top of page
Business & Enterprise
MarTech
Advertising Technology Platform
AB Tasty
Won enterprise experience optimization by giving marketers a no-code way to test and personalize digital experiences, then expanding into feature flags, recommendations and AI-driven personalization, despite free analytics tools and developer-native experimentation platforms.
1
MODEL
BUSINESS MODEL
SaaS
- Consulting-to-product transition: Liwio began as a web-analytics and conversion consultancy; customer work revealed that marketers needed to test ideas without engineering support.
- No-code experimentation wedge: AB Tasty’s visual editor let marketers create A/B and multivariate tests without writing code.
- Platform expansion: Flagship, launched in 2019, added feature flags and server-side experimentation for product and engineering teams.
- Acquisition-led capability expansion: Epoq added AI recommendations and search; Dotaki added emotional modeling; Potions added e-commerce personalization; Wandz.ai added real-time adaptive experiences.
- Traffic-and-module monetization: AB Tasty prices annual subscriptions by traffic volume, selected modules and support tier, creating expansion as customers increase usage and adopt additional capabilities.
1. Find a decision made by intuition. Identify a recurring business decision—pricing, messaging, layout, product release or customer journey—where teams currently debate rather than test.
2. Give the non-technical decision-maker control. Build the first product so the person closest to the customer can run the experiment without filing an engineering request.
3. Instrument the outcome, not just the activity. Track conversion, revenue, retention or task completion so the customer can compare decisions objectively.
4. Expand from testing to decision infrastructure. Add feature flags, targeting, recommendations or AI decisioning only after experimentation becomes habitual.
5. Price by the scale of decisions influenced. Use traffic, experiments, users or modules as the meter, aligning revenue with the customer’s digital activity.
Avoid: selling an experimentation tool before the customer has enough traffic, ownership and decision cadence to act on results; low-volume testing becomes a feature, not a platform.
DISTRIBUTION MODEL
AB Tasty proved its initial wedge—visual, no-code A/B testing for marketers—before expanding into broader experience optimization. Founded as Liwio in 2009 by Alix de Sagazan and Rémi Aubert, it launched AB Tasty in 2011 and closed the agency in 2012 to focus on software. After establishing web experimentation, it expanded into server-side feature management through Flagship in 2019, then acquired Epoq in 2022, Dotaki in 2023 and Potions in 2024 to add recommendations, emotional segmentation and e-commerce personalization. By 2025, AB Tasty served more than 1,000 brands with over 300 employees across Europe, North America and Asia-Pacific.
HOW THEY BUILT IT
HOW TO ARCHITECT IT
Direct Sales, Content Marketing, SEO, Partnerships
HOW THEY OPERATIONALIZED
HOW TO REPLICATE WHAT WORKED
dm
- Consulting-origin demand: Liwio’s conversion and analytics clients became early AB Tasty users because the tool solved a problem they already experienced.
- Direct enterprise sales: AB Tasty sold experimentation and personalization programs to marketing, product and digital-experience leaders.
- SEO around experimentation intent: It targeted searches for A/B testing, personalization, feature flags and conversion-rate optimization.
- Agency and technology partnerships: Implementation partners, e-commerce agencies and technology integrations extended reach among brands needing deployment support.
What worked: Use consulting or services to discover a repeated decision bottleneck, then productize the workflow for the non-technical decision-maker; direct sales can expand the account as experimentation matures.
Trap if copied blindly: Enterprise sales fail when the customer lacks traffic, data discipline or a decision owner; without enough experiments and measurable outcomes, the platform becomes expensive instrumentation rather than a decision system.
| PATTERNS OF THIS MODEL
PATTERNS IN ENTERPRISE EXPERIMENTATION SAAS:
1. No-code control expands the buyer base. When marketers can test without engineering, experimentation moves from a developer capability to a growth-team operating system.
2. Experimentation naturally expands into release management. Once teams test customer-facing changes, they need feature flags, rollout controls and rollback mechanisms.
3. Personalization raises value but raises complexity. Recommendations and adaptive experiences can increase conversion, but require data quality, privacy compliance and meaningful traffic volume.
THE WEAKNESS: Experimentation platforms can be squeezed between free analytics tools and developer-native platforms; if the customer’s engineering team owns experimentation, it may prefer open-source or API-first tools over marketer-oriented suites.
What companies with this model reveal
| OPPORTUNITY INTELLIGENCE
GOLDMINE 1 — DECISION-QUALITY LAYER. The opportunity is not more tests but better decisions: experiment design, sample-size guidance, statistical rigor, rollout governance and a searchable library of validated learnings. Most platforms measure variants; few help organizations accumulate decision intelligence.
GOLDMINE 2 — AI EXPERIENCE AGENTS WITH GUARDRAILS. AI can generate and adapt experiences, but enterprises need guardrails: brand constraints, experiment design, bias checks and rollback. A governed AI experimentation layer can be more valuable than autonomous personalization.
GOLDMINE 3 — PRIVACY-FIRST PERSONALIZATION. Cookie deprecation and privacy regulation create demand for contextual, first-party and consent-aware personalization. The winner will personalize without relying on fragile third-party identity.
THE PIT — PERSONALIZATION WITHOUT TRAFFIC DENSITY. Personalization requires enough visitors per segment to produce statistically meaningful results; low-traffic brands may see noise rather than lift.
THE SECOND PIT — MARKETER-ONLY TOOLS IN ENGINEERING-LED ORGS. If engineering owns releases and data, a visual editor alone feels limited; the platform must serve both decision-makers and developers.
MOVE WITH CAUTION: Do not build another general-purpose A/B testing tool. Without a distinct buyer, data model or workflow advantage, the platform competes with free analytics, feature flags and cloud-native experimentation.
Untapped Business Model / Gaps / Goldmines / Pits
Patterns & Insights
2
MARKET
mkt mt es
AB Tasty entered a market with Google Optimize, Optimizely, VWO, Adobe Target and developer-led experimentation tools. Its differentiation was marketer autonomy: a visual editor allowed marketing and e-commerce teams to launch tests without engineering tickets. Later, Flagship differentiated toward product and engineering teams through feature flags and server-side experimentation, while acquisitions added recommendations and emotional personalization. This created a distinct buying reason for brands that wanted both marketing-led testing and engineering-controlled releases.
MARKET TYPE
Red Ocean
ENTRY STRATEGY
Direct Entry
WHY THEY WON
Experimentation and personalization became a red ocean as analytics vendors, enterprise experience platforms and developer tools converged. AB Tasty changed the buyer’s frame from “A/B testing tool” to “experience optimization platform,” giving marketers direct control while adding engineering-grade feature management. Its structural opening was the organizational gap between marketers who understood customer experience and developers who controlled deployment. The transferable lesson is that crowded markets reward platforms that resolve a cross-functional workflow gap rather than compete solely on statistical features.
EXECUTION
AB Tasty entered directly with its own SaaS product after operating as a conversion-analytics consultancy. The initial bet was that marketers would pay for a visual testing tool if it removed their dependence on developers. Direct entry was viable because Liwio’s client relationships provided early users, domain knowledge and a clear pain point. The essential prerequisite was a reliable testing engine and a simple editor; the approach would fail if marketers could create tests but could not trust the statistical results or deployment stability.
FOOTHOLD STRATEGY
Wedge Strategy
AB Tasty’s wedge was no-code A/B testing for marketing and e-commerce teams. The pain was slow engineering cycles and guesswork in website optimization; these buyers were reachable through the founders’ consulting relationships, SEO and direct sales. They could pay because conversion improvements directly affected revenue. From web testing, AB Tasty expanded into feature flags, personalization, recommendations and AI-driven experience optimization.
GROWTH CAMPAIGN
CAMPAIGNS THAT WORKED
SO WHAT: AB Tasty’s campaigns moved the buyer from “run a test” to “operate a continuous experience-optimization program.”
KEY LEARNING: The non-obvious principle is that a wedge aimed at the non-technical decision-maker can create demand, but platform expansion requires earning the engineering team’s trust through server-side control and reliable release management.
HOW TO USE: If your product serves a business user but depends on technical implementation, build a shared workflow that gives the business user autonomy and the technical team control; measure cross-functional adoption, not only marketer logins.
KEY LEARNING
THE STANDARD: In a red ocean, the entrant wins by resolving a cross-functional workflow gap rather than adding another feature to an existing category.
RULE 1 — IDENTIFY THE HANDOFF FAILURE. AB Tasty attacked the marketer-engineer handoff; look for decisions delayed because the person with customer context cannot execute without technical support.
RULE 2 — MAKE THE FIRST USER AUTONOMOUS. The initial wedge should let the business user act without creating operational risk.
RULE 3 — EARN THE TECHNICAL GATEKEEPER’S TRUST. Expansion requires server-side control, security, governance and reliable deployment.
RULE 4 — KNOW WHEN THE CATEGORY COLLAPSES. If analytics, feature flags and experimentation converge into a single platform, point solutions must deepen a workflow or move up into decision intelligence.
MARKET TYPE: Red Ocean (category).
gc
Market Context
| MARKET INTELLIGENCE
PATTERNS IN CONSULTING-ORIGIN ENTERPRISE SAAS WITH DIRECT SALES:
1. Services reveal the product wedge. Consulting exposes the repeated customer problem that software can standardize.
2. Direct sales expands from user adoption to program adoption. A marketer-led test becomes an enterprise experimentation program with governance and cross-team rollout.
3. Acquisitions can add adjacent decision capabilities. Recommendations, personalization and adaptive experiences extend the platform when they reuse the same customer and traffic context.
4. THE WEAKNESS: Acquisition-led expansion can create integration debt; disconnected capabilities become a portfolio rather than a coherent decision system.
| MARKET ENTRY PLAYBOOK
THE STANDARD: Direct entry is appropriate when the company can build its own product, reach its buyer directly and deliver standalone value without a host platform.
RULE 1 — Start with a customer you already understand. A services or consulting origin can reveal a repeated pain and provide early reference customers.
RULE 2 — Give the buyer immediate autonomy. The first product should remove a dependency that currently slows the customer’s decision cycle.
RULE 3 — Build trust with the technical gatekeeper early. Data accuracy, performance, security and deployment control determine whether the platform expands beyond the initial user.
RULE 4 — Reject direct entry when the buyer lacks decision volume. If the customer runs too few experiments or has too little traffic, the platform cannot demonstrate ongoing value.
Example: AB Tasty entered experimentation directly; a failure pattern is direct entry into low-traffic businesses where testing cannot produce statistically meaningful outcomes.
How to enter
| FOOTHOLD STRATEGY PLAYBOOK
THE STANDARD: A wedge strategy works when one narrow, repeatable pain creates habitual use and reveals adjacent needs within the same customer.
RULE 1 — Choose a pain with a direct revenue or conversion link. Conversion, signup, cart completion and feature adoption create measurable value.
RULE 2 — Expand through the same decision workflow. Once teams test experiences, they need feature flags, targeting, recommendations and rollout governance.
RULE 3 — Measure experiment velocity and decision quality. A wedge is working when teams run meaningful tests, learn from them and change decisions.
RULE 4 — Adapt to organizational structure. Marketing-led companies need visual autonomy; engineering-led companies need APIs, SDKs and release controls.
Failure pattern: expanding into personalization before the customer has enough traffic, data quality and experimentation discipline.
How to get the first strong position
MARKET PATTERNS & PLAYBOOK
3
MONEY
money rev pri
AB Tasty chose digital traffic and selected modules as its primary pricing meters. Customers pay annual subscriptions based on monthly traffic, modules activated—experimentation, personalization, recommendations or full platform—and support tier. This aligns price with the scale of digital experience being optimized and creates expansion as brands add traffic, channels and capabilities.
REVENUE MODEL
Subscription, Usage-Based
PRICING MODEL
Custom Enterprise Pricing, Tiered Pricing, Usage-Based Pricing
WHY THEY WON
Customers pay recurring annual subscriptions based on traffic volume and selected modules. Revenue expands as brands increase traffic, add experimentation across web and app, adopt feature management or activate personalization and recommendations. Software delivery supports strong gross-margin potential, though enterprise support, implementation and data-processing costs rise with account complexity.
AB Tasty uses custom annual pricing based on traffic, modules and support level. Industry references suggest entry pricing around $60,000 annually, with enterprise contracts reaching about $150,000 depending on traffic and features. The meter is digital traffic and platform scope, not seats alone.
TARGET AUDIENCE
Enterprise e-commerce and retail brands (Experimentation/Personalization; want conversion and revenue lift); product and engineering teams (Flagship; want feature flags, gradual rollouts and server-side testing); marketing and digital-experience leaders (personalization; want tailored journeys without developer dependency); global multi-brand organizations (full platform; want governance across sites, apps and regions).
CUSTOMER BUYING BEHAVIOUR
The trigger is low conversion, slow release cycles, personalization ambitions or a desire to replace intuition with evidence. Marketing, product and engineering leaders evaluate through demos, proofs of concept, security review and traffic-based pricing. Adoption begins with web experimentation; expansion follows feature flags, app testing, recommendations and personalization.
| PRICING INTELLIGENCE
What makes this model effective & make customers pay
Customers are not buying an A/B testing tool; they are buying a system to make high-stakes digital decisions with evidence instead of opinion.
RULE 1 — PRICE THE DECISION SURFACE. Traffic and modules reflect how much digital experience the customer can test, personalize and govern.
RULE 2 — SELL REVENUE LIFT, NOT TEST COUNT. The buyer cares about conversion, average order value and retention improvements, not the number of experiments run.
RULE 3 — USE MODULES TO MATCH MATURITY. A customer may begin with testing, then add feature flags and personalization as its experimentation practice matures.
RULE 4 — WATCH THE DATA-QUALITY TRAP. Poor tracking, low traffic or biased experiments undermine trust; the vendor must invest in statistical rigor and implementation support.
The buyer is really buying faster, safer revenue decisions, proven by traffic served, experiments run and measurable conversion lift.
PRICE & REVENUE
| Revenue Risk - The biggest threat to revenue stability
Low experimentation maturity: Customers may buy the platform but run too few meaningful tests, weakening renewal and expansion.
Traffic-based price resistance: As traffic grows, pricing can rise sharply, prompting customers to seek self-serve or open-source alternatives.
Developer-led substitution: Engineering teams may prefer Statsig, LaunchDarkly, Optimizely or open-source experimentation tools over marketer-oriented suites.
Privacy and data constraints: Cookie restrictions, consent requirements and limited first-party data can reduce personalization effectiveness.
Acquisition integration risk: Epoq, Dotaki, Potions and Wandz.ai must integrate into one coherent platform; otherwise they become disconnected modules.
AB Tasty raised more than $63 million, including a $40 million Series C in 2020; its ARR, churn and valuation are not publicly disclosed or independently verified.
Where the model can break
4
MOTION
motion ge cs
AB Tasty converted consulting relationships, SEO and direct enterprise sales into a repeatable acquisition system. Liwio’s conversion clients provided early demand; content and search attracted experimentation buyers; direct sales expanded accounts into multi-module platforms. Growth from 25 customers in France in 2014 to more than 1,000 brands globally by 2025 showed that the channel operated beyond the founding team.
GROWTH EXPANSION MODEL
Platform Expansion, Acquisition-Led Growth
HOW THEY EXPAND
AB Tasty landed through web A/B testing, then expanded within accounts through feature flags, server-side experimentation, personalization, recommendations and AI-driven experiences. Platform expansion came through Flagship and the shared experimentation layer; acquisition-led growth added Epoq, Dotaki, Potions and Wandz.ai. The dependency is the customer’s experimentation practice: if the brand does not run meaningful tests, personalization and recommendations become disconnected add-ons.
COMPETITIVE STRATEGY
Differentiation
HOW THEY COMPETE
Against Optimizely and Adobe Target, AB Tasty differentiated through marketer-friendly experimentation and faster deployment. Against Google Optimize and free analytics, it offered enterprise governance, support and personalization. Against developer-native tools, it added visual testing and cross-functional workflows. Its advantage was serving both marketing autonomy and engineering control.
GROWTH ENGINE
GTM
Content Flywheel
A brand begins with web experimentation, sees conversion or revenue lift, then adds feature flags and personalization. More traffic and more digital properties increase the need for governance, recommendations and AI-driven experiences. Customer success and enterprise sales expand the account across modules and regions. The loop depends on experiment velocity, trustworthy results and cross-functional adoption; weak data quality or low traffic breaks it.
AB Tasty uses a sales-led, product-supported GTM. Content and SEO create awareness; direct sales sells experimentation programs to marketing and product leaders; customer success drives module expansion and enterprise rollout. The primary motion is sales-led because enterprise contracts require security, implementation and multi-stakeholder alignment, while product-led experimentation creates the usage evidence that supports expansion.
SUSTAINING MOATS
High Customer Lock-In
moat
The moat is AB Tasty’s accumulated experimentation and behavioral data: test results, audience responses, conversion patterns and personalization outcomes. Over time, this data can improve recommendations, audience models and decision guidance, making the platform more valuable than a standalone testing tool.
| MOAT INTELLIGENCE
THE STANDARD: A data advantage lasts when the platform captures decision outcomes across many customers and uses them to improve future recommendations.
RULE 1 — CAPTURE EXPERIMENT OUTCOMES, NOT JUST EVENTS. Test design, variant performance and business context create reusable decision intelligence.
RULE 2 — LINK PERSONALIZATION TO MEASURABLE LIFT. Recommendations and adaptive experiences must show incremental revenue or engagement, not merely more variation.
RULE 3 — GOVERN AI DECISIONS. Brand constraints, statistical rigor and rollback controls make AI-driven personalization trustworthy.
THE SIGNAL: AB Tasty served more than 1,000 brands and acquired Epoq, Dotaki, Potions and Wandz.ai between 2022 and 2025. The moat does not protect against low-traffic customers, poor data governance or developer-led teams choosing API-first alternatives.
Why this company remains defensible
ARR & TAKEAWAY
ARR Journey - what to do at each stage
PRE-$1M ARR — VALIDATE THE MARKETER WEDGE. Liwio’s consulting work revealed that marketers needed no-code experimentation; AB Tasty launched in 2011.
WATCH: Whether marketers could create reliable tests without engineering support.
$1–5M ARR — PROVE PRODUCT-MARKET FIT. AB Tasty closed its agency in 2012 and committed fully to the SaaS product, using early customers to refine the visual testing workflow.
$5–10M ARR — SCALE DIRECT SALES. The company moved from French early adopters to broader European and North American enterprise accounts.
$10–50M ARR — EXPAND INTO FEATURE MANAGEMENT. Flagship launched in 2019, adding server-side experimentation and feature flags for product and engineering teams.
$50–100M ARR — BUILD THE EXPERIENCE PLATFORM. Epoq, Dotaki and Potions acquisitions added recommendations, emotional segmentation and e-commerce personalization.
$100M+ ARR — NOT REACHED / NOT PUBLICLY EVIDENCED. AB Tasty serves more than 1,000 brands and has raised over $63 million, but its ARR is not publicly disclosed or independently verified.
Rule: Start with one decision-maker’s autonomy, then expand only when the platform earns the technical gatekeeper’s trust and creates measurable business outcomes.
COPY PLAYBOOK : What Worked → What Failed → What to Replicate → What to Avoid
THE STANDARD: Build an experimentation platform by productizing a repeated decision bottleneck, giving the business user autonomy, and earning technical trust before expanding into adjacent decision capabilities.
SEQUENCE:
1. Find a decision made by opinion rather than evidence. Use services, customer research or domain expertise to identify a recurring choice with measurable revenue or performance impact.
2. Give the non-technical owner controlled autonomy. Build the first workflow so the business user can act without creating technical or operational risk.
3. Expand into decision infrastructure only after adoption. Add flags, personalization or AI recommendations when experimentation becomes habitual and the customer has enough traffic to generate reliable results.
WORKED: AB Tasty used consulting insight to build marketer-led testing, then expanded into feature management and personalization as customer maturity increased.
CAUTION: Do not copy personalization before proving traffic density and data quality; low-volume experimentation produces noise, not lift. Do not rely on acquisitions without integration—disconnected modules weaken the platform’s decision value.
bottom of page