Optimizely publishes no price for any of its products in 2026. Its pricing page lists ten product lines and not one dollar figure. The only guidance it gives you is that a salesperson will assemble a plan for you.
That is not an oversight, and Optimizely is not alone. Between January 2025 and May 2026 the experimentation category consolidated hard, and almost every vendor that changed hands also stopped publishing a price. If you are building a 2026 shortlist from a 2025 comparison post, three of the names on it no longer mean what they used to.
The short answer
If you want a published, self-serve price for experimentation, the 2026 shortlist is genuinely short: GrowthBook and Convert Experiences are the widely used options that still print a number you can budget from without booking a call. Optimizely, VWO and AB Tasty all require a conversation first.
If what you actually need is not another testing engine but an answer to why the losing variant lost, that is a different category of tool. Koji runs AI-moderated voice interviews with the people who used your product, then does the thematic analysis automatically. Koji does not replace Optimizely. It answers the question an A/B test structurally cannot.
How Optimizely pricing actually works
It does not work, in the sense that you cannot observe it. On optimizely.com/pricing the products listed include Agentic CMS, Agentic Experimentation, Agent Platform, Content Marketing Platform, Analytics, Personalization, Commerce, Asset Management, Data Platform and Feature Management. Against those ten lines there are zero currency symbols.
The page's own explanation is a single sentence: Every Optimizely plan is individually packaged. Tell us a bit about your digital needs, and we'll create a plan together.
Third-party trackers try to fill the gap and the results are thin. CostBench reports that the median Optimizely customer pays $50,000 per year, but that figure rests on 3 verified purchases and the page labels it Thin data itself. Three data points is not a budget. Treat it as an order of magnitude and nothing more, and note that a second range printed on the same page is inconsistent with the annual median it sits beside, which is a good reason to distrust any single scraped number for this vendor.
The honest summary for a buyer: Optimizely is an enterprise sale with a five-figure floor, and the only way to learn your price is to become a lead.
The signature finding: three names on your shortlist changed owners in 16 months
This is the part no comparison post covers, and it matters more than any feature table.
- January 2025. Everstone Capital acquires a majority stake in Wingify, the parent company of VWO, in a deal reported at around $200 million.
- 2 September 2025. OpenAI acquires Statsig for $1.1 billion. Statsig's founder Vijaye Raji becomes CTO of Applications at OpenAI. Statsig's own announcement, written by Raji, promises:
Statsig will continue to provide our services and invest in our core products.andOur customers will remain a top priority. - 20 January 2026. VWO and AB Tasty merge under Everstone. Wingify's own announcement describes a combined platform with
more than $100 million in annual recurring revenue and over 4,000 customers worldwide, servedfrom 11 offices across North America, Europe, Asia Pacific, and Latin America, with VWO co-founder Sparsh Gupta as CEO. - 5 May 2026. Amplitude takes over Statsig. Amplitude's own post states plainly:
Amplitude will take on Statsig's brand and customers.The engineering team does not come with it. As MarTech put it, that leaves Amplitudemanaging the platform, roadmap, and support for a product whose creators now work somewhere else.
Optimizely's CEO Alex Atzberger was blunt about the last one: Amplitude is getting Statsig's code without the talent, it is a race car without a driver, and should be very worrisome for existing Statsig customers as innovation slows down and support goes away. He added that Amplitude now has two duplicative experimentation and analytics capabilities which means more uncertainty for customers of either, as one of them will be shut down over time.
Two consequences for your shortlist, and they are concrete:
- AB Tasty is not an alternative to VWO. They are the same company. Both abtasty.com and abtasty.com/pricing now return an HTTP 301 redirect to wingify.com. Any 2026 listicle that recommends AB Tasty as a way to escape VWO is recommending the same vendor twice.
- Statsig is not an independent alternative either. The brand and the customer base belong to Amplitude; the people who built it work at OpenAI.
There is a tension worth noting honestly. Wingify's announcement says the combined platform will continue to support all existing VWO and AB Tasty customers without disruption and that Both brands will maintain product momentum. That may well be true of the products. It is not true of the web properties, because one of the two domains now redirects to the parent. Weigh the promise against the redirect yourself.
The 8 best Optimizely alternatives in 2026
1. Koji, best for finding out why a variant lost. An experiment tells you which version won. It cannot tell you what confused the people who bounced. Koji runs AI-moderated voice interviews at any hour, transcribes and codes them, and produces a themed report with quotes attached to findings. Its structured questions cover six types in one study, so a single session can mix an open_ended probe with a scale rating, a single_choice or multiple_choice pick, a ranking exercise and a yes_no screen. Published pricing: Insights at EUR 29 per month for 29 credits, Interviews at EUR 79 per month for 79 credits, with annual plans at EUR 290 and EUR 790. A text conversation costs 1 credit, a voice conversation 3, and a report refresh 5. Only conversations that clear an internal quality score consume a credit at all.
2. GrowthBook, best for a published price and a warehouse-native model. The only major experimentation platform on this list with a fully self-serve, printed price list. Starter is free, Pro is $40 per seat per month, Enterprise is custom. Self-hosting is free and open source. Infrastructure is metered transparently: managed warehouse overages at $0.03 per 1K events (equivalently $30 per million, which its own then $30/M line confirms), CDN at $10 per additional million requests and $1 per additional GB. You can model your bill before you talk to anyone, which in 2026 is a differentiator.
3. Convert Experiences, best for a printed ladder with stated overage rules. Growth is $399/mo monthly or $299/mo billed annually at $3,588/yr, both for 100K tested users per month. Pro is $599/mo or $420/mo annually at $5,040/yr. Enterprise is Price on Request. It also publishes what happens when you exceed the allowance, which almost nobody does. Two things worth computing yourself, below.
4. Amplitude Experiment, best if you already live in Amplitude. Now also the owner of Statsig's brand and customers. That consolidation cuts both ways: more investment in experimentation, and Atzberger's point about two overlapping product lines is a fair question to put to your account team. If you are weighing the analytics layer itself, see our Koji vs Amplitude comparison and Amplitude vs Heap.
5. VWO, best for a single suite across testing and insights, if you can get a quote. Post-merger it is the largest independent name here by revenue. It publishes no prices either. Full detail in our VWO alternatives breakdown.
6. LaunchDarkly, best for release control rather than marketing tests. Feature flags and progressive delivery first, experimentation second. The right choice if your bottleneck is safe rollout, the wrong one if it is landing-page conversion.
7. Kameleoon, best for server-side and hybrid testing in regulated markets. European vendor with a strong server-side story. Pricing is quote-based.
8. Microsoft Clarity plus a research tool, best for a zero-budget start. Clarity is free and shows you where people struggle. It will not run a valid experiment, and it will not tell you why. Pairing free behavioural data with interviews is a legitimate starting stack, and we compare the behavioural half in Hotjar vs Microsoft Clarity.
Two numbers Convert publishes that are worth computing
Convert prints enough to do arithmetic that it does not do for you.
On the annual Growth plan you pay $299 per month for 100K tested users. An extra 100K block costs $399 per 100k users. That is 1.33x the rate your own plan charges, so exceeding your commitment costs 33% more per user than the commitment does. On monthly Growth billing the base is $399 for the same 100K, so the overage block is exactly the same rate and there is no penalty at all. The overage penalty therefore applies only to the customers who committed to a year.
On Pro the direction reverses. Annual Pro is $420 per month per 100K, while overage is $699 per 250K, which is $279.60 per 100K, or about 33% cheaper per user than the base plan. So the same vendor charges a premium for overage on one tier and a discount on the next one up. Neither is hidden. Both are simply left for you to work out.
Optimizely vs the alternatives at a glance
| Tool | Publishes a price? | Entry price you can see | 2026 ownership status | Tells you why? |
|---|---|---|---|---|
| Optimizely | No | None published | Independent | No |
| Koji | Yes | EUR 29 per month | Independent | Yes, that is the product |
| GrowthBook | Yes | Free, then $40 per seat | Independent | No |
| Convert Experiences | Yes | $299 per month annually | Independent | No |
| Amplitude Experiment | No | Not published | Acquired Statsig brand May 2026 | Partially, via surveys |
| VWO | No | None published | Merged with AB Tasty Jan 2026 | Partially, via surveys |
| AB Tasty | No | Redirects to wingify.com | Same company as VWO | Partially |
| LaunchDarkly | Partially | Per-seat tiers published | Independent | No |
How to choose an experimentation tool in 2026
Ask four questions in this order.
Can I see a price? If not, budget for a five-figure annual commitment and a procurement cycle. That is not disqualifying, but it should be a conscious choice rather than a surprise in week three.
Who will own this vendor at renewal? A year ago that question would have sounded paranoid. After the sequence above it is basic diligence. Ask directly whether the team that builds the product still works for the company selling it.
Is my real bottleneck significance, or understanding? This is the one most teams get wrong. Kohavi, Crook and Longbotham reported in 2009 that at Microsoft only about one third of well-designed, well-executed experiments actually improved the metric they were built to improve. More recent vendor audits point the same way: ConversionTeam's review of 2,288 A/B tests that ran to a determinate win-or-lose result across 71 client engagements found Statistically significant winners: 19.1% per test, with a further 153 tests excluded because their results could not be read with confidence. That is the vendor's own client history rather than an independently published dataset, so treat the precise figure with care. The direction, though, is consistent across every source: most tests do not produce a clean answer.
Buying a faster testing engine does not fix that. Roughly four out of five experiments end without a decision, and no amount of extra traffic explains the reasoning of the people who did not convert.
Will anyone act on the result? A significant lift with no explanation is hard to generalise. A losing variant with six interviews explaining the confusion is a roadmap item. If you want the method, start with our guides to avoiding leading questions and analyzing interview results, and see AI interviews vs surveys for why an open text box does not get you there.
One more thing worth knowing before you design the study: your experiment only observed the people who stayed. The ones who left are missing from every dashboard, which is the classic trap covered in survivorship bias in customer research and sampling bias.
Frequently asked questions
Does Optimizely publish its pricing in 2026?
No. As of this writing optimizely.com/pricing lists ten product lines and contains no dollar figures. The page states that every Optimizely plan is individually packaged and invites you to describe your needs so a plan can be created with you, which means the only route to a number is a sales conversation.
How much does Optimizely actually cost?
Public estimates are thin. CostBench reports a median of $50,000 per year but bases it on only 3 verified purchases and flags the data as thin itself, and another range on the same page is inconsistent with that annual median. Plan for a five-figure annual commitment, treat any specific scraped figure as unverified, and get a written quote before budgeting.
Is AB Tasty a good alternative to Optimizely or VWO?
AB Tasty is a credible experimentation platform, but it is no longer an alternative to VWO because the two merged under Everstone Capital in January 2026 and now form a single company. Both abtasty.com and its pricing path return a 301 redirect to wingify.com. Choosing AB Tasty to get away from VWO means choosing the same vendor.
What happened to Statsig, and should existing customers worry?
OpenAI acquired Statsig in September 2025 for $1.1 billion, then Amplitude took on the Statsig brand and customers on 5 May 2026 while the original team stayed at OpenAI. Amplitude has committed to maintaining and developing the platform and supporting existing customers. Optimizely's CEO publicly called it code without the talent and warned customers about slowing innovation, so ask Amplitude directly about roadmap ownership and long-term plans for its two overlapping experimentation products.
Which experimentation tools still publish a self-serve price?
GrowthBook and Convert Experiences are the clearest examples. GrowthBook publishes a free Starter tier, Pro at $40 per seat per month, free open-source self-hosting, and itemised infrastructure overages. Convert publishes Growth at $399 monthly or $299 monthly billed annually, Pro at $599 or $420 annually, and its overage rates per additional block of tested users.
Can Koji replace my A/B testing tool?
No, and it should not. Koji does not split traffic or calculate significance, so you still need an experimentation platform for that. Koji answers the separate question your test cannot: why people behaved the way they did. Teams typically run the experiment, then interview the users who saw the losing variant to learn what actually went wrong.
Stop guessing why the variant lost
Your experimentation platform will tell you that version B underperformed by 4%. It will never tell you that three people could not find the shipping cost and two thought the plan was per seat.
Koji closes that gap. Launch an AI-moderated study in minutes, let it interview participants by voice or text around the clock with no moderator bias and no scheduling, and get a themed report with supporting quotes instead of a folder of recordings. No research background required, and insights land in hours rather than weeks.
Koji starts at EUR 29 per month, the price is on the website, and new accounts get free interview credits to try it on a real question. If you are deciding between running another test and actually understanding the last one, start with the understanding.
