Every month a PDF lands in your inbox. Impressions up. Clicks up. “Engagement” up. Green arrows next to almost every line. And yet when you look at your bank account, revenue is roughly where it was six months ago.
You’re not necessarily being tricked. Outright fake numbers are rare, because they’re easy to catch and expensive to get caught doing. The more common problem is a report that’s technically accurate and still designed to make the work look better than it is. Here’s how to tell the difference.
How can you tell if a marketing agency’s report is misleading?
Watch for three signs: growth figures with no baseline (“up 40%” from what?), charts whose Y-axis doesn’t start at zero, and numbers pulled only from the channel’s own dashboard. Honest reporting includes context, sample size, and a source you can verify independently.
Are the numbers reported against a baseline?
“Leads increased 40%” means nothing until you know 40% of what, and compared to when.
Forty percent over last month could be seasonality. Forty percent over the worst month of the year is cherry-picking. A trustworthy report shows the comparison period and, ideally, the same month last year. If every figure is a percentage with no raw number and no time frame attached, that’s a choice, and it’s usually not an accident.
Do the charts start the Y-axis at zero?
A line chart with the Y-axis starting at 90 instead of 0 turns a 5% rise into a wall.
This is the oldest trick in reporting, and it’s covered in detail in how to read a chart without being fooled. Check the axis on every chart in the deck. If the impressive-looking ones are zoomed in and the flat ones are shown full-scale, someone chose which story each chart would tell.
Are they reporting vanity metrics or money?
Impressions, reach, and engagement are activity, not results. Leads, sales, revenue, and cost per acquisition are results.
A report heavy on the first group and light on the second is measuring effort, not outcome. That’s worth pushing on. As covered in what a metric actually measures, the question for every number in the deck is: if this went up and revenue didn’t, would we still call it a win? If the honest answer is no, it doesn’t belong at the top of the report.
Can you trace the numbers back to a source you control?
You should be able to open Google Analytics, your CRM, or your payment processor and reproduce the headline figures yourself.
Ask for read-only access to the actual ad accounts and analytics properties, not just the summary dashboard the agency builds. If the reported conversion count is 300 and your CRM shows 180 for the same period, that gap is the conversation. It doesn’t mean anyone lied, but it means the report is measuring something other than your real results, and you need to know what.
Do they ever show what didn’t work?
A report with no losses, no failed tests, and no underperforming campaigns is a sales document, not an analysis.
Real marketing has misfires every month. An agency that’s actually experimenting will have a creative that flopped or a channel that underdelivered, and a good one will tell you about it alongside what they’re changing. Uniformly positive results, month after month, is not a sign of excellence. It’s a sign of curation.
Are conversions being double-counted?
If the agency runs Facebook, Google, and email, each platform will claim overlapping credit for the same sales.
Add up what every channel reports and you can easily “attribute” 150% of your actual revenue. This is the attribution problem in short, and it’s why the platform-reported totals in a multi-channel report should never be summed. Your one real number is the order count in your own system.
What should you ask for?
Four things make a report hard to fake and easy to trust:
- Read-only access to the underlying ad and analytics accounts.
- One agreed source of truth for revenue and conversions, usually your CRM or processor, that every report reconciles against.
- Results measured against a goal you set in advance, not against whichever comparison period looks best in hindsight.
- A holdout or geo test on your largest channel at least once a year, so at least one number in the relationship isn’t self-reported.
An agency doing good work will have no problem with any of these. Hesitation on all four is the actual red flag, more than anything in the PDF itself.
The takeaway
You usually can’t catch a misleading report by staring harder at the report. The numbers in it are probably real. What you can do is change what you’re looking at: check the baselines, check the axes, separate activity from money, and reconcile every headline figure against a system you own.
The agencies worth keeping will welcome that scrutiny, because it makes their good months provable. The ones that resist it are telling you something the report never will.