agencies by what they ask you before they show you anything. The strongest proposal in the pile is usually the one that refuses to name channels until it can name your buyer's alternative.

Three questions, asked before you look at a deck, separate the field faster than any case study:

“What will our buyer compare us against, and how will you find out?”

“Which function in the committee is most likely to veto us, and what will that person need to see?”

“What would you tell us not to do in the first ninety days?”

The first question tests whether they know that the comparison, not the product, sets the terms of the deal. Your buyer's real alternative is the yardstick, and until it is named, every claim you make is measured against something you never saw. I walked through how to find it, with the questions I use, in what your best customers would do if you did not exist. The second tests whether they have ever sold into a room of five to sixteen people across as many as four functions, each with its own definition of value and its own quiet veto. The third tests whether they will protect your budget from themselves. A firm with nothing on that list has one plan and is selling it to everyone.

A proposal that opens with channels, volumes and a monthly fee, before any of those questions has an answer, is a proposal to amplify whatever is in your head. That can be the right purchase, once you know what is in there.

If you hire, hire a generalist, because at seed to Series A you do not yet know which channel will carry you and a specialist commits you to one before the evidence arrives. The real question underneath is seniority, not breadth: a generalist who can settle the story is a different hire, at a different cost, from a generalist who can run one that already exists.

That combination is rarer than the job postings suggest. Xavier Silva, CEO of HEMAV Technology, where I was CMO for four years, described it from the other side of the table: “I have met plenty of very strategic profiles, but they then need a team of seven to eleven people to execute and make those ideas real. Your profile has that, but you are also very hands on, and with few resources you can plan long, medium and short term, and deliver. So you achieve both.”

The Loom Sprint is the go to market I install, and that work is not new: twenty years of it, four of them as his CMO, doing everything the Sprint includes and more. What is new is only the format, ten weeks at a fixed price, because I now work independently. A chief executive who watched that capability hold for four years is firmer ground than any review of a ten week project, and I am on board for the ten weeks, so what he describes is what gets applied. The real guarantee is those twenty years; cutting the thread only guarantees the offer. The full interview is on the site, and it names what you are shopping for: strategy and hands, in one person.

The benchmarks are quieter on this than you would expect. The 16th Annual KBCM and Sapphire Ventures SaaS Survey, on 2024 numbers, gives median headcount under 10 million dollars of ARR as four account executives, one SDR, one sales engineer and five customer success managers. Marketing headcount does not appear at all. ICONIQ Growth's October 2024 report on scaling from 1 million to 20 million dollars of ARR, built on 107 B2B SaaS companies, warns about over hiring and recommends sequencing hires so each one makes the next stretch possible. Both samples are mostly North American: read them as direction, not as your market.

The direction is plain: the hire is not early, and what it runs has to exist first. If the budget question is what is keeping you up, I took it apart in how much it costs to hire marketing for a startup.

A junior marketer can run marketing at this stage, as long as there is a story for them to run. What a junior cannot do is write it, and between seed and Series A the story is usually the only thing missing. Asking a first marketing hire to invent your positioning is asking the most junior person in the company to make its most consequential strategic decision, with the least context, while publishing on LinkedIn twice a week.

Most founders I talk to are not weighing an agency against a consultant. They are weighing it against hiring a growth or marketing person in house at local salary, which often means one junior profile, or a Head of Growth with nobody senior above them. It fails specifically: the new hire spends two quarters doing archaeology on your own company, trying to extract the sentence you have been improvising, and the output is a document you could have handed them on day one.

What that person needs on day one is short and uncomfortably specific. The alternative your buyer would otherwise choose, named. The value you deliver stated in the buyer's own currency rather than in your product's features. And one argument each function in the committee can own differently, so your champion can defend it in a meeting you were never invited to.

I should be straight: there is no reliable data on the failure rate of a first marketing hire. The closest figure is about quota roles: SaaStr's finding that seven out of ten first VP of Sales hires do not work out, and that each miss sets you back a year. I will not stretch that onto a marketing hire; it is not the same job, but it does describe the shape of the risk. The expensive part of a bad first hire was never the salary. It was the year.

A fractional CMO is the option people add once they get past the first three, and I wrote about what a fractional CMO for a technical founder actually is. It is a senior seat rented by the month: a good answer when there is a machine to run, a slow one when there is not.

So: neither, until the sentence exists. Write down what your buyer would do next month if you disappeared tonight, in their words, not yours. Then write what your homepage asks to be compared against. If those two things do not match, no hire and no retainer will fix it, because every one of them starts from the page.

Which raises the option the title leaves out: do it yourself. While you are in every call, it works. What I keep seeing is where it stalls, and it is not effort. Your own words are the vendor's vocabulary, so when you ask your best customers what they would do without you, they answer politely in that vocabulary and you learn nothing you did not already believe. And the sentence stays improvised, built fresh in each meeting and lost at the end of it, so after a year of good conversations there is still nothing another person could pick up and run. It fails quietly, paid in time rather than money, which is why it goes unnoticed for so long.

Once the sentence is written and it holds in a room you are not in, a generalist can run it and an agency can carry it into channels. The choice becomes a budget decision.

This is the work I do, as a go to market strategy consultant for B2B founders selling to committees of risk, IT and compliance, from seed to Series A. The Loom: positioning and go to market, built in ten weeks, priced as a fixed project rather than as a monthly retainer or a seat on your payroll.

A project with an end date means something specific here. Week one is three hours of the right questions, your read of the field against mine. Week two is the Unravel map: the places where your go to market is bleeding, named and ranked. By week four the positioning document is in your hands, and by week ten you have a go to market your team runs without me.

The exercise above tells you whether that sentence exists; the call is for when it does not, or when it exists only in your head. At week four you hold the positioning document, and if it has not shown you the system, you say so, you cut the thread, the second payment never happens and you keep everything delivered. If you are weighing a hire against an agency this quarter and the sentence is still not written down, book a call.

Sources: Forrester, via engsales.substack.com. 16th Annual KBCM and Sapphire Ventures SaaS Survey. ICONIQ Growth, Scaling SaaS from $1M to $20M ARR, October 2024. SaaStr. Xavier Silva interview, recorded July 2026, published at monkeyweaver.com/testimonial.