OPTIMIZE · NAVIGATOR AI

Know what to test and why.

Navigator AI ranks every idea in your backlog by predicted lift, effort, and confidence — grounded in your actual site data and what you learn from each experiment you run. Every ranking shows its reasoning, so you ship the tests that move the number, not the loudest ones.

Watch 2-min demo ▸

◉ STAT-SIG MATH BUILT IN◉ NULL HYPOTHESIS + 95% CI◉ GROUNDED IN YOUR DATA

NAVIGATOR · RECOMMENDED HEADINGS3 / 12

◉ PRICING

Add annual billing toggle to pricing

+18%

LIFT

94% CONF

◉ SOCIAL PROOF

Move testimonial above fold

+11%

LIFT

88% CONF

◉ ACTIVATION

Simplify onboarding to 3 steps

+24%

LIFT

76% CONF

How it works Capabilities Technical specs Integrations

THE PROBLEM

Most teams sit on hundreds of test ideas and ship the loud ones.

Ideas pile up in docs, analysts triage in spreadsheets, and the highest-impact test waits behind a ticket that never reaches the top. Navigator replaces that whole chain with a ranked queue that updates as your data does — so the right test is always at the top, and you can see exactly why.

◆ HOW IT WORKS

From blank backlog to winning test in four moves.

TOTAL TIME · ~5 MIN

STEP 01

Connect your data

Plug in Matomo or GA4. Navigator indexes the last 90 days of behavior, revenue, and funnel events.

~30s

STEP 02

Navigator scores everything

Every page, component, and funnel step gets an opportunity score: traffic × drop-off × lift potential, blended with your competitor data and industry research.

RUNS HOURLY

STEP 03

Ranked queue updates

Ideas are ordered by expected lift, confidence, and effort. Every ranking ships with a reasoning trace showing why it sits where it does.

LIFT × CONFIDENCE × EFFORT

STEP 04

Ship with one click

Launch inside Flight Deck. The first draft of the variant comes with it.

ONE-CLICK LAUNCH

◆ CAPABILITIES

Everything Navigator needs to actually be right.

DATA INTAKE

It sees what you see — and what you don't.

Navigator reads from your analytics stack, session data, support logs, and commercial systems. Nothing gets ranked from a vacuum, and every claim links back to the event-level evidence.

Analytics intake via Matomo (primary) and GA4
Automatic event schema detection — normalized across sources
PII scrubbed at ingestion
Incremental syncs; 90-day backfill by default

PRODUCT VISUAL — PLACEHOLDER

REASONING

Every ranking comes with its receipts.

Navigator doesn't just spit out ideas — it shows you the observed signal, the inferred cause, and the proposed test. Audit the reasoning, disagree with it, or push it live.

Transparent reasoning trace per idea: observation → hypothesis → proposal
Grounded in your data — every claim links to the underlying events and segments
Challenge mode: ask "why does this outrank that?" and get a diff explanation
Feedback loop (Growth+): accept, reject, or edit ideas and Navigator learns your team's taste

PRODUCT VISUAL — PLACEHOLDER

PREDICTION

Forecast the lift before you burn the traffic.

Bayesian priors built from your own site's variance and the results of experiments you run give you a calibrated expected-lift interval for every idea in the queue.

Expected lift with 95% credible interval — not just a point estimate
Required sample size + runtime estimate per idea, based on your traffic
Power analysis built in — no more underpowered tests shipped to prod
Post-experiment recalibration feeds priors back into the ranking model

PRODUCT VISUAL — PLACEHOLDER

EVIDENCE LIBRARY

No hallucination research.

Every claim in the research library is cited to a crawled URL. When Navigator says "teams in your category that added annual billing saw X% lift," there's a real citation behind it — not a made-up statistic.

Citation-backed research documents (Growth+)
Case studies and benchmarks tagged by vertical
Competitive patterns extracted from Radar, not scraped from generic blogs
Deduped so you never see the same idea twice

PRODUCT VISUAL — PLACEHOLDER

HANDOFF

Ships into the tools you already run.

Navigator doesn't lock up your experiment spec. Launch in Flight Deck, or drop a fully-scoped ticket into Jira or Asana.

One-click launch in Flight Deck
Jira and Asana task creation with acceptance criteria and a link back to the source
MCP server means Claude or any MCP-capable agent can read the ranked queue and launch experiments
Full audit log — who ranked what, who shipped what, who killed what

PRODUCT VISUAL — PLACEHOLDER

◆ TECHNICAL SPECIFICATIONS

Under the hood.

AVAILABILITY

Recommendations on all paid tiers; the learning feedback loop and citation-backed research library are on Growth+

MODEL

Multi-LLM router with per-recommendation model attribution

DATA WINDOW

Default 90 days; configurable up to 2 years on Performance

SCORING METHOD

Ranked by predicted lift × confidence × effort — override any factor weight

FLIGHT PATH ROUTING

Recommendations are pre-filtered and re-ranked by Flight Path along the Traffic↔CRO axis, so low-traffic sites get growth work instead of unwinnable tests

CONFIDENCE

95% Bayesian credible intervals, configurable

DEDUPLICATION

Similar ideas are deduped so they never repeat across cycles