Publisher Guide: Lead Quality

How to test ad creative for lead offers

This guide shows you how to run creative tests that tell you which ads produce leads that sell, not just which ads get clicks. By the end you will be able to write a test plan, set up ads so each one is tracked by its own sub ID, decide when a test has enough data, and keep a test log that turns one-off wins into a steady pipeline of new ads.

Published 5 October 2026. How we write and check guides: editorial standards.

Apply as a publisher

Already generating leads on your own forms? Sell your leads instead.

Why lead offers need a different kind of test

Ad platforms judge creative on what they can see: clicks, cost per click and form fills. For lead offers, those numbers stop short of what pays you. You are paid rev share on sold leads, unsold leads are not paid, and returns are deducted under the return terms agreed during onboarding. See how payouts work. An ad with cheap clicks can lose to an ad with expensive clicks if the second ad's leads sell and the first ad's leads do not.

So every creative test has two scoreboards: the ad platform for delivery and cost, and your Summit report for sold, unsold and returned leads. Leads are reported by tracking link, so the only way to join the two is a sub ID on every ad.

Step 1: Write a test plan before you build ads

A test plan is a few lines that keep the test honest. Write it before you open the ad platform.

Fields in a creative test plan
Field	What to write	Example in words
Question	The one thing you want to learn	Does naming the time frame bring leads that sell better than a general hook?
Variable	The single element that changes between ads	The headline only
Held constant	Everything else	Same image, body copy, audience, states, placements and lander
Success measure	The result that decides the winner	Sold leads and returns per ad, then cost per sold lead from your own spend
Minimum data	When you will judge	When each ad has produced enough leads that one more sale would not change the order
Next step	What you do with each outcome	If the time-frame hook wins, test three new time-frame hooks next

Step 2: Choose what to test, in order

Not every element has the same effect. Test the parts that change who clicks before the parts that change how many click.

The angle. The reason the ad gives someone to click: the situation, the next step, the cost question, the time pressure. Angle decides who arrives, so it has the largest effect on sold leads.
The hook or headline. Different wording of the same angle. Test once you know which angle works.
The visual. Image vs video, a person speaking vs a still scene, a phone showing a short form. Keep visuals plain and policy-safe.
The format. Single image, short video or carousel, where the platform offers them.
The call to action. The button and last line. Small effect in most tests, so leave it until last.

For angle ideas by category, see the offer pages, for example car accident offers and bathroom remodeling offers. For platform detail, read Meta ad creative for accident offers and Meta ad creative for home improvement offers.

First tests to run in each vertical

If you are starting from nothing, these angle pairs make good first tests. Each pair changes one idea, so the result tells you something about your audience.

Starter angle tests for MVA and home improvement offers
Vertical	Ad A	Ad B	What the result tells you
MVA	Names the accident and that it was recent	Names the accident with no time frame	Whether a time frame filters in readers whose leads sell
MVA	Focuses on the next step: a short form, then a call	Focuses on dealing with an insurance adjuster	Which worry moves your audience to act
Home improvement	Names the service and home ownership	Names the service only	Whether stating ownership filters in readers whose leads sell
Home improvement	Leads with a problem: a leak, an old tub, drafty windows	Leads with the result: a new roof, bathroom or windows	Whether your audience responds to the problem or the outcome

Keep the wording factual in both ads. Neither ad should promise a settlement, a price, a discount or a timeline you cannot support.

Step 3: Build the test so results can be read

Put all test ads in the same ad set or ad group, so they share an audience, states and placements.
Give each ad its own sub ID value. A readable code helps, for example "ang-time-v1" and "ang-next-v1".
Point every ad to the same offer and lander through your tracking link.
Keep the number of ads small. Two to four ads per test lets each one collect leads in a reasonable time.
Record the test in your log with the start date, the plan and the sub IDs.

Some platforms have their own split test tools. Meta's help page about A/B testing describes its built-in option, which splits the audience so the same person does not see both versions. You can use it for delivery, but still judge the result on your Summit report by sub ID. On Google, check the ad rotation setting so one ad is not shown far more than the others while you test.

Have creative ready to test on Summit offers? Apply with your sources and states.

Apply as a publisher

Step 4: Decide when a test is done

Most creative tests are judged too early. Early clicks are noisy, and sold and returned outcomes arrive after the click. Use these rules:

Each ad has produced enough leads that one more sold lead would not change which ad is ahead.
The test has run across full days of the week, not just one or two days.
Enough time has passed for returns on the early leads to show up.
Nothing else changed during the test: budget, audience, states, placements or lander.
The winner leads on sold leads and returns, not just on click cost.

If no ad is clearly ahead once these are met, the variable you tested does not matter much for this offer. That is a useful answer. Record it and move to the next element on your list.

Step 5: Keep a test log and a creative pipeline

A test log is a sheet with one row per test: date, question, variable, sub IDs, result and next step. Over time it shows which angles work for each offer, state group and source, and it stops you from testing the same idea twice.

Ads wear out as the same people see them again. Plan to have a new test running while your current winner is still strong, so you have a replacement ready when it fades. A common rhythm is one angle test, then several hook tests on the winning angle, then a visual test, then back to a new angle.

Keep every ad inside the rules. Follow our traffic guidelines and the platform's own policies, such as Meta's Advertising Standards. If you are unsure about a claim, send us the creative question before you run it. See ad claims to avoid in accident offers and ad claims to avoid in home improvement offers.

Common mistakes

Picking the winner on click cost. Cheap clicks that do not sell cost more in the end.
Testing several variables in one ad. If the headline and the image both change, you cannot tell which one mattered.
One sub ID for the whole ad set. You lose the link between each ad and its sold leads.
Stopping at the first good day. Short runs reward luck. Wait for enough leads and full weeks.
Letting the platform pick a favorite too early. Automatic optimization can starve an ad of delivery before it has a fair test.
Testing bold claims to see what happens. Claims about outcomes or prices risk rejection, account problems and leads that get returned.

Frequently asked questions

How many ads should I test at once?

Two to four ads per test is a practical range. More ads split your budget so thin that each one takes a long time to produce enough leads to judge.

Should I judge creative on form fills?

Use form fills as an early signal only. You are paid rev share on sold leads, so the final decision should rest on sold leads and returns by sub ID.

Do I need to send my creative to Summit before it runs?

Follow our traffic guidelines. If you have a question about a specific ad or claim, email us before you run it.

How often should I add new creative?

Keep a new test running while your current winner is still performing, so a replacement is ready when results fade. Your test log shows how fast ads wear out for your traffic.