Pricing Page A/B Test Ideas, Ranked by Risk and Traffic

Most pricing page A/B test ideas are ranked by how exciting they sound. Test a higher price, test a new tier, test removing the free plan. Those are the riskiest tests you can run, and on a small team’s traffic they rarely produce a readable result.

Rank them the other way: by what a wrong answer costs you and by how much traffic it takes to get any answer at all. Every idea below is shown on a live pricing page from our directory, so you can see what the change looks like before you build it.

Do the traffic math before any pricing page A/B test

A pricing page converts a small share of visitors, so the effects you are hunting are small. Take a page that turns 2% of visitors into signups and a variant you hope lifts that to 2.5%. A standard two-sided test at 95% confidence and 80% power needs roughly 13,800 visitors per variant to detect that difference. Two variants, about 27,600 visitors.

If your pricing page sees a few thousand visitors a month, that test runs for most of a year, and the market moves under it. So the practical rules for a small team:

• Test changes big enough to move the number a lot, or do not test at all.
• Measure the step closest to the change (plan clicks, trial starts), then confirm on revenue later.
• Treat anything that touches what an existing customer pays as a rollout, not a test.

That third rule is what sorts the list.

The ranking: seven pricing page A/B test ideas, safest first

1. The call to action on each tier

Lowest risk, fastest signal, fully reversible. The words on the button decide what the visitor thinks happens next.

Loops is a clean example as of our directory’s last fetch. The featured Growth plan is paid, shown at $49 a month for up to 5,000 contacts, and its button still says “Start for free.” The free plan (1,000 contacts, 4,000 sends per 30 days) uses the same words. That is a deliberate bridge: the visitor who picks the paid plan is told the first step costs nothing. Enterprise says “Contact us.”

What to test: the paid tier’s button copy (“Start for free” against “Start trial” against “Buy Growth”). Measure clicks per tier and the share of clicks that finish signup.

2. Which tier is featured

The highlighted plan pulls clicks toward it. It is cheap to change and it tells you which plan the page is really selling.

Featuring the second tier is the common move. Shipday’s pricing features Business AI Lite, the middle of three Merchant plans; Loops features Growth; Baremetrics features Growth, the middle of its three tiers. What to test: featuring the tier above your current pick. If more buyers choose it without a drop in total signups, you were under-anchoring. Our post on how many pricing tiers explains why the middle tier carries the page.

3. How the annual discount is framed

Same discount, different words, different take-up. Compare two live framings:

• Tally says “2 months off with annual billing” and shows Pro at $24 a month on yearly billing.
• Baremetrics says “Annual plans can come with up to 35% off the listed price.”

“Two months off” is concrete and small. “Up to 35%” is larger but hedged by “up to,” which tells a careful buyer to ask. What to test: the framing, and separately the toggle’s default position. Measure the annual share of new paid plans, not just signups. The discount strategy teardown has more on where discounts belong on the page.

4. Where the value-metric slider starts

Usage-priced products often show a slider, and its starting point sets the first number the visitor sees. Baremetrics prices by tracked revenue, “never per seat,” and its displayed prices change as the slider moves. Loops prices Growth on subscribed contacts, and the page states that unsubscribed contacts never count toward price.

What to test: the default slider position, and whether the page states the unit in words beside the price. A slider that opens at a large customer’s volume scares small buyers; one that opens at the floor undersells the product to large ones. This is medium risk because it changes perceived price without changing actual price.

5. Free plan or free trial as the entry offer

Now the risk climbs, because the change reaches past the pricing page into activation, support load, and revenue timing. In our directory’s latest stats, 45.0% of products offer a free plan and 52.3% offer a free trial, so neither is the default. Freemium vs free trial covers how to choose.

What to test, if you test at all: only with a measurement window long enough to see paid conversion, which is weeks after the click. Clicks on a “Free forever” button are not the result.

6. What moves between tiers

Packaging tests change what a plan includes. They are high risk because existing customers can see the page too, and a feature moving up a tier reads as a takeaway.

Shipday shows what a packaging change looks like when it ships. Between two captures, the latest on 2026-10-19, Business AI Lite gained “Manage deliveries with AI Operations Manager,” and Business AI went from 1,000 included orders to “Unlimited orders per location.” Branded Elite stayed at $99 with 300 included orders.

If you test packaging, test it on new visitors only and grandfather anyone who already pays.

7. The price itself

Riskiest, and usually the wrong tool. Two visitors from the same company seeing different prices for the same plan is a trust problem the moment they compare notes, and a support problem the day the test ends.

The same Shipday change shows the alternative. The price moved with the package: Business AI Lite went from $159 to $199 a month and Business AI from $349 to $399, while the entry tier held at $99. That is a rollout, not a test. It also shows why a price test rarely isolates price: when the package changes in the same release, nobody can say which change moved the number.

For price, run a sequential change on new signups, keep a dated record of the before and after, and compare cohorts. Directory context helps you pick the range: the median starting price in our directory’s latest stats is $25 a month, and SaaS pricing benchmarks breaks that down by category.

Your competitors are running tests you can read

You will never see a rival’s test results. You can see what they shipped and whether it stuck. A pricing page change that is still live 90 days later probably did not hurt them; a change reversed within weeks probably did.

One warning from running a pricing tracker. A page read at a different slider position or billing toggle can look like a reprice when nothing changed. Before you copy a rival’s “test,” confirm the before and after came from the same view of the page. The free Competitor Website Change Checker compares a pricing page with a public archived copy from 30 to 90 days ago, which is enough to spot a real change. For a structural read of any pricing page, use the pricing page analyzer.

The one move worth making this week

Open your pricing page analytics and write down last month’s visitors. Divide 27,600 by that number: that is how many months a test for a half-point lift would take. If the answer is more than two, skip price and packaging entirely and run idea one, the call to action on your featured tier, with a bold enough change to move clicks by a third or more. Ship it on Monday, read it in four weeks, and keep the before and after screenshots with the date.

Outmano is an AI-powered competitive intelligence platform that tracks rival pricing pages alongside SEO, content, roadmap, and reviews, with AI analysis on every change, delivered through a live dashboard, alerts, a weekly digest, and MCP access for your own AI. See how it works.

Frequently Asked Questions

What should I A/B test on a pricing page first?

Start with the call to action on your featured tier, then which tier is featured. Both are cheap, reversible, and measured right on the page by plan clicks. Leave price and packaging for later, because they reach existing customers and need far more traffic to read.

How much traffic do you need for a pricing page A/B test?

More than most small SaaS teams have. Detecting a lift from a 2% to a 2.5% conversion rate at 95% confidence and 80% power takes roughly 13,800 visitors per variant. If your page gets a few thousand visitors a month, test only large changes or run sequential before-and-after comparisons instead.

Is it legal or ethical to show different prices to different visitors?

Rules vary by country, so check with counsel, but the practical problem comes first: buyers compare notes, and two prices for the same plan look like a bait and switch. Most SaaS teams change price for new signups on a date and compare cohorts rather than splitting live traffic.

How long should a pricing page experiment run?

At least two full weekly cycles so weekday and weekend traffic both count, and long enough to reach the sample size you calculated before starting. For changes that affect the entry offer, extend the window until you can see paid conversion, which often arrives weeks after the first click.

Can I learn from competitors’ pricing page changes?

Yes, with care. A change that stays live for months probably did not hurt them, and a quick reversal is a signal too. Confirm each before and after came from the same page view (same toggle, same slider position) before you treat it as evidence.