Free Ad Leak Check

Sample report. Example Bottle Co. is a fictional business. Every number is invented for illustration.

Sample report: Example Bottle Co. (fictional)

Meta ads, the last 90 days. Prepared by Adlume. Reviewed and signed by our strategist.

The one thing

Meta counted 212 sales. Your store shipped 131. It is learning from sales that never happened, so fix the count before you change a single ad.

What's working: leave it alone

Retargeting. It takes 4% of your spend and brings 22% of your real orders. Don't cut it while you fix the rest.

The leaks, in fix order

1. Meta counts sales that never happened.This week.

The leak
The purchase is sent twice, once from the browser and once from your store, and Meta counts both.
Your number
212 purchases in Ads Manager, 131 orders in your store, last 90 days (Ads Manager, Results column; your store's orders report).
Compared to
Your store's order count, which is the real one.
What to do
Send the purchase event once, from the order confirmation, with deduplication on. Your developer, under an hour.

Why it matters here: Meta chooses who sees your ads from the sales it counts. Until the count is right, it learns from 81 sales that never happened.

2. Clicks that never load the page.This week. Your developer.

The leak
People click, then leave before the page appears.
Your number
4,120 of 6,180 link clicks became landing page views (67%), last 90 days (Ads Manager, Landing page views column). The page takes 5.8 seconds to show on a phone.
Compared to
Our starting line of 70% or more, calibrated on your account; and Google's web.dev guidance that the main content should show within 2.5 seconds.
What to do
Shrink the top image and remove the pop-up script that loads before the page. Your developer, about a day.

Why it matters here: you pay for every click, and 2,060 of them never saw the bottle.

3. A budget split too thin to learn.Next two weeks.

The leak
7 ad sets share $90 a day, so each gets about $13.
Your number
6 of 7 ad sets show "Learning limited" (Ads Manager, Delivery column).
Compared to
Meta's guidance that an ad set needs about 50 results in a week to finish learning. At $62 a real order, $13 a day buys fewer than 2 a week.
What to do
Combine them into 2 ad sets and stop editing them for a week. You or whoever runs your ads, under an hour.

Why it matters here: the platform never gets enough sales in any one place to find your buyers.

4. One ad carrying the account.Next two weeks.

The leak
One ad took 78% of the spend for 9 weeks, and its first frame is your logo.
Your number
14 in 100 people who see it are still watching at 3 seconds (Ads Manager, 3-second video plays divided by impressions).
Compared to
Our starting lines: no single ad above 60% of spend for 3 weeks, and at least 25 in 100 still watching at 3 seconds.
What to do
Run the 3 briefs below, 3 different ideas, and keep this ad running beside them. Your editor, about a week.

Why it matters here: when this ad wears out, nothing is ready to take its place.

5. The offer gives no reason to buy.Next two weeks.

The leak
Every ad leads with 20% off, and none says why this bottle is worth buying.
Your number
20% off is the first line of every live ad, last 90 days (Ads Manager, ad previews).
Compared to
Your own facts sheet: ice for 24 hours and free returns for 30 days, and no ad says either.
What to do
Open each ad with one reason from the facts sheet, then the discount. Whoever writes your ads, under an hour.

Why it matters here: a discount alone gives a stranger no reason to pick this bottle.

Ads or after the click: the verdict

The ads, first. Meta is learning from 81 sales that never happened, and 1 in 3 paid clicks never reaches the page. Each real order cost $62 in ads, not the $38 Ads Manager shows. Fix the count and the page first, then judge the ads again.

What to fix first

This week

  • Send the purchase event once, from the order confirmation.
  • Shrink the top image and remove the pop-up script.

Next two weeks

  • Combine 7 ad sets into 2.
  • Launch the 3 new ads below.
  • Open every ad with a reason to buy.

Later

  • Judge the new ads after $150 each.

Your next 3 ads

Next ad 1 of 3: The 24-hour car test

The idea
One bottle, one hot car, one whole day, so the ice is the proof.
Who it's for
People who've had a drink go warm in the car and just noticed it again.
The evidence
Your current ad takes 78% of spend and keeps 14 in 100 viewers past 3 seconds. Its first frame is your logo.
0 to 3 seconds
Close on the bottle on a sunbaked dashboard, hand-held (0 to 1.5 s). A hand shakes it and the ice rattles (1.5 to 3 s). On screen: "Ice at hour 24."
The body
Timestamped cuts of the same bottle in the same car at 8 am, 1 pm and 8 pm (3 to 10 s), each with its time on screen. The cap comes off and ice pours into a glass (10 to 13 s). On screen: "Free returns for 30 days." (13 to 15 s).
The proof
The owner's own timestamped phone video: same bottle, same car, no cuts inside each clip.
Offer and call to action
"Try it for 30 days." Button: Shop now.
Specs
9:16 and 4:5, 15 seconds, captions burned in, words inside the middle 70% of the frame, the ice sound up front.
Do not
Say "forever" or "guaranteed cold"; cut inside a timestamped clip; leave Meta's automatic text changes on, so "24 hours" is never rewritten.
Judge it by
People still watching at 3 seconds against your current 14 in 100, after $150 of spend; then cost per real order.

Next ad 2 of 3: 30 days to decide

The idea: lead with free returns for 30 days, so trying the bottle costs nothing but the wait.

Full brief in the real report.

Next ad 3 of 3: The plain ice test

The idea: fill the bottle with ice, start a timer on screen, and open it 24 hours later. One shot, no talking.

Full brief in the real report.

The copy, for ad 1

Primary text, short

Ice at hour 24, even in a hot car. Free returns for 30 days.

Primary text, medium

Left your drink in the car again? This bottle kept ice for 24 hours on a sunbaked dashboard. Try it, and send it back free within 30 days if it doesn't earn its place.

Headlines

Still ice after a day in a hot car

Ice at hour 24

30 days to decide, returns free

Method and limits

Dates: the last 90 days. Attribution: Meta's click windows and 1-day view. What we couldn't see: orders placed by phone or in person. Benchmarks appear only as dated context, never as targets. Lines marked "our starting line" are our own, calibrated on each account, not platform standards.

What we never promise: sales, leads, a lower cost per result or a return on ad spend.

Want this for your account?

Get the plan

$1,000 paid in full at checkout. Your 5 business days start when your intake and view-only access are in. Plus sales tax where it applies.

Or start with the free Ad Leak Check