18 questions. 30 minutes. And you find out if your agency actually works.

TThe thought that eats at you after seven months with an agency on Meta and Google sounds like this: I do not know if they are good or bad. Something is happening, but I do not see extra money in the account.

The answer does not come from a call. It comes from a list. The 18 questions below. Many people stop at question 12 and already know what to do.

If you have the same doubts, set aside half an hour without interruptions, open three tabs and put the last three agency reports next to you. Walk through the list. Mark yes or no at each one. At the end you have a score and a next step.

It is not about being paranoid. It is about knowing where you stand. Half the founders I talk to do not know if their agency is good or bad. They only know they are paying and waiting. That is not a strategy. That is a bet.

I structured the 18 questions into four parts. Each part takes 5-10 minutes. Total, 35 minutes max. At the end you have a score you can act on immediately. No advanced technical knowledge required. You just need to know how to read a report and open Ads Manager.

Part 1. Tracking and access. Ten minutes.

The first five questions get answered by logging in. If you do not have direct access, you already have your first problem. And it is a serious one.

1. 01Are you, on your personal email, admin on Meta Business Manager and Google Ads? Not employee, not viewer. Admin. If not, request it now. It is your account, not theirs. A no here is a red flag, full stop. There are founders who leave agencies and cannot take their accounts because the agency held them in its own name. Audiences, pixel, campaign history, all stay at the agency.
2. 02Open Events Manager in Meta. Find Purchase. Do you see Receiving events with a green tick, or Issues detected? Issues sitting unfixed for weeks means half your sales are not tracked. Meta learns on half the data. The result is optimization on half the truth.
3. 03Same Purchase event, click Deduplication. Is dedup between pixel and CAPI above 80%? Below 80%, the data is mixed and the algorithm learns wrong. There are accounts with 40% dedup where Meta thinks every sale is a duplicate. The algorithm reduces bids because it believes it already had the conversions.
4. 04Jump to GA4. Reports, Monetization, Ecommerce purchases. Last 30 days. Does GA4 revenue match Shopify or ERP revenue? Accept a 5-10% gap. Over 10%, tracking is leaking sales. There are accounts where GA4 reports 40% less than Shopify. 40% of sales invisible to Google Ads.
5. 05Google Ads, Tools, Conversions. Is Purchase sourced from Imported from GA4 or Google Tag? Old Universal Analytics tags or anything you do not recognize means you are likely double-counting conversions. There are accounts where Google reports twice as many conversions as actually existed. The agency shows 6x ROAS. Real ROAS is 3x.

These five questions tell you if the foundation exists. Fail two or more and there is no point going further. The agency does not control the account or does not know what it is doing. Every report that follows is built on sand.

Part 2. What the reports actually say. Ten minutes.

Open the last three reports the agency sent. Not the chat, not the meeting. The actual document. PDF, Excel, link. What it says matters. What they said on the phone does not stick.

1. 01Is POAS (Profit On Ad Spend) shown anywhere, not just ROAS? If all you get is revenue and ROAS, they are optimizing on the wrong target. They can hit ROAS 4 and still put you in the red. Picture a report with ROAS 5.2 where POAS is 0.8. You are losing money and do not know it. Because nobody calculated POAS.
2. 02Are numbers broken down per campaign, grouped by objective? A single account-level number hides that three campaigns carry the weight and seven quietly bleed money. Picture a report showing ROAS 2.8 at account level. Broken out: one campaign at ROAS 8. Three at ROAS 0.3. The average looks fine. Reality does not.
3. 03Does it state explicitly what they changed this week, what they paused, launched, new creatives shipped, and why? If it reads optimizing audiences three weeks running, they probably touched nothing. A good report: We stopped campaign X, CPA was EUR 45, double the target. We launched Y with a new lookalike audience.
4. 04Is there a section on what did not work, with explanation? An honest report includes misses. If everything has been rosy for three months straight, things are being hidden. There are reports with no paused campaign, no negative decision, no lesson learned. That is not professionalism. That is a cover-up.
5. 05Do you see AOV, CAC, LTV, or at least average margin, not just revenue and ROAS? Missing business numbers mean the agency works disconnected from your actual business. If they do not know what a customer is worth long-term, how do they decide if EUR 30 CPA is good or bad?

Reports mirror the agency. Thin report, vague, no decisions? Agency is the same. A good report shows not just what happened. It shows what was decided, what was learned, and what changes next. If your reports do not do that, you do not have a strategic partner. You have a paid spectator.

Part 3. How they communicate. Five minutes.

Answer from memory. Last 90 days. Be honest. Do not try to find excuses for them.

1. 01Last time something broke (pixel down, accidental pause, budget overspent), who spotted it first? You or them? If you caught it every time, they are not watching the account.
2. 02Have they ever, on their own initiative, suggested pausing a campaign that brought revenue but lost money on POAS? If never, they are not watching profit either. This is the honesty test. An agency that sees ROAS 2 and POAS 0.7 and says nothing is not on your side. It is on the side of its own invoice.
3. 03Over the last three months, how many calls were we propose testing X versus plain status updates? If all of them are updates, you are paying for stenography, not strategy. A strategic agency shows up with ideas, test proposals, hypotheses. If you are always the one proposing what to test, you are doing strategy. They just press buttons.
4. 04At the last meeting, did anyone ask about your current net margin, stock by category, or the season ahead? If nobody asked for business context, how can they optimize for profit?

Communication tells you everything about the relationship. If you find the problems, propose the tests, and bring the context, then what are you paying for? Why give EUR 2,000, 3,000, 5,000 a month? For someone to click buttons in Ads Manager? You can learn that in two weekends.

Part 4. Small signs that get expensive. Five minutes.

The last four questions. They look like details. These details add up to thousands of euros lost.

1. 01Have you seen charges larger than agreed on Google or Meta invoices in the last three months? Unexpected overage means the budget is not being watched. There are accounts where EUR 3,000 extra gets spent in one month because someone forgot to set a cap. The agency did not notice. Or did not want to.
2. 02Have campaign changes been made without a heads-up before or at least a note after? Not micro-tweaks. Impact decisions: pausing a campaign, switching objective, doubling budget. An agency that pauses a campaign without asking does not respect you. An agency that doubles budget without asking spends your money without consent.
3. 03If you ask right now for a list of all active custom audiences in Meta, with source for each, do you get it within 24 hours or does the reply drift a full week? Turnaround time is the best proxy for real control. If they cannot produce a list in 24 hours, they do not know what is in the account.
4. 04Last time you asked for a report at individual ad level (not campaign, not ad set), how long did it take? More than two days means they look at dashboards, not at creatives. Creative is half of Meta success. If the agency does not analyze which ads work and which do not, they optimize halfway.

What to do the moment you write the score

Whatever the score, do not jump to conclusions on emotion. Do two concrete things, in this order.

First: send them the list, all 18 questions, with a seven-day deadline. Simple wording: I want to understand what is happening on the accounts. Can you reply point by point? Their reaction tells you more than the answers. Defensive, or a request for a call to discuss? You have confirmed the diagnosis. Serious, point-by-point, with evidence? They earned the benefit of the doubt.

Second: start quietly looking at alternatives. Sign nothing. Do nothing radical. Build a shortlist of two or three options. When you have options, decisions are made calmly. When you do not, you cling to what you have. That is the worst negotiating position.

One detail that saves time later: whatever you decide, screenshot everything now. Reports, dashboards, custom audiences, conversion lists in Google Ads, Events Manager in Meta. If you switch agencies in three months, migration takes one day instead of three weeks if you already have captures saved on a drive.

A truth many founders avoid

Switching agencies is uncomfortable. It is stressful. It is full of uncertainty. But staying with one that is not delivering is worse. You do not just lose money. You lose time. And you do not get time back.

There are founders who stay over a year with a bad agency because they are afraid to switch. In all that time they lose tens of thousands of lei on ads that bring no profit and they lose a year the business could have been growing. When they switch and work with someone competent, first results usually come in month two. And they realize they could have been there a year earlier.

Do the audit. Write down the score. And make the decision. Even if it is hard. The discomfort of now is worth more than the regret of a year from now.