A/B testing tools compared: what they cost, who builds the variant, and how they call a winner

Name: BusinessMCP
Brand: BusinessMCP

Six split-testing tools on verified 2026 pricing — from a free tier to $150k a year — and the one that proposes the test and writes the variant itself.

Try BusinessMCP free How we do automated split testing

Free plan available · No credit card

Price: VWO free–custom · Optimizely ~$40k+/yr · AB Tasty ~$15k+/yr · Convert $299/mo annual · Kameleoon $495/mo · BusinessMCP $199/mo Scale
Switching gets you: A published-price shortlist, plus the only tool on it that derives the test from your friction data and opens the PR for you.
Stay with A/B testing tools if: You want a visual editor, multivariate, personalization or server-side tests — every other tool here has at least one of those and BusinessMCP has none.

Feature comparison — pricing verified September 2026. Sources: public pricing pages + published teardowns.
	BusinessMCP	VWO	Optimizely	AB Tasty	Convert	Kameleoon
Entry price	$199/mo (Scale) — analytics, replay + CRM included	Free Starter; paid custom (~$314/mo+)	Custom (~$40k+/yr)	Custom (~$15k+/yr)	$299/mo annual	$495/mo (50k MTU)
Visual (no-code) editor	No	Yes	Yes	Yes	Yes	Yes
Finds the test for you from friction data	Yes — 8 failure modes, ranked by ICE	No	No	No	No	Partly — AI-generated test ideas (credits)
Who builds the variant	AI opens a GitHub PR you review	You	You	You	You	You (or prompt-based)
Flicker-free assignment	Yes — inline pre-paint bootstrap, CSS-only variants	Anti-flicker snippet	Snippet with anti-flicker	Snippet with anti-flicker	Snippet with anti-flicker	Snippet with anti-flicker
Peeking guard (minimum sample AND 14 days)	Yes — 300 visitors + 25 conversions per arm, 14 days	Bayesian; no minimum duration	Sequential Stats Engine	Bayesian; you decide	Frequentist or Bayesian; you decide	Sequential testing + SRM detection
Session replay + heatmaps included	Yes	Add-on (Insights)	No	Via partners	No	No
Requires GitHub	Yes — the PR review is the approval gate	No	No	No	No	No

The full picture

The A/B testing market splits into two prices: published and quoted. Convert Experiences ($299 a month annual) and Kameleoon ($495 a month) print a number. VWO prints a free Starter tier and then names its paid tiers without pricing them; Optimizely and AB Tasty — now merged with VWO — quote every deal, and published teardowns put those quotes at $15,000 to $150,000+ a year. All five have a visual editor. None of them tells you what to test.

BusinessMCP is on this list for a different reason. Its CRO engine reads your heatmaps and session replays for eight named failure modes, ranks the candidates by ICE, proposes a test with a hypothesis and the evidence behind it, and the AI opens a GitHub pull request implementing variant B. It has no visual editor, needs a repository, and only split-tests web pages — and it is $199 a month on the Scale plan with analytics, replay, the CRM and the AI business analyst included. Below, every price is what the vendor’s page showed on September 15, 2026, or a cited teardown where the vendor shows none.

The tools, honestly assessed

Published vs quoted

Convert and Kameleoon publish a price; VWO publishes a free tier and hides the rest; Optimizely and AB Tasty publish nothing. If you cannot get a quote without a sales cycle, the tool is priced for a team, not a site. BusinessMCP is $199 a month on the pricing page.

Who has to have the idea

Every editor-based tool assumes you arrive with a hypothesis. BusinessMCP computes one from dead clicks, rage clicks, scroll depth, form starts and bounce-plus-dwell, and shows you the rows it used. That is the feature that decides whether a small team runs zero tests or a test a month.

Who builds the variant

A visual editor is faster for a copy change and fragile on a modern front end that re-renders the DOM. A pull request is slower for the copy change and reliable for everything. BusinessMCP only does the second; the other five only do the first (plus code, if you write it).

How a winner gets called

Optimizely’s sequential Stats Engine and Kameleoon’s sequential testing let you stop early on purpose; VWO and AB Tasty report Bayesian probabilities and leave the stopping to you; Convert offers frequentist and Bayesian. BusinessMCP is the strictest and the simplest: nothing is shown until 300 visitors and 25 conversions per arm and 14 days.

Real data, not a mockup

What it looks like in BusinessMCP

Recorded on our own site, PII scrubbed — the same product you get on the free plan.

app.businessmcp.com/replays

A/B testing tool pricing, verified September 2026

Plan	Price	Notes
VWO	Starter free (50k MTU); Growth/Pro/Enterprise custom	teardowns: Growth ~$314/mo annual, Pro ~$531–$972, Enterprise $1,265+; Vendr median $16,660/yr
Optimizely	Custom	“individually packaged”; teardowns ~$40k–$80k/yr under 250k MTU, $150k+ above 1M; Vendr median $81,447/yr
AB Tasty	Custom	“No fixed plans”; Vendr ranges $15k–$45k (50k–250k sessions) to $80k–$150k+ (full platform); median $66,500/yr
Convert Experiences	Growth $299/mo annual ($399 monthly) · Pro $420/mo annual ($599 monthly) · Enterprise on request	100k tested users/mo on Growth and Pro; 15-day trial, no card
Kameleoon	Starter from $495/mo · Enterprise custom	50,000 tested visitors/mo, up to 10 experiments, 30/100/200 credits; 30-day trial, 3 experiments
BusinessMCP	$199/mo (Scale)	10k monthly visitors, 30-day replay, CRM, AI SDR and the AI business analyst included; month to month

Vendor pages were checked on September 15, 2026. Where a vendor shows no price, the figure is from the cited teardown or Vendr transaction data and is indicative. VWO and AB Tasty announced a merger on January 20, 2026; expect their packaging to change.

Where BusinessMCP is NOT the right choice

An alternatives page you can trust has to cut both ways. Honestly:

There is no visual or WYSIWYG editor. VWO, Optimizely, AB Tasty, Convert and Kameleoon all let a marketer click a headline and retype it; in BusinessMCP every variant is code in a pull request, so someone on your side has to be comfortable reviewing a diff.
It needs a codebase on GitHub and a human to review and merge the PR. A site on a hosted builder with no repository (a Webflow, Wix or Squarespace site, a Shopify theme nobody keeps in git) cannot run our split tests at all.
We only split-test web pages. There is no multivariate testing, no personalization engine, no server-side SDKs for mobile or backend experiments, and no feature flags — if you need any of those, the tools on this page are the right category and we are not.
A/B testing sits on the Scale plan at $199/mo. VWO’s free Starter tier, Kameleoon’s 30-day trial and Convert’s 15-day trial all cost less to start, and the $199 buys you the whole platform whether or not you use the rest of it.
Our stats module is deliberately simple: a two-proportion z-test behind a fixed minimum-sample-and-14-day guard. Optimizely’s sequential Stats Engine and the Bayesian engines in VWO and AB Tasty let a statistically literate team stop early on purpose; ours will not let you.

Frequently asked questions

Which A/B testing tool is cheapest?

VWO’s free Starter tier (50,000 monthly tested visitors as of September 2026) if it survives the VWO–AB Tasty merger; otherwise Convert Experiences at $299 a month annual is the lowest published price with a visual editor. BusinessMCP’s Scale plan is $199 a month but includes the whole platform, not a testing tool alone.

Which tools publish their pricing?

Convert (Growth $299/mo annual, Pro $420/mo annual) and Kameleoon (Starter from $495/mo) publish prices. VWO publishes a free tier and names its paid tiers without prices. Optimizely and AB Tasty quote every deal. BusinessMCP publishes every plan.

Which tool finds the test for you?

BusinessMCP derives test proposals deterministically from your heatmap and replay data — eight named failure modes ranked by ICE. Kameleoon offers AI-generated test ideas through prompt-based experimentation credits. The others give you an editor and expect a hypothesis.

Do any of these avoid flicker?

All of them address it: the editor-based tools ship an anti-flicker snippet that hides the page until the variant is applied. BusinessMCP assigns the arm in an inline bootstrap before first paint and applies the variant as CSS keyed on that arm, so nothing is hidden and nothing re-renders.

How does each tool call a winner?

Optimizely and Kameleoon use sequential methods that stay valid if you stop early. VWO and AB Tasty report Bayesian probabilities and leave the stopping rule to you; Convert offers both. BusinessMCP hides results until 300 visitors and 25 conversions per arm and 14 days, then reports a two-proportion z-test with lift intervals.

Other honest alternatives

VWO

Free Starter tier, visual editor, quoted paid tiers — the default self-serve pick.

Compare

Optimizely

Enterprise experimentation with flags and server-side SDKs, from roughly $40k/yr.

Compare

Google Optimize (discontinued)

Gone since September 30, 2023 — what to use instead.

Compare

Sources & verification

Every price on this page traces to a public source, last verified September 15, 2026. Spotted a change? Tell us and we'll re-verify.

Keep reading

Automated split testing, explained Session replay & heatmaps (where the friction comes from)Website personalization for B2B Marketing attribution models BusinessMCP pricing

Companies running MCP servers with us

Makers and companies publishing a server in the BusinessMCP directory.

34M+ companies in our database 165 tools on one MCP endpoint Custom MCP servers over your systems Cookieless & GDPR-first Security code published (MIT)

Browse the directory

Test it on your own traffic — free

Install one cookieless script next to whatever you run today and compare on your own numbers. Free plan, no credit card, cancel anytime — the opposite of an annual contract.

Get started free

Free plan available · No credit card