In a 2025 survey of roughly 100 agency and brand respondents, dissatisfaction with delivery and dissatisfaction with value were tied as the leading reasons clients ended relationships, each cited by 61% of respondents, according to the 2025 survey of agency and brand respondents. I'm Chase McGowan, a self taught, independent Google Ads operator. I keep a deliberately small roster, I do the work myself, and I don't fire agencies for sport. When to fire your marketing agency is usually a measurement question first, a people question second, and a contract question third.
In 2024, the share of clients planning to switch agencies fell 15 percentage points from 55% in 2023 to 40%, based on a 2024 survey of 400 marketers. Falling intent to switch doesn't mean performance improved. It means you need a clearer test than frustration.
My working rule is simple. One bad month isn't grounds for termination. Persistent delivery failure across two or more review periods, measured after tracking is verified, is. Fees going up, a weak month, or a report you don't like are reasons to investigate. They're not proof.
Start with economics, not platform vanity. For a medical practice, that's cost per booked procedure. For ecommerce, it's revenue or gross profit against ad spend. For lead generation, it's qualified cost per booked outcome, not cost per click or raw form fills. If your agency optimizes to platform reported conversions while your CRM shows flat booked appointments, you're not arguing about taste. You're arguing about what counts.
A primary conversion is the conversion action Google Ads uses for bidding and reporting. If nobody can tell you which actions are Primary, how calls are counted, or what minimum call duration applies, every optimization decision after that is suspect. The same goes for ownership. If the agency is the sole administrator and you can't see billing or conversion settings, that's structural. You don't have a vendor. You have a hostage situation with invoices.
Junior turnover matters for the same reason. When your primary contact changes twice in six months, account history walks out with each person. Search term exclusions get repeated, tests get rerun, and nobody remembers why a bid strategy changed. A missing change log makes that worse because there's no falsifiable record to learn from.
| Signal | Threshold to act | Classification |
|---|---|---|
| Primary conversion not verified | Unverified after 14 days | Fix tracking first |
| Qualified cost per booked outcome | Up 25% vs 8 to 12 week baseline for two review periods, after tracking is verified | Escalate to 30 day corrective plan |
| Platform conversions vs CRM qualified leads | Platform up, CRM flat or down over a full sales cycle | Measurement or optimization red flag |
| Administrative access | Agency is sole admin; you cannot see billing or conversion settings | Structural red flag |
| Change log | Budget, exclusion, bid or test changes not logged | Red flag |
| Primary contact | Changed twice or more in six months | Junior turnover signal |
| Monthly report | No conversion action, attribution model or date range labeled | Unusable for decisions |
| Recommendations | Rejected with no written test or reason | Learning is not falsifiable |
Use these thresholds as working rules, not universal benchmarks. A high ticket practice with a 60 day sales cycle needs a longer window than a clinic booking exams this week. What doesn't change is the order. Verify measurement, compare against a baseline, then judge delivery.
If you're rebuilding scope from scratch, it helps to build your own paid search package around the outcomes you pay for, calls, booked procedures, qualified leads, revenue, instead of buying a bundle of activity. For the account level work, Come Together Media's account audit guide lays out the checks in the order I'd run them.
Do this week: Export 90 days of spend, primary conversions and CRM outcomes. Put platform numbers and CRM numbers side by side by week. If they tell different stories, you don't have a firing decision yet. You have a measurement decision.
That export takes an hour. It'll tell you whether you're looking at weak media buying, broken tracking, or an agency optimizing to low value events and calling it growth.
A drop in results isn't evidence of agency failure until you've ruled out what changed outside the ad account. I check client side causes first because they're common, they're fast to verify, and firing an agency over a broken form helps nobody.
Google Ads conversion totals can move when the attribution model or a conversion action changes, even if spend and customer behavior don't. Attribution is the rule that assigns credit for a conversion across clicks. Change the rule and reported CPA and ROAS, return on ad spend, can improve or collapse overnight without one extra customer. Check settings history before you blame strategy.
Market pressure is real too. In an immigration medical exam account I run at about $13K per month, over two years cost per booked exam rose only 10% while market CPCs rose 54%, platform reported, with conversion rate up from 11.5% to 15.9%. Judged on CPC alone, that account looks worse. Judged on booked exams, the measurement held and the economics held. That's the difference between auction inflation and operator failure.
The photo below captures the unglamorous part of this work, reconciling reports against what your business recorded.
| Symptom | Check on your side first | Where to look |
|---|---|---|
| Leads drop suddenly | Form, phone number or landing page broke; mobile page speed changed | Test form submissions and call numbers on mobile |
| Cost per click rises across the auction | Market competition or seasonal demand | Auction insights report in Google Ads |
| Reported conversions fall after a settings change | Attribution model switched or conversion action edited | Conversion settings and attribution history |
| Calls drop | Minimum call duration changed, number swapped, after hours routing | Call settings and call detail review |
| Sales fall while ads look steady | Offer, pricing or capacity changed | CRM close dates against click dates |
| Revenue lags ad activity | Sales cycle longer than the reporting window | Conversion lag data in Google Ads |
Other client side causes show up constantly. A reception team that's understaffed misses calls that ads paid for. A price increase lowers conversion rate. A surgeon takes two weeks off and booked procedures fall. None of that is media buying.
If the drop lines up with a change on your side, log it, annotate the account, and don't fire on that data. Fix the form, restore the number, staff the phones, then measure again over a clean period.
Documentation turns an argument into a decision. Do this before a tense call, not during it, and keep every export dated.
The two step summary below covers where I'd start.
Conversion tracking determines which campaigns get credit, so the mechanics matter. Google's data driven attribution documentation recommends at least 200 conversions and 2,000 ad interactions in a 30 day period for the model to analyze data more effectively. Below that volume, data driven attribution gets noisier, so judge performance on a longer window instead of weekly swings. Enhanced conversions, hashed first party data sent back to Google to improve measurement, and consent mode, which adjusts tag behavior based on user consent, both affect what gets counted. Your agency should be able to explain your setup in plain language.
This short walkthrough shows what clean checks look like in practice.
I use the PPC Vault audit checklists to keep these checks consistent across accounts. There's also a free PPC Vault preview, a free preview of vault audit checklists, if you want the structure before you confront anyone.
If an agency reports a lower cost per conversion but can't name the conversion actions, counting rules or attribution settings behind it, that number isn't evidence of improvement.
Firing is one option among several, and the right one depends on what your documentation shows. If tracking is broken, replacing the agency before fixing measurement just hands the same broken inputs to a new vendor.
A corrective action plan works when tracking is sound and strategy is the gap. It needs written targets, named owners and a 30 day deadline, or it's just another month of retainer. A partial scope change works when one discipline is strong and another is weak. If creative is performing and paid search isn't, move paid search and keep creative. If video is the weak spot, AdCrafty's video advertising agency guide is a useful reference for what a specialist scope should include before you renegotiate.
| Option | Best when | Typical cost | Main risk |
|---|---|---|---|
| Corrective action plan with current agency | Tracking is sound and strategy is the gap | Current retainer for 30 days | Repeats the same behavior without written targets |
| Partial scope change | Creative strong, paid search weak, or the reverse | Renegotiated retainer | Coordination gaps between vendors |
| Fractional senior operator | Need senior oversight without a full time hire | $3,500 per month minimum | Requires clear handover and agreed KPIs |
| Paid 30 day diagnostic and rebuild | Tracking unknown; want proof before a longer commitment | $7,500 | Short window; ongoing decision due at day 30 |
| Independent second opinion | Unsure whether the agency is the problem | $350 | Diagnoses the account but does not fix it |
| Full replacement with new agency | Documented failures persist after a remediation period | Ask for a quote by scope | Transition loss; plan 2 to 4 weeks of audit |
The paid diagnostic route is how I work. The 30-day Google Ads Sprint is $7,500, positioned as hire me for 30 days before you hire me forever. Ongoing management after that is $3,500 per month minimum, month to month. No account managers. The person you talk to does the work.
Fixing measurement first isn't theory. In an ophthalmology practice, after tracking was rebuilt first, conversion rate moved from under 1% to 17% plus and cost per conversion fell 89%, from about $400 to about $44, platform reported, with 190 additional patient calls per month. The media didn't get smarter until the inputs were honest. tROAS, target return on ad spend bidding, and Quality Score, Google's estimate of ad and landing page relevance, only help when the conversion signal underneath them is real.
Pick the smallest option that answers your actual question. If you don't know whether the agency is the problem, buy a diagnosis. If you know tracking is broken, buy a rebuild. Don't buy a replacement relationship to solve an evidence problem.
Termination and disruption are separate decisions. You can end a contract on schedule without losing the measurement history the next operator needs, if you sequence the exit properly.
The timeline below puts access and contract terms first, which is the only safe order.
That last step is safer than most owners fear. Google's manager account ownership documentation states that a client account still owns its data even when a manager account has ownership, and the client can remove that ownership by unlinking the manager account. A manager can have full administrative access without taking your data ownership away. So if an agency insists it must own the account to manage it, that's a preference, not a technical requirement.
If any of those fail, fix access before you send notice. Agencies are most cooperative while they're still being paid. For broader context on vetting whoever comes next, see the hiring advice from Come Together Media.
End the contract on the date the contract requires. Preserve the account, the history and the measurement as if you'll need them for years, because you will.
Keep the notice short, dated and specific. Emotion in a termination letter creates disputes. Specifics create deliverables.
We are giving notice under section [X] of our agreement dated [date] that services end on [date], which is [30/60/90] days from today. Before that date, please confirm administrator access for [accounts], and provide full exports of the items listed in the attached checklist. Please confirm receipt in writing by [date].
Attach the checklist below to that notice and assign an owner on your side for every row. An asset isn't transferred because someone said it was. It's transferred when your owner opens it and signs off.
| Asset | What to confirm or export | Format and sign off |
|---|---|---|
| Google Ads access | Admin access for at least one business owned user | Screenshot of user list; owner signs off |
| Conversion actions | Full list with Primary or Secondary status and counting rules | Export; owner signs off |
| GA4 property | Admin access and event list | Event export; owner signs off |
| Google Tag Manager | Container export and version history | Container file; technical lead signs off |
| Call tracking | Numbers, routing, minimum call duration | Settings export; owner signs off |
| CRM stage mapping | Lead stages mapped to conversion actions | Mapping document; sales lead signs off |
| Audiences and customer lists | Names, sources and membership rules | Export; owner signs off |
| Product feed | Feed file, attribute mapping and rules | Feed export; owner signs off |
| Scripts and rules | Source code and schedules | Text file; successor signs off |
| Experiment history | Results, hypotheses and dates | Spreadsheet; owner signs off |
| Change log | Budget, bid, exclusion and test changes | Spreadsheet; owner signs off |
| Negative keyword lists | Shared lists and their applied campaigns | Export; successor signs off |
| Ad copy library | Active and paused ads with performance | Export; owner signs off |
| Creative and landing pages | Source files and live URLs | Shared folder link; owner signs off |
| Historical reports | Last 12 months of performance | Export with date ranges; owner signs off |
Creative files deserve special attention. If a commercial video production company or any outside vendor produced your video assets, confirm who holds the source files and usage rights now, not after the agency relationship ends and the vendor points back to the agency.
Pay attention to match types, the keyword settings that control how closely a search must match your keyword, when you export ad copy and negative lists. A successor rebuilding without your negative keyword history will repay for traffic you already learned to block. That tuition is avoidable.
Close the handover with one final report tied to a stated date range, conversion action list and attribution model. That's the baseline your next operator gets judged against.
Fire when delivery has failed across verified measurement and a written remediation period. Don't fire after one disappointing month or a fee increase. That's the whole decision rule, and everything above exists to help you apply it with candor.
A second opinion comes first because most owners who believe the agency has failed haven't yet confirmed that tracking, attribution and account access are sound. If measurement is broken, the verdict on the agency is premature. If measurement is clean and delivery still fails, the verdict gets much easier, and much harder for the agency to argue with.
Here's the order I'd use, from lowest commitment up. Start with the free audit checklist preview at cometogether.media/ppc-vault/free-preview and run the access checks yourself. If you want a senior operator to walk the account with you, book a 60-minute Office Hours session for $350. If the account needs rebuilding, the 30 day Sprint is $7,500, then $3,500 per month minimum if we continue. I take 2 to 3 new clients per quarter maximum, and the person you talk to does the work.
Book a call at cometogether.media/meetings/chase-mcgowan and bring your last 90 days of CRM outcomes, not just a dashboard. We'll know quickly whether you're looking at an agency problem, a measurement problem or an auction problem.
One action for today, before any calls get booked. Log in and confirm you personally hold administrator access to Google Ads. If you don't, that request is your first email, regardless of what you decide next.
If you're deciding when to fire your marketing agency and you want measurement checked by the person who would rebuild the account, Come Together Media LLC offers a $350 Office Hours review and a $7,500 30 day Google Ads Sprint with tracking fixed first, no account managers in between. Visit the site, look at the documented platform reported results, and book a call if a direct, tracking first second opinion is what your spend needs right now.