Managing a marketing agency comes down to four habits: you own the accounts and data, you give the agency briefs it can execute, you judge it on metrics it actually controls, and you review it on a fixed rhythm. Founders without a marketing lead tend to skip all four, then discover months later that nobody can say whether the agency is working.
Set up ownership first: ad accounts, analytics and creative files
Before you judge performance, make sure you could end the relationship tomorrow and keep everything you paid for. When I step into a company that runs agencies, this is the first thing I check, and it’s one of the most common gaps I find.
The rule is simple: your business creates every account, your payment method pays for media, and the agency gets user or partner access.
| Asset | Who should own it | What the agency gets |
|---|---|---|
| Google Ads account | Your company, created under your Google login | User access, or a link to the agency’s manager account |
| Meta ad account, Page and pixel | Your own Meta business portfolio | Partner access to specific assets |
| LinkedIn ad account and company page | Your company | Campaign Manager role on the ad account; the lowest page role that lets them run ads |
| GA4 property and Google Tag Manager container | Your company’s Google account | Editor or Marketer in GA4; Edit access in GTM, with Publish only if they need it |
| Domain, website, landing page tool | Your company | Editor, never owner |
| CRM and email platform | Your company | A user seat with limited permissions |
| Creative, copy and source files | You, once paid, per the contract | Upload rights to a shared drive you own |
Run this checklist once, then again every quarter:
- At least two people at your company have admin rights on every platform
- Media is billed to your card or invoice account; if the agency rebills media, you see the platform invoices
- The contract says work product, including layered and source files, belongs to you on payment
- Final and source creative files land in a folder you own every month
- An offboarding clause requires the agency to hand over access, naming conventions and documentation within a set number of days
Writing a brief an agency can actually execute
Most weak agency work traces back to a weak brief. “Get us more leads” is a wish, not a brief. A brief the agency can execute answers nine questions:
- Business goal, as a number. A hypothetical example: “20 sales-qualified demos a month from paid channels at or below $500 each.”
- Audience. Who buys, who doesn’t, and which segments to exclude.
- Offer and message. What you’re asking people to do and why they should do it now.
- Budget and pacing. Monthly cap and how much it can flex.
- Constraints. Brand rules, claims you can’t make, regions and compliance limits.
- Available assets. Existing creative, customer proof, landing pages, and who produces anything new.
- Success metric and system of record. Which number counts, and where it’s measured: your CRM or store, not the ad platform.
- Decision rights. What the agency can change without asking, such as bids or budgets within a set range, and what needs your sign-off, with a turnaround time.
- Deadlines. Launch dates and when the first results review happens.
A useful test: give the brief to someone who missed the kickoff call. If they can’t tell you what success looks like and what’s off-limits, rewrite it.
KPIs the agency controls vs influences
Split every KPI into three layers before the agency starts. This single step prevents most arguments later.
| Layer | Examples | Owner | How to use it |
|---|---|---|---|
| Controls | Launch dates hit, creative variants shipped, spend pacing, tracking accuracy, click-through rate, cost per click, conversion rate on pages they build | Agency | Hold them accountable; this is the contract |
| Influences | Cost per qualified lead, sales-qualified lead rate, pipeline from paid, new-customer CAC, marketing efficiency ratio (MER) | Shared | Set targets together; review the full chain when they miss |
| Doesn’t control | Close rate, pricing, product fit, sales follow-up speed, retention | You | Don’t blame the agency, but expect them to flag problems they see |
The mistake runs both ways. Founders hold agencies to revenue they can’t control, so good agencies get fired for sales problems. Agencies hide behind the metrics they do control, so a strong click-through rate covers for an empty pipeline.
My working rule: judge the agency’s execution on the controlled layer, and make renewal decisions on the influenced layer. Measure the influenced layer in your CRM or store backend, and bring those numbers to the review yourself.
The weekly, monthly and quarterly review rhythm
A fixed rhythm replaces vague “how’s it going” calls with decisions.
| Cadence | Length | Agenda | Output |
|---|---|---|---|
| Weekly | 30 min, or async | Spend vs pace, what launched, what’s blocked, what’s next | Blocker list with owners and dates |
| Monthly | 60 min | Controlled KPIs vs target, influenced KPIs from your CRM, tests run and what they taught, next month’s plan | Updated plan and test queue |
| Quarterly | 90 min | Scorecard, strategy check, scope and fee review | Keep, fix or change decision |
Three habits make the rhythm work. Ask for the report 24 hours before the meeting, so the meeting is for decisions rather than reading. Keep a shared test log with the hypothesis, result and next step for every test. And split the data duties: the agency brings platform numbers, you bring CRM or store numbers, and the conversation happens in the gap between them.
Red flags in agency reporting
Reports can look busy and still hide a problem. Watch for these:
- The metrics change month to month. New KPIs appear when old ones slip.
- Only platform-reported conversions. Platforms overlap and claim credit generously, so nothing is reconciled against your CRM or store.
- Conversions that aren’t business outcomes. Page views, scroll depth or engaged sessions counted as conversions next to real leads.
- Branded search inside the totals. Blended results hide that many “wins” came from people already searching for your name.
- No test log. Nobody can say what was tried, what was learned or what changes next.
- Screenshots instead of access. You see curated images, not the live account.
- Every miss has an outside cause. The algorithm, seasonality or a learning phase is a fair answer once, not three months running.
- Spend doesn’t reconcile. Reported spend differs from platform invoices, with no explanation.
- Activity reported as results. A count of “optimizations made” tells you nothing about outcomes.
If paid search is a large part of the scope, check the account yourself each quarter with a Google Ads audit checklist, or have someone independent do it.
When to renegotiate, replace or bring work in-house
Use the scorecard trend, not a single bad month, to decide.
| Situation | Likely move |
|---|---|
| Execution is solid, but results are soft because scope or budget is spread too thin | Renegotiate: fewer channels, more depth |
| Retainer hours go to reporting and meetings instead of work | Renegotiate the fee structure or automate reporting |
| Controlled KPIs missed two months running after a written fix plan | Replace |
| Reporting red flags, or access withheld when you ask for it | Replace, starting with recovering access |
| Work is steady and core, volume justifies a salary, and someone can manage the hire | Bring it in-house |
| Work is spiky or needs rare skills, such as creative bursts or tracking builds | Keep it external |
When you replace an agency, recover access and export data first, then give notice. Where the contract allows, overlap old and new agencies for a few weeks so campaigns don’t go dark during the handover.
Sometimes the real problem isn’t the agency. It’s that nobody on your side can set strategy and hold vendors to it. That’s a leadership decision, and fractional CMO vs full-time CMO vs agency walks through the options. Running agency briefs, reviews and scorecards is a core part of my fractional CMO work, so founders get a clear read without managing it themselves.
Agency scorecard template
Score each area from 1 (poor) to 5 (excellent) at the quarterly review, multiply by the weight and add the results.
| Area | What to score | Weight | Score (1-5) |
|---|---|---|---|
| Ownership and access | You own every account; access is current; files delivered | 10% | |
| Execution | Launches on time, creative volume, sticks to the brief | 20% | |
| Controlled KPIs | Hit agreed targets on metrics they control | 25% | |
| Influenced outcomes | Qualified leads, pipeline, CAC or MER trend in your systems | 20% | |
| Reporting quality | Stable metrics, reconciled to CRM, clear test log | 15% | |
| Proactivity | Brings ideas, flags problems outside their scope | 10% |
A hypothetical example: scores of 5, 4, 3, 3, 2 and 4 give a weighted total of 3.35, which lands in the fix-plan band. These are my working thresholds, not industry benchmarks:
- 4.0 and above: keep, and consider expanding scope.
- 3.0 to 3.9: agree on a written fix plan with named actions and recheck in 60 days.
- Below 3.0 twice in a row: start the replacement process.
Share the scorecard with the agency at kickoff. Good agencies want to know the rules they’ll be judged by, and the ones that resist it tell you something too.
Get it built
If you’re paying agencies without a marketing lead and can’t tell what’s working, I can set up ownership, briefs and the review rhythm, then run it. Start with a Growth Audit, $1,500 fixed and credited if we continue. See pricing or get in touch.