My ads are running but sales are flat — where do I start?
Last updated:
Short answer
Start with measurement, not the campaign: confirm an order is actually recorded, then read the cost at the event that reaches the bank rather than at add-to-cart, then open your checkout on a phone the way your customer does, then calculate your margin to see whether the campaign can be profitable at all. Targeting is the last suspect, not the first, and in most accounts I have reviewed the problem sat in one of those first three steps.
First: is the number you are looking at real?
Before touching the campaign, make sure the dashboard reflects what happens in your business. The most common silent fault is a test event code left switched on in the setup: every server-side event goes to the testing tool, and the campaign reads as though it produced nothing while it is in fact working.
Next comes the duplicated event: the pixel records the purchase, the conversions API records it again with no shared identifier, and orders appear at twice their real number. The reverse happens too — an event that never fires on the thank-you page, so the campaign reads as zero.
The practical test is simple: export the last fourteen days of orders from your own system and compare them with the platform using the same attribution window and the same date basis. If the gap is large and unexplained, the problem is measurement, not advertising.
The event you measure is not the event you sell
Plenty of accounts are optimised toward add-to-cart because the numbers look better and arrive faster. The platform does exactly what it was asked: it finds people who add to cart, some of whom never intended to buy.
Read the cost at the completed order, and make that the campaign objective once there is enough data for it. If weekly orders are very few, start with an event closer to the purchase — initiate checkout — rather than one far from it.
The page and the offer come before targeting
An ad sending interested visits to a slow page with unclear pricing reads as a failed ad in the dashboard, while the problem sits one step past it. Open the page on your own phone on mobile data rather than wifi, and count the seconds until the image and the price appear.
Three things drop the order most often: the price is not visible before checkout, the payment method the customer wants is missing, and there is no reassurance about delivery or returns in a market that buys cash on delivery.
Margin: can this campaign succeed at all?
Some campaigns do not fail; they are arithmetically impossible, because the margin cannot carry the cost of an order in that market. That is not a fault of the advertising or the advertiser, and diagnosing it takes numbers rather than opinion.
Work out contribution per order — after product cost, shipping, collection and the return rate — and you have the most you can pay. If your current cost sits far above that ceiling, the work starts at price, bundle or shipping cost, not at the campaign.
When is targeting genuinely the problem?
When measurement is sound, the event is right, the page is fast and clear, and the margin carries the cost — and the result is still weak. Now targeting and message are the question: a broad audience in a crowded market, or a message that never says why you specifically.
Change one thing at a time and give it enough results to judge. Changing three things together gives you an outcome whose cause you do not know, which is worse than changing nothing.
The order of diagnosis, step by step
Compare your numbers with the platform's
Fourteen days of orders from your system against the platform, same attribution window, same date basis.
Inspect the setup
No test event code left on, no duplicate between pixel and conversions API, and the purchase event genuinely firing on the thank-you page.
Change the event you read
Read cost at the completed order rather than add-to-cart, and make it the objective once its data is sufficient.
Test your own checkout
On a phone, on mobile data: load time, price clarity, payment methods, delivery and returns reassurance.
Work out your ceiling
Contribution per order gives you the maximum cost and the break-even point. Above the ceiling, start with pricing.
Only then: targeting and message
One element at a time, with enough results to judge before the next change.
Frequently asked
- How long before I judge a campaign?
- Judge by results, not by days: a campaign with three orders cannot be judged even after a month. Wait for enough results to separate a difference from luck, which in a small business means weeks rather than days.
- Should I pause the campaign or fix it?
- Pause it if cost per order sits above your margin ceiling by a gap no tweak can close. Otherwise change one element — the event, the page, the offer, then targeting — and re-measure after each.
- Is the cause that my audience is targeted wrongly?
- It can be, but it is the last thing to check rather than the first. Targeting is what people blame because it is the most visible part of the setup, while the silent faults in measurement and on the page cost more and get noticed less.