The report says everything is up. The phone says otherwise. If that describes your last monthly review, the problem is usually not that the report is false. It is that the report is answering a different question from the one you are asking.
Only three numbers in an agency report decide anything: what was spent, how many enquiries it produced, and the cost per enquiry in rupees. Everything else, reach, impressions, engagement, is context. If the report leads with context and buries those three, read the rest with care.
This page teaches you to read the document you are already receiving: which metrics matter, how to check the report against the platform yourself, and the one honest-looking inflation to look for before anything else.
What are vanity metrics in an agency report?
Vanity metrics are numbers that rise easily and prove nothing about revenue: reach, impressions, likes, video views and raw clicks. They are not fake, they are incomplete. A metric earns a place in a report only if it connects to enquiries, customers or cost per result.
| Metric | What it measures | Can it pay a salary? | When it legitimately matters |
|---|---|---|---|
| Reach | How many accounts were shown the ad at least once | No | A brand-awareness push in a new city, where being seen is the stated goal |
| Impressions | How many times the ad appeared on a screen | No | Diagnosing delivery problems: an ad that never shows can never sell |
| Likes and engagement | Reactions, comments and shares on the creative | No | Testing which creative to put spend behind, before the spend |
| Video views | Plays past a threshold, often just 3 seconds | No | Comparing hooks in video-led campaigns, view-through cost only |
| Clicks and CTR | Visits to the landing page and the rate of them | Not on its own | Spotting a broken step: high clicks with zero enquiries points at the landing page |
| Enquiries (leads, calls, WhatsApp) | People who asked to buy from you | Yes | Always. This is the number the retainer exists to move |
| Cost per enquiry in rupees | Spend divided by enquiries produced | Yes | Always. It is the exchange rate between your money and your customers |
Notice the pattern in the last column. Every vanity metric has a legitimate diagnostic use inside the account. None of them belongs on the first page of a monthly report to an owner. When context metrics get the headline and the three deciding numbers need hunting for, the report is presenting, not reporting.
How do you verify an agency report against the platform?
Pick one number from the report, usually the lead or sale count, open the ad platform or GA4 with your own login, set the identical date range, and find the same figure in the raw data. Small gaps have innocent explanations. Large, repeated gaps in the report's favour do not.
Here is the method in full. It takes about fifteen minutes the first time:
- Pick one number. The lead count from last month's report is the best candidate, because it is the number closest to money.
- Sign in yourself. Use your own access to Google Ads, Meta Ads Manager or GA4, not a screenshot and not the agency's portal. If you do not have access, ask for read-only access, and note what happens when you ask.
- Match the date range exactly. A report for 1 to 30 June must be compared against 1 to 30 June in the platform. Date-range drift is the most innocent source of mismatch, so remove it first.
- Find the same metric. In Google Ads that is the Conversions column; in Meta it is Results; in GA4 it is the key event count. Confirm the metric name matches what the report claims to show.
- Compare and write both numbers down. A modest gap can be attribution: the platforms use different windows and models, and some conversions are modelled rather than observed.
- Ask before assuming. Show the agency both numbers and ask them to explain the difference in plain language. A good agency does this happily. A vague or jargon-heavy answer, month after month, is itself a finding.
If you want to go one level deeper into the account itself, the 30-Minute Owner's Check walks you through the change history, the search terms report and the conversions column, step by step, with no technical background needed.
The double-counted conversion: how 40 sales become 80
One of the most common inflators in agency reporting looks completely honest, and often is a setup mistake rather than a lie: the same action counted twice as a conversion.
An illustrative walkthrough. An online store has two conversion goals switched on in Google Ads: Purchase and begin_checkout. A customer clicks an ad, starts checkout, and buys. The platform records two conversions for one sale. If 40 people did that last month, the report says 80 conversions. Nobody typed a false number anywhere. The account settings did the inflating.
The check takes five minutes. In Google Ads, open the conversion actions screen (Goals, then Conversions, then Summary) and read which actions are set to count as conversions. In Meta, open Events Manager and see which events the campaigns optimise and report on. You are looking for two things: an action that is a step toward a sale (begin_checkout, add to cart, page view) being counted alongside the sale itself, and any single action set to count "every" conversion where "one per click" fits the business. If your report's conversion number is roughly double what your own sales or enquiry book says, this screen is the first place to look.
Want the reconciliation done properly, in writing? The Full Second Opinion reconciles your last three agency reports against raw platform data, line by line, and ends in a founder-signed verdict: Keep, Fix or Replace. Fixed fee, read-only access, and a 12-month ban on taking over any account we audit.
See the Independent AuditWhy do the report numbers look good when you get no customers?
Four places truth hides: the wrong action is counted as a conversion, the leads are junk or bots, the funnel leaks between form and phone call, or seasonality is doing the moving. Each has a distinct check, and none of them requires taking anyone's word for anything.
- The wrong action is counted. A "conversion" that is actually a page view, a scroll or a checkout-started tells you nothing about sales. Read the conversion actions screen, as above.
- The leads are junk. Export the raw leads and call ten. Count how many answer, remember enquiring, and are in a city you serve. Ten calls tell you more than any dashboard.
- The funnel leaks after the ad. Submit your own form and call your own number. If the form fails, or the enquiry sits unanswered for two days, the ads may be fine and the leak is in the handling.
- Seasonality is doing the moving. Compare against the same month last year, not last month. A festive-season lift is not agency skill, and a monsoon dip is not agency failure.
What should a good monthly report contain?
A good monthly report fits on one page and answers, in plain language: what was spent, how many enquiries it produced, the cost per enquiry in rupees, what changed in the account, what was tested, what failed, and what happens next month. If a report never admits a failure, doubt the rest.
Use this as a checklist against your own report:
- Total spend, split between ad spend and agency fee
- Enquiries produced, with the source data named
- Cost per enquiry in rupees, computed and shown
- What changed in the account this month, specifically
- What was tested, and what the test found
- What did not work, stated plainly
- Next month's plan, in two or three lines
These map directly onto the 10 monthly questions worth asking in every review meeting, along with the answers that should worry you.
How should agency fees appear on your invoice?
An honest invoice separates the agency fee from the ad spend, names the fee model, and shows GST on the fee if the agency is registered. Ad spend should be billed by the platform to your own card, inside your own account, never bundled invisibly into a single line.
There are three common fee models in India: a flat monthly retainer, a percentage of ad spend, and a hybrid of the two. None is wrong, but each rewards something different, and a percentage-of-spend model rewards higher spend as a plain fact of its arithmetic. What matters on the report is that you can always see the split: how much went to the platform and how much went to the agency. If your monthly payment is one opaque figure, ask for the split in writing. For what businesses typically pay under each model, see the indicative fee ranges for Indian agencies.
Frequently asked questions
What are vanity metrics?
Vanity metrics are figures that grow easily without proving revenue: reach, impressions, likes, video views and raw clicks. They are context, not results. A report built around them can look excellent while enquiries fall. Judge every metric by one test: does it connect to enquiries, customers or cost per result in rupees.
Why do my agency's numbers not match GA4?
Small gaps are normal. Google Ads, Meta and GA4 attribute conversions differently, use different windows, and count at different moments, so modest differences are expected. Large gaps are not. If the report shows roughly double the platform figure, check for duplicate conversion goals or a number taken from a different date range.
What is a good ROAS?
There is no universal good ROAS, because it depends on your margin. Compute your own break-even: divide 100 by your gross margin percentage. At a 40 percent margin, break-even ROAS is 2.5; at 20 percent, it is 5. A ROAS above your break-even earns money; below it, the ads lose money whatever the report calls it.
What if the report and the platform disagree?
Do not assume the worst first. Write both numbers down, note the date ranges and the exact metric names, and ask the agency to walk you through the gap in plain language. Attribution settings explain small differences. If the explanation is jargon, or the gap repeats every month in the report's favour, verify independently.
Are the leads in my report real people?
Check three things: call ten leads from the raw export and count how many answer and remember enquiring, compare lead locations against the cities you actually serve, and match the report's count against your own phone and WhatsApp records. Real leads survive all three checks; bot and junk leads fail them quickly.