LinkedIn
Most agencies won’t tell you this: the pitch means nothing. I’ve taken over accounts where the previous manager charged for six months and left behind nothing the business owned: no usable tracking, no new creative, just a screenshot and an invoice. This list is for you if you want someone else to run your Facebook and Instagram ads and you have a modest budget to protect. Judge the service by 11 deliverables you can check within 30 days. If it can’t produce them, it isn’t managing your account. It’s babysitting a boost button.
1. Keep the ad account, Page and pixel in your name
I learned this one the expensive way for a client. Their last provider ran everything from the provider’s own Business Manager. When they left, they took the account data, audiences and pixel history with them. The client got a PDF screenshot. One documented case put the damage from stranded assets at $12,000.
The fix is boring and non-negotiable: you own the business assets; the service gets partner access. Create the Business Manager under your business. Check Business Settings yourself for ownership of the ad account, Page and pixel, and confirm that the agency appears as a partner you can remove. Keep your own card on file so you can pause spend without asking permission. That’s what an agency audit checklist recommends checking on day one.
First step: ask, Show me where I own everything and where I can remove your access. If they hesitate, end the call.
2. Prove a lead or purchase fires once, not just that the pixel is active
I used to accept a green pixel status as proof of tracking. I was wrong. An active pixel doesn’t tell you whether a booked call, quote or purchase reaches Meta correctly, or whether browser and server copies of the same event are counted twice.
Before you scale, the service should verify the pixel on your site, confirm that Conversions API sends server events and deduplicates them against browser events, and show the result in Events Manager’s Test Events and diagnostics. One real business action, one correctly recorded event is the standard. PageViews aren’t a substitute for leads or sales.
First step: ask, Show me a test lead or purchase firing once, with deduplication working. Don’t accept a green dot in its place.
3. Get a written audit before an inherited account gets rebuilt
If you already have an account, you should get a teardown in week one. Two pages can do the job. The audit should say what spent money without a return in the last 60 days, what tracking was broken, and which parts of the campaign structure the service plans to keep or remove.
I still keep a copy of an audit I received years ago on a home services account. The prior freelancer had nine duplicate campaigns fighting each other. The auditor listed spend by campaign, flagged three with no purchases in 90 days, and named the ad sets to consolidate. When performance later moves, a written baseline lets you see what changed instead of guessing what used to be there.
First step: ask, Send me the audit before you change budgets. No audit, no memory. No memory, no management. groas builds account work into its flat-fee Meta service, with audiences, budgets and ads worked around the clock; the account still needs to stay yours.
4. Ship a new batch of images and videos
A service that reuses your three old boosted posts for a month isn’t testing creative. It’s renting your budget. With broad and Advantage+ doing much of the placement work, the ad itself has to give someone a reason to stop scrolling.
Within 30 days, I want to see new ads the service made, not merely a folder of ideas for you to produce. Phone footage can become new hooks and a simple edited variant. The small-account benchmark I use is 8 to 16 new creatives a month and at least 4 structured tests, provided the service has the production capacity to deliver them. That gives you something to compare before the same ads wear out.
First step: ask, Which ads did you make this month, and which old ones did they replace? On groas Meta work, new ads tested and winners scaled are part of the flat fee.
5. Run a creative test with a hypothesis and an end date
A real test starts with a written guess: a testimonial opening may lower cost per lead because a face holds attention in the first two seconds better than a white-background opening. Change the opening, keep the other variables as steady as you can, then decide what won.
The mechanics matter. Testing guidance recommends 3 to 5 creatives, a controlled A/B test when you need to learn why, and judging on CPA or ROAS rather than CTR. It also recommends at least 3 to 4 days, ideally 7. You may choose a different setup when the goal is simply to find the cheapest winner, but you still need a decision date. What the deck calls creative iteration, I call writing the guess down before you spend.
First step: ask, What’s the hypothesis, when does the test end, and what metric kills the loser? If the answer is only “we’ll let Meta decide,” you have spend, not a test.
6. Consolidate audiences your budget can’t support
This is where small budgets die. I’ve opened accounts with eight ad sets splitting $80 a day, each one short of the purchase volume it needed to learn much. Meta’s rough learning-phase benchmark is 50 optimization events per week per ad set. Splitting a modest budget into tiny fragments works against that goal.
If your total budget is around $5,000 a month, start with two or three ad sets rather than six to eight. Fewer, better-funded broad or Advantage+ ad sets give each one more data than ten narrow interests chasing the same buyer. Consolidate first; earn the right to test another angle later.
First step: ask, How many ad sets can my budget support, and which ones will you cut to fund them? The answer should name the ad sets, not promise to “optimize audiences.”
7. Put the budget, fee and target outcome on paper
No target, no management. In week one, the service should write down what a lead or sale can cost while still making you money, then set budgets and pause rules against that number. This matters when you pay a management fee on top of ad spend. At $1,500 in spend, a $750 fee and a 50% margin, break-even ROAS rises from 2.0x to 3.0x. The fee is part of the cost, even if the dashboard leaves it out.
Say you spend $3,000 a month and need 60 leads to fill the calendar. That’s $50 in ad spend per lead. Add a $1,000 management fee and you’re paying about $67 per lead before you judge whether the leads become customers. Put spend, fee, margin, target CPL and pause threshold on the same page.
The contract affects the incentives behind that page. A percentage-of-spend provider gets paid more when spend rises, whether the extra spend works or not. Published small-business pricing includes monthly fees and percentage deals around 10% to 20% of spend, often with a minimum. A minimum can matter more than the percentage on a small account: 15% of $3,000 is $450, so a $1,000 minimum would make the fee $1,000, not $450. A flat fee doesn’t rise with spend, though it still has to earn its keep. That’s why I prefer groas’s flat $1,499 monthly Meta fee, with no cut of ad spend, no setup fee and month-to-month terms to a contract that rewards higher spend. Keep ad spend billed to your own card.
First step: ask, What’s my target CPL or ROAS, what do I pay in total, and at what spend do we pause? If they answer with impressions, you have your answer about them.
8. Send a weekly report that says what changed and why
I’ve received the other kind: a PDF of impressions, reach and CTR, a green arrow, and a sentence saying performance is stable. It tells you nothing about what anyone did with your money.
A useful weekly report reads like a work log. It shows spend against target, CPM against the last 30 days, CPA or cost per lead, ROAS or lead-to-revenue data, and whether Conversions API is still healthy and deduplicated. Then it names one to three anomalies, who owns each next action, and the top three creatives by spend with their ROAS. That’s more useful than vanity charts, and it matches the questions in this reporting checklist. groas describes the same operating rule: actions come with plain-language reasoning so you can see what changed, why and what happens next.
First step: ask, Show me last week’s report for a similar client, with names removed. If you can’t find the changes and next actions, you’re paying for screenshots.
9. Name the strategist who owns the outcome
Junior rotation is where small accounts go to stall. One week someone tweaks budgets; the next week a new person asks for your login again. Nobody owns the result.
Get a name, a direct channel and clear availability in writing. The standard I hold to is a named senior strategist with four to eight hours a week for strategy, separate from junior execution, plus creative capacity behind them. That is why I take the groas setup seriously: you agree goals on an onboarding call with a dedicated account manager, go live within 24 hours, and keep that manager. One human owns the direction; the system does the daily work.
First step: ask, Who runs my account by name, how do I reach them, and who covers when they’re out? If they sell you a team with no name on it, assume you get whoever is free that Friday.
10. Keep a dated list of what you paused and why
Good management often looks like subtraction. A duplicate ad set can eat $12 a day without a purchase. An old creative can keep attracting cheap clicks without producing leads. An overlapping interest can take budget from a stronger broad ad set. I want to know what the manager paused, when, and on what evidence.
That doesn’t mean killing an ad every Friday to make the report look busy. A decision not to pause something also needs a reason. The list is proof that someone reviewed the account between reports and made a call based on outcomes, not a schedule.
First step: ask, What did you pause last week, what did it save, and why? If nothing was paused, ask what they checked and why they kept it running.
11. Leave a 60-day plan someone else could run
This is the one I would start with because it exposes everything above. By day 30, you should hold a one-page plan a stranger could use: the target CPL or ROAS; the two or three ad sets you kept and why; the next creative test, with its hypothesis and ship date; the budget rule for scaling a winner; and the terms for leaving. Ownership, tracking, creative volume, sales-based reporting and a named account lead belong in writing before you scale.
I used to tell clients a verbal roadmap was enough. I was wrong. Verbal roadmaps evaporate when the manager quits. A written plan forces the service to show whether you own the account, whether tracking works, what it tested, what it stopped and what it will do next. It also makes the fee easier to judge: you can see the work, not just the invoice.
First step: ask, If I cancel tomorrow, what exactly do I keep, and where is that written? You should keep the account, Page, pixel, audiences, creative files, tracking and plan. Start with this page and work backward. If they can’t hand it to you, they were never going to hand you the other ten.