Arcane Marketing logoArcaneApply
← All posts

A founder forwards a twelve slide PDF. Impressions up, a line about optimising for quality, and nowhere in it the answer to the question asked three weeks running: which campaigns produced the fourteen calls that got booked? That is usually the moment someone starts reading about when to fire marketing agency relationships, and it is also where the call most often goes wrong. Being handled feels identical to being underserved. Sometimes those are the same thing. Sometimes the buying is competent and the reporting is lazy, which is a cheaper problem with a different fix. Work out which one you have before you give notice, because the handover costs real money and you only want to pay for it once.

Tenure Tells You Nothing About Your Account

The first instinct is to check the calendar. Resist it. ANA and the 4As report that the average client-agency relationship tenure now stands at approximately seven years, more than double the 3.2-year average reported in 2016. The same study splits that number in a way that should change how you read your own engagement: integrated full-service agencies report an average tenure of 87 months (7.3 years), while media-only agencies have a significantly shorter average of 44 months (3.7 years). A paid media engagement running out of road in year two or three sits inside the normal range for its type.

Two caveats before you apply any of that to yourself. Those figures describe agency-of-record tenure among large member brands, so read them as enterprise context rather than a benchmark for a founder-led advertiser. The study also reports no termination rate, so anyone quoting you a percentage of businesses that fire their agency is inventing it.

The process finding is more useful. ANA and the 4As found that clients without mandatory review periods, 60% of respondents, have significantly longer relationships at 8.1 years than those with frequent reviews, as low as 3.8 years. Reading that as proof that reviews cause churn is too tidy. It does argue against installing an annual bake-off as your governance model. Run a diagnostic when something looks wrong, and decide on what the diagnostic finds.

When to Fire Marketing Agency Partners, and When to Fix First

Here is the diagnostic we run before anyone drafts a notice email. Every signal is a question about the account, not about how the last call felt.

You cannot see the account. Read-only platform access to every channel you fund is the floor. If performance only ever reaches you as a slide, you have no way to check anything on it. There is an enterprise version of that discomfort and it is documented: ANA's 2026 report found that 90 percent of marketers say their top concern is whether recommended principal media is truly in their best interest, up from 79 percent in 2024, while only 57 percent of marketers say their companies have guidelines governing principal media. Adoption is moving faster than the governance, with 58% reporting their companies used principal media in the past year, up from 47% previously, where principal media refers to the practice in which agencies purchase media inventory themselves and resell it to clients rather than acting solely as agents. Asking what was bought, from whom, at what price is standard practice at the top of the market.

Nobody owns the number. Strategy questions come back as execution updates. If no named person is accountable for pipeline rather than delivery, you have a vendor and an empty chair. Fractional CMO vs Marketing Agency is the cleaner way to fill that chair than firing over it.

The creative has not moved. Same three hooks for four months, no written test queue, no record of what was learned. The Ad Creative Testing Framework Platforms Document is the standard to hold them to.

You are reported to in numbers that cannot move money. Reach, engagement rate and a cost per result that nobody can tie to a booked call. Marketing Vanity Metrics and Which Numbers Predict Revenue covers the swap.

The diagnosis never changes. Every soft month is the algorithm, the season, or your sales team. A partner that never finds a fault of its own is not looking.

Signals that say fix rather than fire: one weak cycle right after a real change to offer, price or targeting, and a team that volunteers its own misses before you raise them. Month six friction has its own pattern, and Why Marketing Agencies Fail: The Month Six Fight is worth reading before you act on it. Signals that say fire: access refused or slow-walked, figures you cannot reproduce in the platform, and claims about what was bought that come without documentation.

Our own rule for the fix case is one written diagnosis, one agreed change, one reporting cycle to show movement in a number you both named in advance. Call that a house rule. No research here sets a probation window, so choose yours deliberately and write it into the agreement, which is what Marketing Agency Accountability: SLAs and Exit Terms is for.

Check Ownership Before You Check the Contract

Most transition plans open with the notice period. Start one level lower, with whose name is on the asset.

On Google Ads, a client account can only have one owner, and ownership decides what you can do without asking permission. The documentation is reassuring in one direction and limiting in the other: the client account still owns its data and has the ability to remove ownership access by unlinking, while users of the client account do not have permission to transfer ownership but can always unlink a manager with ownership. You can always cut the cord. You cannot hand the account to the next partner yourself, so the incoming team needs its own access path built before the old one is severed.

Meta runs harsher. A partner holding full control of your ad account still cannot pass that asset to another business portfolio, because only the organization that owns an asset can share it onward. If the ad account, dataset, Page or catalog sits inside the agency's portfolio, your transition plan is a request that the other side has to accept. Confirm ownership while the relationship is good. Do not build a plan around claiming assets back later.

The audit itself is one list: ad account per platform, pixel or dataset, Page and Instagram asset, catalog, domain verification, analytics property, tag container, call tracking numbers, CRM, landing page builder, creative source files. Against each, write the entity that owns it and the people holding admin. Anything owned by the agency is either a negotiation or a rebuild, and both take longer than a notice period.

Sequence the Switch So Spend and Measurement Survive

Unlinking comes last. Google's own documentation explains why. When you unlink an account that was using the manager account's remarketing tag or any shared remarketing lists, the lists that rely on the shared tag will no longer populate in that account, and the account will lose access to any of the manager's shared lists. Billing is sharper still: if the Ads account is on monthly invoicing, and the manager account to be unlinked is the paying manager, the account will stop serving after you unlink it. The timing is also not entirely yours, since manager accounts can terminate a relationship with a client account at any time by unlinking it, as long as the client account has at least one user with access to the account, or is linked to a manager account that has administrative ownership of it. Make sure you hold a user on the account before the conversation happens.

Measurement history needs the same handling. In GA4, when data reaches the end of the retention period, it is deleted automatically on a monthly basis, and if you use a date range for non-aggregated reports, such as Explorations, that is longer than your retention period, then data for that additional time is not visible in reports. Pull exploration-level history before notice goes out. Your headline trends are safer, because the data retention setting does not affect standard aggregated reports in your Google Analytics property.

A working order of operations:

  1. Audit ownership and add yourself at the highest permission level each platform allows.
  2. Move billing onto your own payment method, and check whether any account is invoiced through the agency's manager account.
  3. Stand up your own tracking: your tag, your conversion actions, your dataset where ownership permits, and rebuild audience definitions on assets you hold.
  4. Export everything you report on: exploration history, platform reports at your usual granularity, creative files, scripts, landing pages, call recordings, audience logic.
  5. Give notice in writing with a handover list and dates attached.
  6. Bring the new team in with access while the existing structure is still live.
  7. Unlink.

Budget for a re-learning period after structural change. Meta is explicit that delivery is still learning how an ad set may perform after a significant edit, and that ad sets are less stable and usually carry a higher cost per action while that is happening. Nothing published gives you a magnitude for your account, so treat the dip as a figure you set in your own plan rather than one you can look up. Marketing Agency Reporting: Read It Like an Operator helps you tell a learning reset apart from a genuine decline when the first fortnight looks ugly.

Price the Switch Before You Sign the Next One

Run this on your own numbers. Every percentage below is a placeholder.

Assume $40,000 a month of media and a $6,000 management fee, so $46,000 of total cost. Assume a blended $80 cost per lead, which is 500 leads. Assume 18% of those leads reach a held sales call, so 90 calls, and that you close 12% of held calls, so 10.8 deals. Assume a $6,000 average order value. Multiply it out and that is $64,800 of new revenue against $46,000 of cost.

Now price what you are buying. If you believe the next team gets the same $40,000 to a $65 cost per lead at the same call and close rates, that is 615 leads, 110.7 calls, 13.3 deals and $79,800 of revenue, so $15,000 a month better. Price the handover with your own haircut. If you assume the first month runs at 70% of today's output, you give up 30% of $64,800, which is $19,440, and at $15,000 a month of improvement you are square in roughly 1.3 months.

That model does two jobs. It shows whether the gap you are hoping for is big enough to justify a reset, and it forces you to write down the improvement you expect, which becomes the number the incoming team is measured on. If the honest answer is a few points of cost per lead, the constraint probably sits in the offer or the follow-up rather than the media buying, and Speed to Lead Statistics: Where the Numbers Come From is a better place to start than a search.

Do the ownership audit this week whether or not you are unhappy, because it is free while the relationship is good and expensive the moment it is not. The fire decision comes down to access and evidence: a team that gives you the platform and names its own misses deserves a fix cycle even after a bad quarter, while a team that answers the ownership question with reassurance has already shown you what the handover will feel like. Keep the diagnostic in a document, run it every quarter against the account rather than against the vibe of the last call, and the switch becomes a scheduled project instead of an argument. If you want a second read on what your account is telling you, that is the conversation we have on an application call.

Sources

Stuck above the $10k ceiling?

We diagnose your account in 24 hours and tell you straight whether it's a creative, audience, or sales problem.

Talk to us