--- title: "6 Wingify alternatives for AB testing in Brazil (formerly VWO and AB Tasty)" slug: "wingify-alternatives-ab-testing-brazil-vwo-ab-tasty" locale: "en-us" description: "If you signed a VWO or AB Tasty contract, the company you signed with no longer trades under that name. The two combined. On September 16, 2026, they relaunched as a single brand, Wingify." published_at: "2026-10-20" updated_at: "2026-10-20" image: "https://storage.googleapis.com/croct-assets-b931d070/blog/Wingify_alternatives_for_AB_testing_in_Brazil_formerly_VWO_and_AB_Tasty_5827459745/Wingify_alternatives_for_AB_testing_in_Brazil_formerly_VWO_and_AB_Tasty_5827459745.png" authors: ["Juliana Amorim"] category: "Experimentation" tags: ["Comparison","A/B testing"] canonical: "https://blog.croct.com/post/wingify-alternatives-ab-testing-brazil-vwo-ab-tasty" --- # 6 Wingify alternatives for AB testing in Brazil (formerly VWO and AB Tasty) If you signed a VWO or AB Tasty contract, the company you signed with no longer trades under that name. The two combined. On September 16, 2026, they relaunched as a single brand, **Wingify**. As of October 2026, both websites redirect to `wingify.com`, and its pricing page contains no prices. Where VWO once published self-serve tiers you could put on a company card, Wingify now says: *"No fixed plans. Just a custom proposal built around your goals and scope."* For a buyer in London or Chicago, that's a mild annoyance. For a buyer in Brazil, it changes a lot: a custom quote in dollars, from a vendor with no Brazilian entity, is not the same purchase as a published price you can budget against. This post is for teams already on VWO or AB Tasty who are deciding whether to stay. If you're choosing an experimentation platform from scratch, start with our broader guide to [the best AB testing platforms for the Brazilian market](/post/ab-testing-brazil.md) instead. The alternatives below are Croct, Convert, GrowthBook, PostHog, Statsig, and Optimizely. For most Brazilian teams running web experimentation and personalization, we think **Croct** is the strongest replacement. I know I'm one of the founders, so weigh that accordingly and read the watchouts in every section, including ours. > 💬 > > We checked all prices in October 2026, but they change often, so we always suggest you check them yourself. ## What actually changed when VWO and AB Tasty became Wingify Three things, but only the first gets discussed. ### The brands consolidated into one The merger was announced on January 20, 2026, "subject to customary closing conditions and approvals," and the combined business launched under the Wingify name on September 16, 2026. The launch release states more than US\$100M in revenue and over 4,000 customers, but it did not disclose the terms. Both products still run. Wingify describes them as "two established, independently built platforms" and says that "over time, both VWO and AB Tasty are expected to fully converge" under the Wingify umbrella (with no published timeline for what converged looks like or what it costs). As of today, there are still two separate logins. ### The catalog expanded, so the quote does too Wingify now sells six suites: Web Experimentation and Rollout, Web Personalization, Feature Experimentation, Behavior Analytics, Journey Analytics, and User Feedback. Each has its own Growth, Pro, and Enterprise tiers. If you came to VWO for AB testing, your renewal conversation is now with a company that has five other modules to attach. ### Published pricing went away This is the change that matters most and gets mentioned least. The Wingify pricing page lists features by tier with no figures and routes every buyer to `Schedule a Demo`. It advertises a 30-day free trial with no credit card required, and shows no permanent free tier. If you are currently on a legacy VWO free or Starter plan, confirm directly with Wingify what happens to it, because the Wingify pricing page does not describe one. For historical context, Vendr's February 2026 data put the median VWO contract at **US\$16,330/year** across 102 purchases, with negotiated deals typically landing 15–30% below the first quote. AB Tasty never published list prices, but Vendr's February 2026 median for it was **US\$66,500/year**. To be completely honest, none of this makes Wingify a bad product. It's mature, deep, and still the right answer for some teams. But it does mean a renewal is the moment to check whether you're still buying the thing you originally chose. ## The Brazil filter: what a USD contract actually costs Rankings written for buyers in the US and Europe optimize for features. From Brazil, four things drive the decision more than any feature. > 💬 > > **What follows is general information about how these costs typically work, not tax or legal advice.** Rates change, municipal rules vary, and several points below are genuinely contested. Confirm your own position with your accountant and counsel. ### The list price is not the landed price A US\$500 subscription does not cost US\$500. That's a fact. Paying a foreign software vendor typically triggers a stack of taxes: - **IRRF** withholding at 15%, and since SaaS vendors bill a fixed amount rather than absorbing withholding, that generally gets grossed up. - **PIS/COFINS-Importação** at 9.25%, creditable under the non-cumulative regime that companies on *lucro real* usually follow, and not creditable for *lucro presumido* or *Simples*. - **ISS** at the municipal rate, 2% to 5% (São Paulo charges 2.9% on software licensing). - **IOF-câmbio** at 3.5%. - And possibly **CIDE-Remessas** at 10%, depending on how the contract is characterized. Model that out and a US\$500 subscription plausibly lands somewhere between **US\$675 and US\$751**, a **35% to 50% gross-up**. A company crediting PIS/COFINS brings that down to roughly US\$620–700, or 24% to 39%. Comparing a USD list price against a BRL one at the spot rate understates the foreign option by about a quarter to a half. ### YOU need to issue the Nota Fiscal A foreign vendor isn't registered as a service taxpayer with your municipality, so in practice it cannot issue a Nota Fiscal for you. The work doesn't disappear, it moves to **you**. The Brazilian buyer self-issues the NF, withholds and remits ISS, calculates IRRF and PIS/COFINS-Importação, and reconciles all of it against the contrato de câmbio. That's a finance-team effort the vendor never sees and never prices in. It isn't a blocker. Thousands of Brazilian companies buy foreign SaaS every day. It is a recurring monthly cost in someone's time, and it's the reason procurement teams push back on foreign vendors in ways that look irrational from the outside. ### FX is a line item you do not control Over the last 24 months, the monthly average USD/BRL rate ranged from **R\$6.10 in December 2024 to R\$4.98 in May 2026**. An unhedged US\$500 subscription cost about R\$3,049 at the December 2024 average and about R\$2,530 at today's rate. Same contract, same product, a 20% difference in what hits your P\&L. On a three-year commitment, that variance is larger than most of the discounts you'll negotiate. ### LGPD and data transfer LGPD does **not** require data localization. There is no general storage-in-Brazil mandate. What LGPD does regulate is the transfer. International transfers of personal data need a recognized basis. A US vendor generally has to sign Brazil's specific annex. Some will, some won't, and finding out takes weeks. A Brazilian vendor processing data in Brazil has no international transfer to legitimize at all. ### Where the offices actually are Wingify has sales team members working in Brazil, but there aren't any offices in Latin America. They do have Portuguese marketing content, but they do not have a Brazilian entity, do not charge in BRL, have no Nota Fiscal capability, and usually do not provide local support in Portuguese. TechCrunch reported at the time of the merger that about 90% of the combined company's revenue comes from the US and Europe. Portuguese content is a marketing decision. A Brazilian operation is a commercial one. They aren't the same thing. ## Wingify alternatives at a glance | Alternative | Best fit | Execution and analysis model | Brazil posture | Main watchout | | -------------- | ------------------------------------------------------------------ | -------------------------------------------------- | ----------------------------------------------------------------------------- | ---------------------------------------- | | **Croct** | Brazilian teams wanting testing, segmentation, and personalization | Server-side resolution, built-in Bayesian analysis | Brazilian company, prices published in BRL, Portuguese support in local hours | No mobile app SDKs (API-only) | | **Convert** | High-traffic sites that need a published price | Client-side first, with server-side APIs | USD only, no Brazilian entity | Priced per tested user, overage is steep | | **GrowthBook** | Teams with a data warehouse and engineers | Server-side SDKs, warehouse-native analysis | USD only, no Brazilian entity | Needs a warehouse and someone to own it | | **PostHog** | Product teams wanting analytics, replay, and flags in one | Server and client SDKs, built-in product analytics | USD only, no Brazilian entity | Usage-based billing is hard to forecast | | **Statsig** | Engineering-led teams wanting a large free tier | Server-side SDKs, warehouse-native on Enterprise | USD only, no Brazilian entity | Stewardship changed twice in 14 months | | **Optimizely** | Large programs spanning web, apps, and services | Server, client, and edge SDKs | USD only, no Latin American office | Median contract US\$81k/year | ## What to compare before you replace Wingify Feature matrices are close to useless here, because everything on this list can run an AB test. What separates them is where the decision executes, whose numbers you end up trusting, and how much of the program a non-engineer can run alone. ### Where the variant is decided If organic search matters to your acquisition, this is worth deciding early rather than late. A client-side tag loads on every page whether or not a test is running, and its cost shows up in LCP and INP. Measure it rather than taking anyone's word for it, including ours. - **Client-side** The browser loads the original page, then JavaScript rewrites the DOM. Fast to set up, and it's how VWO's and AB Tasty's visual editors work by default. It also means the control paints before the variant replaces it, tests break when a refactor changes a CSS selector, and whatever the ad blocker stops is simply missing from your data. - **Server-side** The variant is chosen during the render, before the response leaves the server. No flicker, no selector fragility, no ad-blocker loss on assignment, and experiments can reach into application logic rather than only visible elements. This is Croct's default. - **Edge** The same server-side decision relocated to a CDN worker. A latency optimization, not a different methodology. > 💬 > > We've written separately on [why visual editors cost performance](/post/visual-editors-cro-perfomance.md) and [how server-side execution avoids it](/post/server-side-vs-client-side-ab-testing.md). ### Assignment, exposure, and sample ratio Three questions, and insist on seeing the answers in your own environment rather than a demo account. 1. Is bucketing deterministic? Does hashing the visitor ID and experiment key always produce the same variant? 2. Does assignment survive a returning visitor, including on Safari, where Intelligent Tracking Prevention caps script-set cookies at seven days? 3. Does the platform surface a sample ratio mismatch, or do you have to notice it yourself? A platform that can't tell you when a 50/50 split delivered 53/47 isn't giving you experiments. It's giving you dashboards. ### Metric definitions and governance Not whether the tool has goals, but whether a conversion defined in the testing platform means the same thing as the conversion in your analytics tool and your warehouse, and who's allowed to change it. Teams that skip this discover eighteen months later that half their historical wins were measured against a metric somebody quietly redefined. ### Statistical method and stopping rules Bayesian and frequentist engines are both defensible. Undisciplined peeking is not, and it's the most common way a program produces confident nonsense. Find out what the platform does when someone opens the results on day two. Sequential and Bayesian methods are built for continuous monitoring, a fixed-horizon frequentist test isn't, and a dashboard showing a p-value updating in real time is actively encouraging the wrong behavior. > 💬 > > Our [Bayesian versus frequentist comparison](/post/bayesian-vs-frequentist.md) covers the trade-off. ### Who can launch a test This sets your testing velocity, and velocity drives results far more than statistical sophistication does. Count honestly: how many people can ship a variant to production today without a developer, and how many working days pass between having a hypothesis and having it live? Most programs are capped by sprint cadence, not by ideas. Moving that number from three weeks to one afternoon is worth more than a better stats engine. We've written about [the dev-dependency problem](/post/cro-dependency-on-devs.md) at length. ## The alternatives ### 1. Croct: best for teams operating in Brazil [Croct](https://croct.com) is a Brazilian experimentation and personalization platform. We offer AB tests, personalization, and feature flags on one server-side resolution layer, and keep visitor profiles in the product rather than expecting you to bring a CDP. For this list, it matters for a reason that has nothing to do with features: we're the only option here that is a Brazilian company, publishes prices in BRL, and staffs support in Brazilian business hours in Portuguese. We're the best fit for Brazilian growth and product teams on a modern JavaScript or PHP stack that want testing, segmentation, and personalization in one tool. It's a particularly direct swap for two groups: VWO customers whose program is web experimentation and personalization rather than mobile apps, and VTEX merchants who lost the native AB Tester and have been improvising since. Key strengths: - Priced in reais, not converted into them - Portuguese support in Brazilian hours, from a team in Brazil, not a translated help center - Server-side resolution, so no flicker - Marketers launch without a deploy - Bayesian analysis - One mechanism for testing, segmentation, and personalization - Forever-free plan so you can try it out with no commitments Watchouts: - You will need a developer's help once, for the initial integration. It is usually very simple and fast. - We still do not provide mobile app SDKs. Our customers usually integrate with apps using our APIs directly. - We do not provide a visual editor in the VWO sense, and do not plan to. Croct is headless-first, and content is modeled as structured slots rather than edited by pointing at the live DOM. - Our plan tiers are based on the number of monthly visitors, regardless of whether they're impacted by an experiment or not. For a proof of concept, run one component-level test on a high-traffic page and one audience-based personalization on the same slot. That exercises both halves of the product inside a week, on the free tier, before any money moves. ### 2. Convert Convert is one of the few CRO platforms that still publishes its prices. It is a good option for teams for whom the visual-editor workflow is non-negotiable. Growth starts at **US\$299/month** billed annually (US\$399 monthly) for 100,000 tested users. Overage is steep at US\$399 per additional 100,000 users. It's a mature visual editor, so the transition from VWO is the least disruptive here. But multivariate testing, feature flags, multi-armed bandit, sequential testing, and SSO all sit above Growth. There's a 15-day trial but no free tier. No Brazilian entity and no BRL billing, though Convert does list Brazilian certified partner agencies, which helps with implementation but doesn't change the tax arithmetic. ### 3. GrowthBook GrowthBook is open-source and warehouse-native: it handles assignment and the statistics, and reads events from the warehouse you already have rather than becoming a second source of truth. It is a very good tool for scale-ups with a data team, an established warehouse, and a reason to keep experiment analysis next to everything else. The free cloud Starter covers 3 users and one project with unlimited flags, experiments, and traffic. Pro unlocks sticky bucketing, sequential testing, CUPED variance reduction, and bandits. **Warehouse-native is a prerequisite, not a feature**: without a warehouse and someone to own it, GrowthBook is the wrong tool. It's **developer-first**, so marketers won't be launching tests unaided, which for most teams coming off a visual editor is the whole problem. No Brazilian presence, no Portuguese support, no BRL. ### 4. PostHog PostHog is product analytics, session replay, feature flags, and experiments on one usage-based bill, with no per-seat charge. It is a great option for product teams that want experimentation adjacent to analytics and replay, and that would rather pay for what they use than negotiate a seat count. The free monthly allowance is substantial: 1 million feature flag requests, 1 million analytics events, and 5,000 session replays, free every month before anything is billed. Beyond that, flag requests run US\$0.0001 and fall to US\$0.00001 at volume, which is hard to forecast. Finance teams dislike it for good reason: a traffic spike is also a bill spike. Experimentation is one feature of a large product rather than the center of it, so the experiment workflow is thinner than a dedicated CRO tool's and it is **developer-oriented**: there's no marketer-facing visual editor of the kind you're leaving. Also, no Brazilian presence. ### 5. Statsig Statsig is a warehouse-capable experimentation platform with an unusually generous free tier. Its stewardship has also changed twice in fourteen months, so it's unclear where it is headed. It is usually recommended for engineering-led teams that want serious experimentation infrastructure without a contract, and that can tolerate some uncertainty about the product's trajectory. The free Developer tier includes 2 million analytics events per month, **unlimited feature flag checks**, 50,000 session replays, and unlimited seats. This is, however, the acquisition-risk case study. OpenAI acquired Statsig in September 2025 in a deal reported at US\$1.1B. In May 2026, Amplitude announced a partnership under which it takes on Statsig's brand, customers, platform, and roadmap, while the engineering team remains at OpenAI. The product is still called Statsig, still priced separately, and customers were not migrated. No public document states the legal structure of the Amplitude arrangement, and nobody has published what Statsig looks like or costs in 2027. No Brazilian presence. ### 6. Optimizely Optimizely is the enterprise incumbent: web, feature, and full-stack experimentation across SDKs for nearly every runtime, inside a broader digital experience suite. It has long been the go-to option for large organizations running experiments across web, native mobile apps, and backend services, where SDK breadth justifies the cost and implementation effort. The cost is the headline. Vendr's February 2026 data puts the median Optimizely contract at **US\$81,447/year**, with a range from US\$26,460 to US\$261,075. No published pricing, so every buyer enters a sales cycle. Implementation is heavy, and connecting it to a headless CMS usually means custom middleware. Its offices page lists fourteen locations across North America, Europe, Asia Pacific, and the Middle East. None in Latin America. ## How the alternatives differ in practice | If your priority is… | Strongest candidates | What to verify | | -------------------------------------------------- | -------------------------------------------- | --------------------------------------------------------------------------------- | | Paying in BRL | Croct | That BRL invoicing and nota fiscal are available on your plan, not only via sales | | A published price you can budget against | Croct, Convert, GrowthBook, PostHog, Statsig | Whether your actual traffic and experiment count fall inside the published band | | Keeping the visual-editor workflow | Convert | That the editor survives your component refactors, not just a demo page | | Analysis in your own warehouse | GrowthBook, Statsig | That warehouse-native isn't gated behind an enterprise tier | | Eliminating flicker and protecting Core Web Vitals | Croct, GrowthBook, Statsig | LCP and INP measured with the tool installed and removed | | Marketers launching without engineering | Croct, Convert | Time a real marketer shipping a real variant, unaided | | Mobile apps plus web in one program | Optimizely, PostHog, Statsig | Whether the mobile SDK has parity with web, or is a thinner version | | Minimizing LGPD transfer complexity | Croct, GrowthBook self-hosted | For foreign vendors: whether they'll sign ANPD's clauses | ## How to migrate off Wingify Most failed migrations fail in the same place: the new tool reports different numbers than the old one, nobody can explain why, and the program loses credibility before it produces a result. This sequence exists to prevent that. ### 1. Inventory what is actually running Export every live and recently concluded test. For each, record what it changes, where the change executes, which metric decides it, and who built it. You'll usually find three groups: 1. Visual tweaks a marketer made in the editor 2. Structural changes a developer shipped 3. Tests nobody has looked at in months. The third group is the useful finding: turn those off before migrating anything, because a dormant test still fragments traffic and still carries a tag. ### 2. Decide what each test is really measuring A visual-editor test that moves a button and a server-side test that changes a pricing rule look identical in a dashboard and are not the same experiment. Group your inventory by what the variant actually contains: copy and layout, content and offers, or application logic. That grouping tells you which replacement you need. ### 3. Validate with an A/A test Run the same content against itself on real traffic. A correct platform shows no reliable difference and a sample ratio close to what you configured. If an A/A test produces a winner, the problem is instrumentation, not your hypotheses. ### 4. Export your history Wingify's published retention policy keeps raw visitor data for most products for 13 months, but Behavior Analytics raw data is kept for less time, and there's a short grace period after unsubscribing before deletion. Check the product you actually care about, and export before you give notice rather than after. ### 5. Read your exit terms Then find your renewal date and your notice window. Contracts in this category commonly auto-renew with 30 to 90 days' notice. Given that convergence between the VWO and AB Tasty platforms has no published timeline, ask in writing what your product and price look like after it, and what happens if the answer changes mid-term. ## A proof-of-concept scorecard for Brazilian teams Run the same checks against every finalist, in your own environment, before anyone sees a sales deck. 1. **Deterministic assignment:** The same visitor gets the same variant across sessions and, once identified, across devices. 2. **Sample ratio:** A 50/50 split delivers close to 50/50 exposure, and the platform tells you when it doesn't. 3. **No flicker:** Load a test page on a throttled connection. Any visible content swap disqualifies the configuration, and usually the execution model. 4. **Core Web Vitals:** Measure LCP and INP with the tool installed and removed. Record the delta. 5. **Exposure integrity:** Compare platform-recorded exposures against your own analytics. Ad-blocker loss should be visible and explainable. 6. **Time to launch without engineering:** A marketer creates and ships a variant unaided. Time it. 7. **Failure behavior:** Block the vendor's domain in DevTools. The page should render the control, not an error or an empty region. 8. **Landed cost in BRL:** Build the twelve-month total including FX, IOF, and import-service taxes. Compare that figure, not the list price. 9. **Transfer basis:** For any foreign vendor, confirm in writing which LGPD mechanism applies and whether they'll sign ANPD's clauses. ## Which Wingify alternative is right for your team If your experimentation runs on the web, your team is in Brazil, and you want testing and personalization on one bill in BRL, **Croct** is the strongest replacement (and the only one that takes the tax and USD arithmetic out of the decision rather than asking you to absorb it). If you need a published price at high traffic and the visual-editor workflow is non-negotiable, **Convert** is the best of the foreign options. If analysis belongs in your warehouse, **GrowthBook**. If you want analytics, replay, and experiments on one usage-based bill, **PostHog**. If you want the largest free tier and can live with developer dependency and uncertainty about the product's stewardship, **Statsig**. If your program genuinely spans mobile apps and backend services, **Optimizely** earns its price. Whichever way you go, the discipline matters more than the logo. Build the landed cost in BRL rather than comparing list prices. Get the LGPD transfer basis in writing. Run the A/A test. And ask every vendor what happens to your contract if they're acquired, because in this category, over the last eighteen months, that question has stopped being hypothetical. ## FAQ **What happened to VWO and AB Tasty?** They combined and relaunched as a single brand, Wingify, on September 16, 2026. The merger was announced on January 20, 2026. Wingify is VWO's former parent company; Everstone Capital took majority control of it in January 2025. Both products continue to run during a transition, and Wingify says they are expected to converge over time, with no published timeline. **Is Wingify the same company as VWO?** Wingify built and owned VWO, and is now the brand for the combined VWO and AB Tasty business. As of October 2026, `vwo.com/pricing` and `abtasty.com/pricing` both redirect to `wingify.com/pricing`. **Does Wingify still have a free plan?** The Wingify pricing page advertises a 30-day free trial with no credit card required and lists no permanent free tier. Its lowest plan is Growth. If you're on a legacy VWO free or Starter plan, confirm its status directly with Wingify. **How much does Wingify cost?** Wingify does not publish prices. Its pricing page lists Growth, Pro, and Enterprise tiers across six product suites with no figures, and directs buyers to contact sales. For historical reference, Vendr's February 2026 data put the median VWO contract at US$16,330/year and the median AB Tasty contract at US$66,500/year. **Which AB testing tools bill in Brazilian reais?** Croct publishes prices in BRL, starting at R\$600/month for 20,000 monthly active users on annual billing, with a free tier up to 10,000 MAU. The other platforms in this comparison price in USD or EUR only, which adds FX exposure, IOF, and import-service taxes to the effective cost. **How much more does foreign SaaS cost a Brazilian company?** Modeling IRRF, PIS/COFINS-Importação, ISS, IOF, and possibly CIDE, roughly 35% to 50% above the USD list price, or 24% to 39% for a company able to credit PIS/COFINS. Rates change, ISS is municipal, and CIDE's application to SaaS is contested. Treat this as a planning range and confirm with your accountant. **Does LGPD require AB testing data to be stored in Brazil?** Not as a general rule. LGPD contains no data localization mandate. It does regulate international transfers. ANPD recognized the European Union as adequate in January 2026, but the United States is not covered, so transfers to US vendors generally rely on ANPD's standard contractual clauses. **Is there a free AB testing tool for Brazilian teams?** Croct is free up to 10,000 monthly active users with AB testing, Bayesian analysis, and personalization included. GrowthBook, PostHog, and Statsig all offer substantial free tiers, and GrowthBook and PostHog can be self-hosted, though none of the three has a Brazilian entity or Portuguese support. **Can I keep my test history when I switch platforms?** Rarely in a usable form. Most vendors export raw event or result data but not the experiment configuration or the statistical reasoning behind past decisions. Treat historical results as a document to archive rather than a dataset to migrate, and export before you give notice.