Convert.com Review 2026: Features, Pricing, Real Limits

Convert.com is the A/B testing tool that publishes its own price list, and that changes more than it sounds like it does. In a category where nearly every relevant competitor hides the number behind a contact form, being able to add up the cost before speaking to a salesperson is a practical planning advantage. This review covers what the tool does, what it actually charges, how the tested-user model interacts with the way you size a test, where the real limits are, and what kind of operation the math works for. For the whole category landscape, see our neutral comparison of CRO tools by category.

A disclosure before anything else: Donnu is an A/B testing tool and competes with Convert.com on part of this scope. This review was written to be useful even to a reader who picks Convert at the end, which is exactly why the section on where it is clearly the right call exists.

What Convert.com Is, One Line Per Piece

Piece	What it does	Who tends to care
A/B and split URL testing	Client-side experiments with a visual editor and code editing	Marketing, without waiting on a deploy
Multivariate testing	Combinations of elements on the same page	Teams with high traffic and many variables
Personalization	Content delivery to declared segments	Operations with a segmentable audience
Dual statistics engine	Frequentist or Bayesian read of the same experiment	Teams that already have a statistical position
Analytics integrations	Sends the experiment into the data tool the company already uses	Whoever decides outside the testing tool’s interface

The honest reading: Convert is an on-page experimentation tool, not a behavior suite with heatmaps, session recording, and user research bundled in. That is a deliberate scope, and it explains both the lower price against a full suite and the fact that it does not replace one.

Pricing: The Public Table, In Full

This is the tool’s main competitive differentiator and deserves to be reproduced without interpretation. Figures checked on the official pricing page on 13 August 2026:

Plan	Monthly	Annual	Tested users included	Overage	Projects / domains
Growth	$399/mo	$3,588/yr ($299/mo)	100K/mo (1.2M/yr)	$399 per 100K	5 projects, 10 domains
Pro	$599/mo	$5,040/yr ($420/mo)	100K/mo (1.2M/yr)	$699 per 250K	30 projects, unlimited domains
Enterprise	Quoted	Quoted, annual only	From 1M/mo	$699 per 250K	Unlimited

Prices exclude taxes, calculated at checkout based on billing location. The advertised free trial is 15 days, no credit card.

Two observations the published table makes explicit and almost nobody notices:

1. Growth and Pro include the same volume. What separates them is the feature set, the project and domain limits, and the overage price, not the allowance. If your problem is volume, moving up a plan does not solve it on its own.
2. Pro overage is proportionally cheaper. 699 dollars per 250,000 works out to roughly 2.80 dollars per thousand additional tested users, against 3.99 per thousand on Growth. A team that knows it will blow through the allowance regularly needs that calculation before the first invoice, not after.

Tested Users: The Unit That Decides Whether the Plan Fits

Convert bills by tested user: a unique visitor who enters an active experiment during the period. Not sessions, not pageviews, and not visitors to the whole site, only those actually bucketed into a test. That distinction is what ties the tool’s price to the statistical sizing of your experiment, and it is the calculation nobody runs before signing.

Set the calculator to your real baseline rate to see how many visitors your next test consumes:

Current conversion rate%Minimum detectable effectrelative (%)absolute (pp)Confidence909599%Power8090%Visitors per week (total)Testtwo-sidedone-sided

-Visitors per variation

-Total (2 variations)

-Estimated duration

Two-proportion normal approximation, 2 variations (50/50). Tweak the inputs and watch it update live.

How Much of the Plan a Single Test Really Eats

A store converts at 3% and receives 20,000 visits per week. To detect a 15% relative gain (3% to 3.45%), at 95% confidence and 80% power, the math asks for 24,193 visitors per variation, which takes 17 days. With two variations, the experiment consumes roughly 48,386 tested users, less than half the 100,000 monthly allowance.

Now change the ambition. Aiming at 10% relative (3% to 3.3%), the requirement rises to 53,211 per variation, which takes 38 days and consumes roughly 106,422 tested users: on its own, that single test overruns the monthly allowance, and overage lands on the invoice.

Test ambition	Sample per variation	Duration at 20K/week	Tested users consumed	Fits in 100K/month?
15% relative	24,193	17 days	~48,386	Yes, half to spare
10% relative	53,211	38 days	~106,422	No, overruns on its own

The practical conclusion applies to any tool billed by tested volume, not just Convert: your statistical ambition is a line item. Aiming at a smaller effect does not only mean waiting longer, it means paying more. That is not a reason to inflate your minimum detectable effect and fool yourself, it is a reason to pick tests whose expected effect is large enough to fit the budget, a subject covered in our guide to statistical significance in A/B testing.

Dual Engine: Frequentist and Bayesian In One Tool

The pricing page lists a frequentist and Bayesian statistics engine in the plans. Having both rulers available is a real advantage, and also a trap if the team does not decide upfront which one it uses.

The two answer different questions. The frequentist p-value answers “if there were no difference at all, what is the chance of observing a result at least as extreme as this one”. The Bayesian probability answers “given what I observed, what is the chance B is better than A”. They are not translations of each other, and on the same dataset they can produce noticeably different reads of confidence.

The Worked Example

The test above ran to 24,000 visitors per variation and closed with 720 orders in control (3.00%) against 806 in the variation (3.36%).

Control (A)

VisitorsConversions

Variation (B)

VisitorsConversions

Confidence90%95%99%

Control (A) · Rate-

Variation (B) · Rate-

Relative lift-

p-value-

95% CI of the difference-

Two-sided two-proportion z-test. "Not significant" almost always means not enough sample, not that the versions are equal.

Running those four numbers through this blog’s significance engine: z = 2.24, p-value ≈ 0.0253, with a 95% confidence interval on the difference between +0.04 and +0.67 percentage points, and an observed relative improvement of +11.9%. On the frequentist 5% ruler, the result is significant and the variation wins.

Look at the interval, because that is what carries the business decision. The true gain could be as small as 0.04 percentage points, roughly one eighth of the observed point estimate. A significant result with a wide interval licenses the claim “B beats A” and does not license the projection “we will gain 11.9% conversion”. That distinction is the difference between an experimentation program that survives a finance audit and one that does not.

The watch-out with a dual engine is behavioral, not mathematical: having both reads on screen creates the incentive to look at both and pick the one that already favors the preferred decision. The hygiene rule is to declare the ruler in the test design, alongside the primary metric and the sample size, and not change it afterwards. Readers who want to understand the Bayesian logic before choosing can start with our guide to Bayesian A/B testing and with expected loss as a stopping rule.

Real Strengths

• Public, computable pricing. You can build a full year’s budget without talking to anyone, which is rare in this category and worth a lot in a procurement process with internal approvals.
• Lean, coherent scope. The tool does on-page experimentation and does not try to be an entire behavior suite, which reduces configuration and contract complexity.
• Both statistical rulers available. For a team that already has a position on frequentist versus Bayesian, not having to accept the vendor’s choice is concrete freedom.
• 15 day free trial, no card. It lowers the barrier to technical evaluation, and no card means nobody forgets to cancel.
• Published overage price. Knowing in advance what exceeding the limit costs is different from discovering it at renewal.

Limitations and Watch-Outs

• A 100,000 tested-user allowance is modest for a large operation. As the example above shows, one ambitious test consumes more than that. Teams running several simultaneous experiments end up living in overage.
• Growth and Pro sharing an allowance is confusing. It is easy to buy Pro expecting more volume and receive only more features.
• It is not a behavior suite. Without heatmaps, session recording, and user research in the same product, an operation that depends on those will contract a second tool.
• Client-side focus. If your central case is server-side experimentation with feature flags in the deploy pipeline, the right category is different, covered in our feature flag tools comparison and in the client-side versus server-side testing guide.
• Dollar pricing. For a team billing in another currency, the real cost moves with the exchange rate and applicable taxes, and that needs budget headroom.

How to Evaluate It In a 15 Day Pilot

The classic mistake is trying to run an experiment to a verdict inside the free trial. Fifteen days rarely closes a sample. Use the period for the questions that do not depend on sample:

1. Installation and performance. Does the snippet load without delaying the page? Is there visible flicker on the variation?
2. Real traffic split. On day one, does the distribution match the configuration? A skewed split is an instrumentation defect, not bad luck.
3. The visual editor against your actual HTML. Can it change the elements you need to change, or does everything become code editing?
4. Analytics integration. Does the experiment show up in the tool where your team actually decides?
5. Tested-user consumption. Check how many tested users the pilot burned and project that onto your quarterly test plan.
6. Both statistical reads. Run an A/A test, with no real difference between sides, and see how each ruler reports the non-effect.

The A/A test in item 6 is the most revealing and the most skipped. It does not measure the tool against competitors, it measures your instrumentation against reality, and it is the cheapest way to find a collection problem before making a decision on top of it.

Final Read: Who Convert.com Is For

It fits a mid-market operation with traffic in the tens of thousands of tested users per month, that needs A/B and split URL testing on the web with a visual editor, values a predictable budget without a sales call, and has its own statistical position it wants to apply.

It fits less well for volumes far above the allowance, where overage dominates the bill and a volume-negotiated quote tends to win; for teams that need heatmaps and session recording bundled in; and for engineering teams whose central case is feature flags and server-side experimentation, where the open options covered in open-source A/B testing tools and in the GrowthBook vs Statsig comparison tend to fit better.

If your case is that second group and cost per volume is what weighed, Donnu is one of the lighter options in the category, with predictable pricing and no call required to learn the number, focused on web A/B testing with honest statistics. It does not replace Convert for multivariate testing or segment personalization, and saying otherwise would be dishonest. If your case is on-page A/B testing with a conservative statistical read, start a free trial and compare what actually matters to you.

---

Read also: CRO tools compared · A/B testing tools pricing compared · VWO review 2026 · Convert.com vs VWO · Leia em português

References

• Convert. Pricing. Plan prices, tested-user allowance, overage price, project and domain limits, and the 15 day free trial, checked on 13 August 2026. convert.com/pricing.
• Convert. Convert Experiences. Product page, with the scope of A/B, split URL, multivariate testing, and personalization. convert.com.
• Kohavi, R., Tang, D. and Xu, Y. Trustworthy Online Controlled Experiments: A Practical Guide to A/B Testing. Cambridge University Press, 2020. Chapters on tool selection, A/A testing, and interval reading. Companion material at experimentguide.com.
• Gelman, A. et al. Bayesian Data Analysis, 3rd edition. Reference for the difference between the Bayesian and frequentist read of the same experiment. sites.stat.columbia.edu/gelman/book.