The pattern is remarkably consistent. A company hires an agency. Months one to three feel good. Months four to six feel slow. Around month nine the founder decides it is not working, gives notice, and starts looking for a replacement.
Then the same thing happens with the next agency.
Sometimes the agency really was bad. Often the brief was vague, approvals took three weeks, direction changed twice, and nobody ever wrote down what success would look like. Both situations feel identical from the inside, which is why the second agency fails the same way.
Before you fire anyone, separate the two.
Normal friction versus real failure
Four complaints show up in almost every agency relationship, and none of them is grounds to leave on its own.
"They are slow to respond." Annoying, and usually a resourcing or expectations problem you can fix in one conversation about response times. Agree a standard, then hold it.
"The last campaign did not perform." Campaigns fail. What matters is whether they can tell you why, and whether the next one reflects what they learned. An agency that explains a failure precisely is more valuable than one that never has one.
"I do not love the creative." Taste is not performance. If the work is on brief and converting, your opinion of the headline is the least important data available. If it is off brief, that is a briefing problem worth naming.
"They keep asking me questions." That is what good agencies do. The ones who never ask are the ones producing generic work.
These are all fixable inside the existing relationship. Firing over them puts you back at month one with a new team, paying for onboarding again.
The signals that actually mean it is over
Five things do not improve with another conversation.
- The team changed and nobody told you. The senior people who won the account are gone and juniors are doing the work at the same rate. Ask directly who is on your account and how many hours a week you get. Watch whether you get a straight answer.
- Reporting measures activity, not outcomes. Impressions, posts published, emails sent, hours worked. If you cannot see what happened to pipeline or qualified leads after six months, either the agency is hiding weak results or nobody set up the measurement. Both are serious.
- You have raised the same issue three times. Once is a miss. Twice is a pattern. Three times means the agency has decided this is acceptable to them, and repeating yourself a fourth time changes nothing.
- They cannot explain why something worked. A campaign performs and the answer is that the creative resonated. Without a real explanation there is no way to repeat it, and you are paying for luck.
- You find work has quietly stopped. A channel nobody mentioned in three months. A report that stopped arriving. When you have to discover the gaps yourself, the relationship is already over and only the paperwork remains.
Any two of these together and you are done. The conversation still has to happen, but you should be planning the exit while you have it.
Check whether you are the difficult client
This is the part nobody does, and it is the part that decides whether the next agency works.
Ask four honest questions about your own side.
How long do approvals take you? If your average is 10 days on a blog post, your agency is not slow. You are. An agency built a schedule around your feedback arriving in two days, and everything downstream slipped when it did not.
Was the brief specific? "Generate more leads" is not a brief. A brief names the audience, the offer, the target number, and what you will not do. If your brief was a sentence, the work was always going to be a guess.
Did you change direction mid-quarter? New product priority in month two, different target market in month four. Every change resets the clock and destroys any chance of learning what worked.
Do they have the access and the data they need? Agencies asked to prove pipeline impact without CRM access are being asked to do something impossible. This is the same problem that makes any attribution work collapse, and it is almost always the client's side to fix.
Agencies rarely say any of this to your face. The commercial relationship makes it awkward, so they absorb it and the results quietly suffer. If two or more of those four are true on your side, changing agencies changes nothing.
Run the conversation before you run the process
Give a real chance to fix it, and make it specific enough to act on.
Name three things that have to change. Describe what changed looks like in observable terms, not feelings. "Reporting shows qualified leads and pipeline value by source, monthly" is actionable. "Better communication" is not. Then state a date when you will review it, 60 or 90 days out.
Send a written summary after the meeting. Two paragraphs. What was agreed, what will be different, when you will look again. This stops the classic pattern where a difficult conversation produces two weeks of increased attention and then a return to exactly where you were.
If the agency responds with a plan and specifics, that is a good sign. If they respond with reassurance and a new senior person on the call, treat that as a sales response to a service problem.
Protect the accounts before you give notice
Do this while relations are still good. The ownership audit is a normal governance task and nobody reads anything into it. Requested a week after a difficult conversation, it announces exactly what you are planning.
Go through the list and confirm two things for each: your company owns the account at the top level, and at least one person at your company has admin rights.
- Google Ads, Google Analytics 4, Search Console, Tag Manager
- Meta Business Manager and LinkedIn Campaign Manager
- Your CRM and your email platform
- Domain registrar, DNS, website hosting and CMS
- Any tracking, call recording, or reporting tools you pay for
The one that catches people is Google Ads. Whoever created the account owns it, and if the agency created it inside their manager account, the history may not transfer cleanly. Ask for a transfer now, not during an exit.
Also ask for the source files. Not exported PDFs and flattened images: the editable design files, the raw campaign data, the content in a usable format. Contracts usually cover payment and notice in detail and say almost nothing about the handover, which is precisely where the friction ends up.
And read your renewal clause before you plan the conversation. Some contracts renew automatically unless notice lands inside a specific window, and missing that date locks you in for another term. That is the one deadline you cannot renegotiate afterwards.
What to do instead of hiring the next agency
Answer one question first: was the failure execution or direction?
If the agency did competent work against a bad plan, another agency will do competent work against the same bad plan. What you need is someone who decides what to do. That is a different kind of hire, usually a fractional CMO or a senior advisor a few days a month, and it costs less than an agency retainer.
If the direction was sound and the execution was weak, another agency or a specialist contractor makes sense. Hire narrower this time. One channel, one accountable person, one number.
Companies that cycle through three agencies in four years almost always have a direction problem and keep buying execution. The pattern is easy to spot from outside and very hard to see from inside.
Whichever way you go, pay for a transition period. Thirty to sixty days of overlap where the outgoing team stays available for questions costs less than a new team spending a quarter rebuilding an understanding of your account from nothing.
The takeaway
Fire the agency when the same issue survives three conversations, when reporting counts activity instead of outcomes, or when you have to discover for yourself that work stopped. Those do not get better.
Before you do, spend an hour on your own side. Approval speed, brief quality, direction changes, data access. If the problem lives there, the next agency inherits it, and month nine arrives right on schedule.