Most people asking this question already know the answer and are looking for permission. But not always — plenty of agencies doing genuinely good work get fired because the reporting is bad, and plenty of bad ones survive for years because the reporting is beautiful. So here's the version I'd use, and I run an agency.
I've inherited enough accounts to know what neglect looks like from the inside. The most expensive one I've seen: a Noida commercial developer came to us after a previous agency had spent ₹6 lakh over four months and closed exactly zero site visits. The reports had looked fine the whole time. Impressions were up. That's the trap — reach is easy to grow and means almost nothing.
Key takeaways
- Judge on cost per booked customer, not cost per lead. A cheap lead that never buys is the most expensive thing in marketing.
- Ask for the 30-day change log. A real operator produces it in minutes because the work actually happened.
- Creative volume is the engine. If the same three ads have run for two months, your results were always going to decay.
- You should own your ad account, pixel and creative files. Access is fine; ownership is not negotiable.
- Sometimes the honest verdict is your agency is fine and your offer is the problem. That section is included below.
THE THREE QUESTIONS THAT DECIDE IT
Send these in one message. Don't warn them, don't schedule a call for it. How fast the answers come back is itself the test.
| Ask this | A good agency | A drifting agency |
|---|---|---|
| "What's our cost per booked customer this month?" | Answers with a number, and explains what's included in it | Sends cost per lead instead, or asks what you mean |
| "Which creatives drove revenue last month?" | Names specific ads and shows the split | Sends a screenshot of the whole account |
| "What changed in the account in the last 30 days?" | A dated list — creatives shipped, audiences cut, budget moved | "We've been optimising" with no specifics |
Reach is easy to grow and almost meaningless. If the first number in your report isn't tied to money, the report was built to be looked at, not read.
WHY COST PER LEAD LIES TO YOU
This is the single most common way good-looking reporting hides a bad account. Lead cost is easy to push down — loosen the targeting, simplify the form, offer something free — and every one of those moves makes the leads worse. The number that survives contact with reality is what it costs to get a customer who actually books and pays.
Here's the shape of it from our own accounts. On a Noida commercial real-estate account we run, the cost per lead looked ordinary; the number that mattered was ₹14,800 per booked site visit at 5.6x return — because site visits, not form fills, are what turn into sales.
THE 10-POINT SCORECARD
Score one point per row where the "good" column is true. Be honest — you're the only one reading it.
Reporting (how they show you the work)
| # | Signal | Good | Bad |
|---|---|---|---|
| 1 | First number in the report | Cost per customer or revenue | Impressions, reach, CTR |
| 2 | Attribution to creative | Names the ads that worked | Account-level totals only |
| 3 | Bad news | Arrives from them, early | You discover it yourself |
| 4 | Reporting cadence | Predictable, same format | Only when you chase |
Operating (whether anyone is actually running it)
| # | Signal | Good | Bad |
|---|---|---|---|
| 5 | New creative shipped monthly | A steady batch, tested and killed | The same ads for months |
| 6 | Change log for last 30 days | Produced in minutes | "We've been optimising" |
| 7 | Tracking health | Conversions API live, events verified | Pixel only, nobody's checked |
| 8 | Who touches the account | A named person you can reach | Unclear, or rotating juniors |
Relationship (whether you're the client or the tenant)
| # | Signal | Good | Bad |
|---|---|---|---|
| 9 | Ownership of assets | You own account, pixel, creative files | Held in the agency's name |
| 10 | Willingness to say no | Pushes back on bad ideas | Agrees with everything |
WHEN YOUR AGENCY IS ACTUALLY FINE
I'd rather say this plainly than pretend every unhappy client has a bad agency. Three situations where switching won't help:
- Your follow-up is the bottleneck. If leads sit for hours before anyone calls, no amount of media buying fixes it. Response speed is usually the cheapest fix available and it belongs to you, not them.
- Your offer isn't sharp enough. If the market has a dozen versions of what you sell at a similar price, ads amplify indifference rather than create demand.
- It's been six weeks. Paid social needs roughly a 90-day arc before a trend is real. Month one is data, month two is where cost per acquisition should start bending.
If you want the buyer-side version of this, we wrote how to choose a performance marketing agency for people at the start of that process rather than the end of one.
IF YOU DO SWITCH
Protect the assets before you give notice, not after. Confirm the ad account, Business Manager, pixel or Conversions API setup and domain verification are in your business's name, and get every creative file handed over. An agency that hesitates on this has told you everything you needed to know.
Then ask the next team something specific: how much creative will actually be produced each month, and who is producing it. Volume is what makes paid social work — our own approach is built around that, which we've written up as the method we run, and we back the outcome with a 5x ROAS guarantee in writing.
WANT A SECOND OPINION ON YOUR ACCOUNT?
Bring your last three months. We'll run this scorecard with you honestly — including telling you to stay where you are if that's the right answer. Every engagement is scoped and quoted on the call.
Book a Scoping Call Ask on WhatsAppQuestions people ask alongside this one
What reports should my marketing agency actually send me?
A monthly report should open with cost per acquired customer, not impressions or reach. Below that you want spend, leads, qualified leads, closed customers, and the specific creatives that drove them, plus a plain list of what changed in the account that month. If the report leads with reach, engagement or CTR and buries the money, it's built to look good rather than to be read.
How often should my agency be changing things in the ad account?
On a healthy account you should see meaningful change every week: new creative shipped, losers switched off, budget moved toward what's working. A month with no changes usually means nobody opened the account. Ask for the change log — a real operator can produce it in minutes because the work actually happened.
Is a long lock-in contract normal for a marketing agency?
Three months is fair, because almost nothing meaningful can be judged in less. Twelve months with no exit clause is not fair, and agencies that insist on it are usually protecting their revenue rather than your results. We work on rolling terms after an initial scoping period, which is easier to defend when the work is good.
Should my agency own my ad account, pixel or creative files?
No. Your business should own the ad account, the pixel or Conversions API setup, the domain, and every creative file produced for you. Agencies can have access; they shouldn't have ownership. If leaving your agency means losing your data history, you were never really the client — you were the tenant.
How much should I expect to pay a marketing agency?
Market rates vary hugely by scope and country: published industry ranges put small retainers in the low thousands of dollars a month and senior full-funnel work far higher, with Indian retainers typically a fraction of US and UK equivalents for the same scope. What matters more than the number is what sits behind it — who touches the account, how much creative gets produced, and whether the fee is tied to output you can verify. We scope and quote every engagement on a call rather than publishing packages, because the honest answer depends on volume, markets and creative load.
My agency says results take time. How long is reasonable?
For paid social, judge on a 90-day arc. Month one buys data and kills losing creative, month two is where cost per acquisition should start bending, and month three should show a trend you can plan against. Beyond that, 'it takes time' stops being a reason and starts being an excuse — especially if nobody can show you what they changed while the time was passing.