When to Fire Your Agency (and When You're the Problem)

You are six months into a retainer and you have a feeling. Not a spreadsheet. A feeling. The monthly report lands, the deck is prettier than it was in month one, and somewhere between slide four and the "next steps" slide you realize you could not explain to your co-founder what you are paying for. You pay $6,000 a month. That is $36,000 so far. And when someone asked you last week what changed, you said "we're building brand awareness," which is the thing people say when they do not have a number.

The feeling is real. The problem is you have no framework for it, so you keep giving it another month. This is the companion piece to how to hire a marketing agency, from the other end: how to know when it is time to leave, how to leave clean, and how to tell whether the agency is the problem.

The five signals, in the order they show up

Firing an agency on vibes is how you end up hiring a worse one. Here are the signals that mean something, roughly in the order operators notice them.

Reporting that never mentions revenue. Impressions up 40 percent. Engagement rate up. Reach, frequency, sentiment. All of it moving, none of it money. A report that cannot draw a line from spend to dollars is not a report, it is a mood board. The good version names revenue, pipeline, or cost per acquisition on the first screen. If you have to scroll to find a dollar sign, that is signal one.

"Strategy" calls that restate last month. You get on the call. They walk you through what they did. You already knew, because you were there. Nobody proposes a bet that could be wrong. A real strategy call has a hypothesis in it, and a hypothesis can fail. Restating the past is not strategy, it is attendance.

Deliverables that require them forever. Every asset lives in their tools, their accounts, their heads. Ask for the thing and it comes back as a PDF or a screenshot, never a source file. This is not an accident. Work that only functions while they are attached is the retainer trap working exactly as designed.

Cost per acquisition drifting up while the deck gets prettier. This is the tell that costs the most and hides the longest: production quality climbs while the economics sink. The slides look more expensive because they are the deliverable now. The results were supposed to be.

And the tell of tells: you cannot answer "what do we own that we did not own six months ago."

If you cannot name one asset you now own that you did not own six months ago, you are not building an engine. You are renting applause.

Accounts, pixels, tracked audiences, a documented funnel, source files, an email list that grew. If none of that came into your possession, the money bought activity, not equity. That is the difference between a retainer that compounds and one that just recurs. We wrote the positive version of this in own your acquisition engine: the whole point is that the machine keeps running the day the vendor leaves.

Before you tell yourself it's them: is the offer broken?

Here is the part most "fire your agency" advice skips, and skipping it is how good operators fire good agencies and then repeat the exact same six months with a new logo.

Sometimes the agency is fine and the offer is broken. No amount of media buying fixes a product nobody wants at a price that does not clear. If your cost per acquisition is $180 and your customer is worth $150, that is not a targeting problem, that is a math problem, and no creative refresh closes a negative gap. The honest test: pull three things before you assign blame. Your close rate on the leads they send. Your average order value over the period. Your gross margin per customer. If leads are arriving and your own sales process is dropping them, or the unit economics never worked, the agency could be excellent and you would still feel exactly this bad.

The clean way to tell them apart: look at the top of the funnel and the middle separately. If click-through, cost per click, and lead volume are healthy but nothing converts to revenue, the machine upstream is working and the problem is your offer, your sales follow-up, or your speed to lead. If the top of the funnel itself is degrading, costs climbing, volume falling, creative stale, that one is theirs. It is the difference between a decision and a mood.

If you want a second read on which side of that line you fall on, book a call and bring the three numbers. Fifteen minutes usually settles it.

The clean exit, step by step

Once you have decided, the order of operations matters more than the decision. Most operators give notice first and ask for their assets second, which is exactly backward and hands the leverage to the party you are leaving.

Inventory what is actually yours. Make the list before any conversation. Ad accounts (the account, not "access to" the account), the pixel and its event history, custom and lookalike audiences, the domain and DNS, analytics properties, the CRM and its data, email lists and the sending platform, and every source file: design files, ad creative, landing page code, automation blueprints. Write down which of these are in your name today and which are in theirs. The gap is your exposure.

Request handoff in writing, before you give notice. This is the single move that separates a clean exit from a hostage negotiation. Email, not a call: "As we plan for the next phase, please confirm in writing that the following accounts and files are owned by us and transfer administrative ownership to these addresses." You are not announcing a breakup. You are doing normal governance. A healthy agency does this same day. An agency that stalls, gets vague, or discovers a "policy" just told you everything about why you are leaving.

Overlap on purpose. Do not run a hard cutover. Keep the retainer live for two to four weeks after the handoff request while you confirm you can actually log in to everything, that campaigns still run under your ownership, and that nothing was tied to a login that walks out the door. Pay for the overlap. It is the cheapest insurance you will buy all year.

Then give notice, with the handoff already done or in motion. Reference the written confirmation. Set a data export date. Ask for a decommission list: what they will turn off, and when, so nothing keeps billing your card after they are gone.

We ran this exact sequence with an operator who came to us mid-retreat from a previous shop. Their prior agency held the ad account and the pixel; two years of conversion history lived in an account the client did not control. We had them send the written handoff request before a word about leaving, got ownership transferred inside a week, and overlapped for three. The pixel history, the audiences, the account, all of it moved. Nothing reset to zero. Compare that to the operators who give notice first and rebuild their tracking from scratch. This is the same ownership principle that runs through escape SaaS hell: if leaving means starting over, you never owned it.

The week after

The relationship ends on a Friday and Monday still exists. Here is what the first week looks like when you have done the exit right.

Log in to every account yourself, from your own credentials, and confirm you are the owner and not a guest. Rotate any shared passwords and revoke lingering user access. Export everything twice: the ad data, the CRM, the email list, the analytics history, and store a copy somewhere that is not a single vendor. Check that billing now runs on your cards, not theirs, so nothing silently lapses or silently renews. Then, and only then, do the strategic work: look at the assets you now hold and decide what the next engine looks like, whether in-house, a new partner, or a build like the ones in our work.

The quiet win of a clean exit is that you are not starting over. You are starting from six months of accumulated ownership, which is what those six months were supposed to buy.

Why an agency is telling you how to fire agencies

Because the incentive alignment is the point. We would rather you leave clean and come back a believer than stay confused and resentful and eventually leave loud. An agency whose work only functions while it is attached has to keep you a little dependent to survive. An agency that hands you a machine you own has to be good enough that you keep choosing it. Those are different businesses. We described which one we are trying to run in the un-agency manifesto, and this post is that manifesto applied to the hardest moment: deciding whether to stay.

If you are six months in with a feeling and no framework, run the five signals, then run the offer test, then run the exit in order. If you want a neutral read before you do anything, book a call. We will tell you if it is them. We will also tell you if it is you.