Fyxer.com A/B Reconnaissance Report — Competitive Intelligence by Revise Engine
Report Details
- Target
- https://www.fyxer.com/
- Date
- March 8, 2026
- Pages probed
- 7
- Status
- Complete — HIGH confidence
About the Target
Fyxer is a YC-backed AI executive assistant for email and meetings. It drafts replies in your voice, summarises meetings, manages your inbox, and joins calls as a note-taker. 100,000+ users. The product connects to Gmail, Outlook, and Google/Outlook Calendar.
Detected Tools
PostHog EU
Confidence: VERY HIGH (score 170). eu.i.posthog.com, EU data centre. Signals: POST /flags/ endpoint, window.posthog object, ph_ cookie prefix.
GrowthBook
Confidence: VERY HIGH (score 145). cdn.growthbook.io, public CDN. Signals: GET /api/features/, SSE /sub/ subscription endpoint.
Architecture
The two tools are deliberately split by surface. PostHog EU covers the Webflow marketing site with 138 anonymous feature flags, analytics and surveys. GrowthBook covers the React product app with 498 features and 37 active post-login experiments including the acquisition funnel.
Geo Routing & VPN Analysis
No geo-routing detected. Identical tool stack, cookie count (32), and content across all US regions tested: generic US, New York, Los Angeles, Chicago. No location-based price differentiation. HTML content and CTAs match across sessions.
Experiment Overview
- 138 PostHog flags active for anonymous visitors
- 37 GrowthBook experiments currently running
- 277 GrowthBook force/conditional feature rules
- 7 pages probed (/, /pricing, /sales, /teams, and more)
- 5 pricing variants in active test ($40–$65 per user per month)
- 4 churn recovery variants being tested
- 3 simultaneous referral program experiments
Most Significant Experiments
#1 US Base Price MVT — GrowthBook, Acquisition
The most commercially significant experiment in this scan. Fyxer is running a 5-way price discovery test to find the revenue-maximising price point before scaling paid acquisition. Five variants at 20% each: $40, $45, $50, $55, and $65 per user per month. Bucketed by organisation_id so entire teams see the same price, preventing a split experience within a customer's org.
#2 Inbox Zero Fake Door — GrowthBook, Product Discovery
A demand validation test for an Inbox Zero feature that does not yet exist. Users are shown an upsell at one of three price points; clicking measures intent, not purchasing. Variants: control with no upsell (25%), $79/mo (25%), $149/mo (25%), $249/mo (25%).
#3 Activated User Conversion — GrowthBook, Retention
What converts an activated-but-not-paying user after trial? Four concurrent recovery strategies each at 25%: control (no intervention), trial extension (more time), discount offer (price reduction), combination (extension plus discount).
#4 Speak-to-Sales CTA MVT — PostHog, Acquisition
4-variant CTA test on the /sales book-a-demo page. Fyxer is likely testing headline copy, button text, or form layout variants. The /sales page H1 is "Speak to a Fyxer expert".
#5 Autocomplete Model — GrowthBook, Infrastructure
gpt-4.1-mini versus gpt-5-mini, 50/50 split, live on all users. Measuring the quality-vs-cost tradeoff directly on production traffic. Outcome likely drives model selection for cost savings.
Pricing & Checkout Experiment Cluster
Fyxer is running an unusually dense cluster of simultaneous pricing experiments. All flags below were active for the same anonymous visitor in the same session.
- free-trial-cta: control — CTA design for free trial
- three-month-billing-page: test — 3-month billing option
- checkout-modal: test — Modal vs full-page checkout
- promo-sign-up: test — Promotional offer at signup
- free-fyxer-for-3-months: test — 3-months-free offer
- 14-day-trial-for-work: control — 14-day trial for work emails
- 14-day-trial-for-personal: test — 14-day trial for personal email
- billing-cycle-blocks: control — Block-style billing toggle
- default-to-monthly: control — Default to monthly billing
- toggled-other-plans: test — Show/hide other plans
- anchor-to-coffee: control — "Less than a coffee" framing
- minimal-special-offer: control — Minimal offer UI variant
- offer-countdown: control — Countdown timer on offer
- skip-special-offer: test — Skip the offer entirely
- signup-page-ux: control — Full signup UX redesign
GrowthBook: 498 Features Breakdown
The 498 features span every layer of the product. Active experiments are noted per category.
- Meeting / Calendar: 104 flags, 8 experiments
- Pricing / Billing: 76 flags, 5 experiments
- UI / UX: 74 flags, 6 experiments
- Onboarding: 54 flags, 4 experiments
- Draft / Email Writing: 48 flags, 3 experiments
- Other: 43 flags, 2 experiments
- Sales / Demo: 26 flags, 5 experiments
- Referral / Invite: 19 flags, 2 experiments
- Recording: 16 flags
- Integration / OAuth: 16 flags
- AI Model / Infra: 9 flags, 2 experiments
- Search / Knowledge: 6 flags
- Churn / Retention: 3 flags
- Notifications: 4 flags
Key signals from the flag inventory
- Meeting/Calendar is 21% of all flags — the notetaker is Fyxer's biggest product bet with 104 flags indicating aggressive iteration.
- non-us-base-price-mvt exists alongside us-base-price-mvt — separate price discovery tests running in parallel for US vs international markets.
- usage-based-pricing-fakedoor — a second fake door test measuring appetite for usage-based pricing as an alternative to per-seat, running alongside inbox-zero-fake-door.
- rebrand — a full visual rebrand is in active A/B test. Fyxer may be preparing to change its brand identity in production.
- ramp-dupe-lp — a duplicate landing page specifically targeting Ramp customers, likely for paid search or ABM campaigns.
- physics-hero, flow-inspired-landing, before-after-hero — three distinct hero section concepts in simultaneous test. Flow's landing page was directly cloned and A/B tested against Fyxer's own design.
- chatgpt-app-banner — a banner targeting visitors arriving from ChatGPT, suggesting paid or organic ChatGPT plugin acquisition is being tested.
Notable Observations
Churn Flow Is Fully Instrumented But Not Yet Active
Five simultaneous churn flags (churn-recovery-experiment, churn-time-saving, churn-downgrade, churn-discount-offer, profile-churn-recovery) are all set to control. The framework is in place — variants are about to launch.
SENT_LABELS_V2 Is Nearly Fully Rolled Out
GrowthBook shows a 5/95 split — 95% of users are in treatment. This feature has won and will reach 100% imminently.
Three Simultaneous Referral Experiments
referral-credit-reward (variant A), referral-flash-sale, and referral-link-in-summary are all running in parallel. Referral acquisition is a clear strategic priority.
Org-Level Price Bucketing Signals B2B Focus
The us-base-price-mvt uses organisation_id not user_id as the bucketing key — a deliberate choice to prevent teammates in the same company seeing different prices. Classic B2B SaaS.
Fake Door Before Build
inbox-zero-fake-door is textbook lean validation. Rather than building a full Inbox Zero product, Fyxer is exposing a subset of users to pricing screens and measuring click-through before committing engineering resources.
About Revise Engine
Revise Engine is an experiment on your experiments. An internal tool designed to scan, intercept and analyze experiments. Always learning.