Plan small paid ad tests before you spend

A good test answers one question, such as "Can Google search ads bring us demo requests for less than $150 each?" Narrowing the test to a single channel, audience, and offer keeps the budget focused and makes the result easy to read.

Small budgets get spread too thin when a test tries to answer everything at once. If you run three platforms, four audiences, and six ads on $2,000, each combination gets a few dollars and a handful of clicks. None of them collects enough results to tell you anything. A test with one question puts all of that money behind a single answer.

Kite helps you pick the question worth asking first. If you're not sure which platform to test, it compares how your buyers search, who they are, and where your competitors keep paying to advertise, then suggests the one with the best evidence. If you already know the platform, it helps you choose the audience and the offer. For example, a payroll software company for restaurants might test Google search ads for owners searching "restaurant payroll software," sending clicks to a page that offers a free payroll setup call. That's one channel, one audience, and one offer.

Kite also checks what your competitors are running before you spend. Its built-in ad research shows a competitor's current ads on Google, Meta, and LinkedIn, along with their landing pages and how long each ad has been running. An ad that has run for months is usually earning its keep, which tells you which angles the market already responds to.

Related use cases

• Choose which ad platforms to spend on: Decide which platform deserves the test, based on how your buyers look for what you sell and what a customer would cost on each one.
• Find the search keywords your competitors pay for: See which Google searches your competitors bid on, how often people search them, and what each click costs.

Spend enough to get roughly 10 to 15 results at the cost you're willing to pay, and no less. A smaller budget can't tell a good channel from a bad one, so the money is spent without buying an answer.

The math starts with your target cost per result, which is the most you'd pay for one lead, demo request, or sale. Kite works it out from what it knows about your pricing and how many leads usually become customers. Say a customer is worth $3,000 to you over the first year, and about one in five demo requests becomes a customer. You could pay up to $600 for a demo request and break even, so a target of $150 leaves plenty of room for profit.

Next, Kite estimates what each result will cost on the platform you're testing. For Google, its built-in search data gives the typical cost per click for each keyword. For LinkedIn, when you connect your LinkedIn Ads account, Kite asks LinkedIn what your exact audience costs, and LinkedIn returns a suggested bid. If your Google Ads or Meta Ads accounts are connected, Kite reads what you actually pay there.

From there, Kite shows the math. Suppose clicks on your keywords cost about $5, and your landing page turns about 3 percent of visitors into demo requests. Each demo request would cost about $167. A test budget of $2,000 over four weeks buys about 400 clicks, which would bring in about 12 demo requests. That's enough to see whether the real cost lands near your target. If your budget is only $600, Kite tells you that it would buy about 4 demo requests, which isn't enough to judge, and suggests ways to make the test fit, such as narrowing the keyword list or measuring an earlier step first.

When an estimate is a guess, such as the conversion rate of a landing page you haven't built yet, Kite says so and uses a cautious figure. That way the plan doesn't assume the test will go well.

Success is a cost per result at or below the target you agreed on, measured on the action that matters to your business. Kite helps you choose that action and agrees on the target number before the test starts.

The action you measure should be as close to revenue as your budget allows. A demo request or a purchase tells you much more than a click. But a small test may not produce enough purchases to judge, especially for an expensive product. In that case, Kite suggests measuring a step that happens more often, such as a signup or a demo request, and checks it against how often that step leads to a sale.

The ad platforms need enough results to learn, too. Meta says an ad set needs about 50 results in a week to finish its learning phase, when it figures out who to show your ads to. A $1,500 test optimized for purchases may never get there. Kite points this out and recommends setting the campaign up to optimize for an action that happens often enough for the platform to learn.

Kite also sets a few early signs to watch, such as click-through rate and cost per click. These won't tell you whether the test succeeded. But they do warn you early when something is broken, such as an ad nobody clicks or a keyword that costs three times the estimate.

You stop or scale when the test hits a rule you agreed on before launch. Writing those rules down in advance is the single most useful part of the plan, because it's very hard to judge a test fairly once you've watched it spend money.

Kite writes the rules into the test plan in plain sentences. For the payroll software test above, they might read like this:

• Check early for broken pieces. If the test spends $450, three times the target cost per demo request, without a single request, pause it and check the tracking and the landing page before spending more.
• Scale when it beats the target. If the test ends with at least 10 demo requests at $150 or less each, it earns more budget. Raise the budget about 20 percent at a time, a few days apart, so the platform can adjust without starting over.
• Revise once when it's close. If the cost per demo request lands between $150 and $225, the channel may work with better ads or a better page. Run one more round with a single change, and judge it by the same rules.
• Stop when it misses. If the test spends its full budget and the cost per demo request is above $225, end it and move the money to the next test on your list.

Each rule names a number and the action it triggers, so anyone on your team can read the result and know what happens next. Kite records the plan and the rules in your team's knowledge base. When the test ends, the decision was already made. It's just waiting for the numbers.

Related use cases

• Find which products and ads to scale: Decide which winning ads and products can take more budget, using profit after ad costs.
• Cut wasted ad spend: Find the ad sets that spend without results once a test grows into a full campaign.

Once you approve the plan, Kite builds the ads and the page they send people to. A test is only as good as the ad and the landing page, so Kite builds both to match the question the test is asking.

Kite writes the ad copy from what your customers actually say and checks every line against each platform's limits. It writes Google responsive search ads and Facebook and Instagram ads, and it makes the images and short vertical videos for Meta placements. For a single test, it usually writes a small set of ads around one angle, so the result tells you about the channel, not just which headline won.

Every test needs a landing page that keeps the ad's promise. If the ad offers a free payroll setup call, the page should offer that same call near the top, not bury it under a general homepage. Kite builds a landing page for the test's audience and publishes it on a site it hosts or through a connected content system such as Webflow, once you approve it. On sites Kite hosts, its built-in analytics track form submissions and calls to action from the start, so the test's results are measured from day one.

What Kite can set up inside each platform depends on the connection:

• LinkedIn Ads. Kite builds the test as a draft Sponsored Content campaign in your connected account. It sets the objective, the audience, the daily budget, and the bid, along with the end date and a total budget cap, so the test can't spend past its limit. It writes and designs the image or carousel ads. A team owner or admin launches the draft from Kite after confirming the budget.
• Google Ads and Meta Ads. Kite writes the ads, makes the creative, and gives your team the exact campaign settings from the plan, including the daily budget, the end date, and the action to optimize for. Your team sets up the campaign in Google Ads or Meta Ads Manager.

Related use cases

• Produce paid-ad creative: Get the copy, images, and video for each platform, built from the angles that work.
• Build landing pages for campaigns Coming soon Get a page for each campaign that matches the ad's audience and offer.

While the test runs, Kite checks it against the early warning rules. When it ends, Kite reads the final numbers against the stop and scale rules and recommends the action they call for.

With your ad accounts connected, Kite reads spend, clicks, and conversions for the test campaign. Google Analytics adds what happened after the click, so Kite can confirm that the conversions the ad platform reports match the ones your website recorded. A test can look like a failure because a conversion tag isn't firing on the thank-you page. Kite checks for that before calling it.

You can put the check on a schedule. Every few days, Kite reads the test's numbers and tells you whether anything has hit an early warning rule. On LinkedIn, if the test spends past the early checkpoint with no results, Kite can pause it on its own, because pausing only stops spending. For Google and Meta, it names the exact campaign to pause, and your team pauses it.

At the end, Kite writes a short result: what the test spent, what it brought in, the cost per result next to the target, and the rule that applies. If the test earned more budget, Kite proposes the first increase. On LinkedIn, it sends the budget change for a team member to approve, then makes it. If the test missed, Kite records what it learned and suggests the next question to test. Each result goes into your team's knowledge base, so later tests start from what earlier ones taught.

Related use cases

• Measure whether your ads actually drive sales Coming soon Find out whether a channel that passed its test brings in sales you wouldn't have gotten anyway.