Home Free tools"Did the ads work?" checker

Plan and budgetFree · no sign-upInstant

Did the ads actually work?

A campaign worked if your results rose by more than their normal week-to-week ups and downs. Paste your weekly enquiries or sales from before and during the ads: the checker compares the two, allows for the normal week-to-week ups and downs, and says plainly whether the lift is clear, likely, or too early to call, and what each extra result cost.

The lift, in results and per cent
How sure you can be
What each extra result cost

1Fill in a few boxes
2See your result straight away
3Unlock the full report, free

Results per week before the ads

Enquiries or sales, one number per week, separated by commas or new lines. At least four.

Results per week during the ads

At least two weeks.

What the ads cost in all (optional)

The verdict

Clear lift

A clear lift: up 51.2% a week, well beyond the normal swing. About 43 extra results in all.

Before During

21.2Average a week before

32Average a week during

+51.2%Lift

In your full report

Cost per extra result
How far beyond the normal swing it is
How to test it properly

Rather have a person look at it? We will go through your numbers with you on a free 30-minute call.

Book a free call

How do you measure the real lift from an ad campaign?

Compare the same measure before and during the ads: enquiries, bookings or sales, counted the same way each week. Use your own records, such as your CRM, booking system or till, rather than the ad platform's figures.

Then compare the rise with how much results normally move. A business that swings between 17 and 23 enquiries a week on its own has proved nothing with one week of 22.

How sure can you be that the lift is real?

The checker measures the rise in normal swings: how far results moved compared with their usual week-to-week variation, allowing for how many weeks you have. The more weeks on each side, the smaller the rise it can trust.

A rise of about a tenth can be either noise or real. With the six weeks above, three weeks at 22 a week is too early to tell, while eight weeks averaging 22.3 is probably working.

How the checker reads a lift
Verdict	Rise, in normal swings	What to do
Clear lift	2.5 or more	Keep the ads running and test a larger budget.
Probably working	1.65 to 2.5	Run a few more weeks before changing anything.
Too early to tell	Above zero, below 1.65	Keep going, or run a holdout test.
No sign of a lift	Flat or down	Check the tracking, the offer and the page before spending more.

The thresholds the checker uses. 1.65 swings is roughly the usual line for 95% confidence that a rise is real.

What is a holdout test?

A holdout test switches the ads off for part of your market and compares it with the part that still sees them. You might pause ads in two of eight postcode areas for four weeks and compare enquiries in each.

Because both groups live through the same weeks, seasons, bank holidays and news affect them equally. That makes a holdout the strongest test a small business can run without special tools.

Pick areas or groups of similar size and history.
Keep everything else the same: prices, offers and opening hours.
Run it for at least four weeks.
Compare the gap between the groups, not the totals.

Why do ad platforms report more sales than the ads caused?

Google and Meta count a sale when someone who clicked an ad, or on Meta even just saw one, buys within a set window. Meta's default window is seven days after a click and one day after a view, so people who would have bought anyway are counted too.

That is not dishonest, but it is not lift. Platform figures show which ads people touched; a before-and-during comparison or a holdout shows what the ads added.

How the tool works

Paste a number per week from before the ads (at least four weeks).
Paste a number per week from while the ads ran.
Add what the ads cost, if you want the cost per extra result.
See the lift, how sure you can be, and what each extra result cost.

How Scale Oxford helps

Want to know what your ads really add?

We set up tracking that counts results in your own records, then design tests that separate the ads' effect from the season. Where a campaign adds little, we say so and move the money.

We match platform figures against your CRM or till every month.
We plan holdout tests by area or audience for campaigns big enough to measure.
We report the cost of each extra result, not the platform's claimed return.

Book a free call See our Meta ads service

Case study · BoConceptScale Oxford brought the cost of each online sale from Google Ads down by 43% for BoConcept. The account was counting some sales twice and missing others, so Google was bidding on the wrong numbers. We fixed the tracking before touching a single bid.↓ 43%Lower cost per sale↑ 81%Higher return on ad spend↑ 162%More revenue from Google AdsRead the case study →

Questions people ask

How do I know if my ads are working?

Compare the same measure (enquiries or sales) before and during the ads, over enough weeks to see the normal ups and downs. If the rise is much bigger than the usual swing, the ads most likely caused it.

Why not just compare two months?

One month against another mixes the ads with everything else: seasons, holidays, a busy week. Several weeks each side show how much results move on their own, which is what the checker measures the lift against.

What does "too early to tell" mean?

The rise is within the normal week-to-week swing, or there are too few weeks to say. Run longer, or test with a holdout: pause the ads in one area and compare.

Does this prove the ads caused it?

No simple test can prove it. A clear lift with nothing else changing is strong evidence; a holdout test is stronger.

How many weeks of data do I need to check an ad lift?

At least four weeks before the ads and two during them, which is the checker's minimum. Eight weeks on each side is better: the more weeks you have, the smaller the rise you can tell apart from normal ups and downs.

What is incrementality in advertising?

Incrementality means the results an ad caused that would not have happened without it. It is usually lower than what a platform reports, because some people who saw or clicked an ad would have bought anyway.

Should I paste enquiries, sales or revenue into the checker?

Paste the result closest to money that you record reliably every week. Enquiries suit service businesses; orders or revenue suit shops. Keep the same measure and the same source for both periods.

What if something else changed while the ads ran?

Then the check cannot separate the two. A price cut, a new product or a busy season during the ads all push results up too. Note what changed and, if you can, repeat the test in a quieter month or as a holdout.

Can I use the checker for SEO, email or a leaflet drop?

Yes, for anything with a clear start date: paste the weeks before it began and the weeks after. SEO builds slowly, so compare longer periods, such as eight weeks or more on each side.

What does cost per extra result mean?

It is the ad spend divided by the results above the usual level. £1,200 of ads that added 24 enquiries cost £50 for each extra enquiry, which is a fairer figure than spend divided by every enquiry in the period.

Why are my results down even though the ads are running?

Results can fall during ads for reasons the ads did not cause, such as a quiet season, a competitor's offer or a broken contact form. Check the same weeks last year and test your form before deciding the ads failed.