How to plan a first test campaign
A first test campaign is not meant to make money. It is meant to answer a question cheaply and clearly. This guide shows how to plan one for a lead offer. By the end you will be able to write a one-page test plan with a single question, a tight scope, a spend cap, the numbers you will read and the decisions you will make before you spend anything.
Published 5 October 2026. How we write and check guides: editorial standards.
Already generating leads on your own forms? Sell your leads instead.
Why plan before you launch
On a rev share program you pay for traffic before you know which leads will sell, and unsold leads are not paid. Without a plan, a first test usually ends in one of two ways: you stop too early because the first day looked bad, or you keep spending because you are not sure what "bad" looks like. A written plan settles both questions before money is on the line.
This guide covers planning. For offers and program terms, start at the publisher program. For the platform build, use the guide for your source, such as how to run MVA traffic on Meta or how to run home improvement traffic on Google Ads.
Step 1: Write one question
A test should answer one question. Write it at the top of your plan in a single sentence. Good first questions are specific and can be answered with your reports:
- Can my Meta traffic produce car accident leads that sell in this group of states?
- Which of two state groups sells more of my roofing leads from Google search?
- Does a situation angle or a next-step angle produce more sold leads for bathroom remodeling?
Avoid questions like "will this work?" They have no clear answer, so the test never ends.
Step 2: Keep the scope tight
Every extra variable splits your spend and blurs the answer. For a first test, hold everything fixed except the one thing you are testing.
| Element | First test setting | Why |
|---|---|---|
| Offer | One offer only | Results stay comparable. See how to evaluate a lead gen offer to pick it. |
| Traffic source | One source you already know | You learn the offer, not the platform |
| States | A small group, split into two at most | Each state gets enough traffic to read |
| Ads or keywords | A few, each with its own sub ID | Enough to compare, not so many that each gets little traffic |
| Hours and placements | Keep platform defaults unless they are the question | Fewer moving parts |
| Lander | The Summit lander for the offer, unchanged | Consent is captured there; the lander is not a test variable |
Step 3: Set your spend cap
Decide the total you are willing to spend on this test before you launch, and write it in the plan. Treat it as the cost of the answer. This guide does not suggest an amount, because the right figure depends on your source, your offer and your own finances. Use these questions to set it:
- Can I spend this full amount and get nothing back without it affecting my business?
- Is it enough to give each ad set or state group a fair amount of traffic?
- Is the test long enough to cover different days of the week?
- Does it leave room for a second test if the first answer is "change something"?
Split the cap evenly across the parts you are comparing. If one part gets most of the spend, you are not comparing like with like.
Step 4: Decide what you will measure
Write down the measures before launch so you are not tempted to pick whichever one looks best afterwards. For lead offers, the order of importance is:
- Rev share earned compared with spend, by sub ID. This is the answer to most first-test questions.
- Sold leads, because only sold leads are paid.
- Returned leads, because returns are deducted under the return terms agreed during onboarding.
- Leads submitted, as a check that traffic reaches the form and the form works.
- Clicks and platform metrics, only to diagnose problems.
Set up your sub IDs so every measure can be split by the thing you are testing. See tracking links and sub IDs explained.
Step 5: Write decision rules in advance
Decide now what you will do with each possible result. Write rules in words, tied to your spend cap and your measures.
| If you see | Then |
|---|---|
| Ads rejected or links broken | Pause, fix and relaunch. This does not count against the test. |
| Spend reaches your checkpoint (a share of the cap you chose in advance) with no sold leads | Check targeting, ads and links. If all are correct, consider stopping and changing the question. |
| Leads arrive but most go unsold | Finish the test, then make the angle more specific in the next one. See how ping/post affects affiliate payouts. |
| One part clearly ahead on rev share vs spend | Finish the test, then move spend toward it and start a new test against it. |
| Cap reached, results close | The answer is "no clear difference". Test something with a bigger expected effect. |
Step 6: Launch and keep a log
- Before launch, click every ad link in preview and confirm sub IDs are filled.
- Launch everything at the same time so each part sees the same days and hours.
- Within the first day, confirm ads are approved and leads are arriving on the right tracking link.
- After that, do not edit the test. Edits reset learning on most platforms and mix two tests into one.
- Keep a dated log of every change, rejection or outage, so you can explain odd results later.
- Read results only at the review points you set in the plan.
Ad platforms review ads before and after they run. Meta explains its process on ads in review, and Google lists its rules in the Google Ads policies. Summit's traffic guidelines apply to every ad in the test.
Have a test plan and a traffic source ready? Apply with your offer, states and expected volume.
Apply as a publisherYour one-page test plan
Copy these headings into a document and fill each one before you launch:
- Question: one sentence.
- Offer and source: one of each.
- States: the group, and how it is split.
- Ads, keywords or pages: each with its sub ID value.
- Spend cap: the total, and how it is split.
- Measures: in order of importance.
- Review points: when you will read results.
- Decision rules: what you will do for each result.
- Next test: what you will test if the answer is yes, and if it is no.
A filled-in example
Here is a plan for a media buyer who knows Google search and wants to try window replacement offers.
- Question: Which of two state groups sells more of my window replacement leads from Google search?
- Offer and source: window replacement, Google search only.
- States: two groups of neighboring states with similar weather, one campaign per group.
- Keywords: the same three keyword groups in each campaign, each with its own sub ID value, plus {keyword} passed into a sub ID.
- Spend cap: a fixed total, split evenly between the two campaigns.
- Measures: rev share earned compared with spend, then sold leads, then returns.
- Review points: after the first day for approvals and links, then at the checkpoint, then at the cap.
- Decision rules: as in the table above.
- Next test: if one group wins, test a third state group against it. If neither sells, test a more specific keyword set.
Common mistakes
- Testing several things at once. You get a result but not an answer.
- Stopping after the first day. Early numbers move a lot. Use your review points.
- Judging on leads submitted. Unsold leads are not paid. Use sold leads and returns.
- No spend cap. Tests without a cap turn into campaigns without a plan.
- Editing mid-test. Write the change down for the next test instead.
Frequently asked questions
How much should I spend on a first test?
This guide does not set an amount. Pick a cap you can lose without harm to your business, large enough to give each part of the test a fair share of traffic. Write it down before you launch.
How long should a first test run?
Long enough to cover different days of the week and reach your spend cap or a decision rule. Set review points in advance and read results only at those points.
Should my first test include several offers?
No. One offer, one source and a small group of states give the clearest answer. Add offers in later tests.
What if the test produces leads but they do not sell?
Finish the test, then make the next test about the ad angle or the states. Read why leads get returned if returns are also high.
Related guides
Pick the right offer to test.
Set up the labels your test depends on.
Creative tests once your first test is done.
What to do when a test works.
Read the results of your test.
Sources
- About Ads In Review, Meta Business Help Center
- Google Ads policies, Google Advertising Policies Help
Run your first test on Summit offers
Every publisher is reviewed before going live. Email us your verticals, traffic sources, states and expected volume.
Apply as a publisherOr write to team@summitleads.ai. We reply by email.
Already generating leads on your own forms? Sell your leads instead.