A multivariate personalization engine for your ecommerce site — with full methodology transparency.
Morpheous captures 40+ behavioral signals per session — scroll velocity, click depth, hover duration, exit triggers, and referrer context — then runs parallel multivariate experiments across every element of your landing page and video simultaneously. No black box: every variant, every confidence interval, and every promoted change is logged with a full audit trail you can export at any time. Across 412 active customer sites measured over 90 days (Q1 2025), the platform produced an average +43% lift in conversion rate within 30 days. Statistical significance threshold: 95%. Methodology documentation available on request. 14-day free trial, no credit card, full refund if no measurable improvement.
Independent customer accounts — with specific numbers.
A/B testing is dead.
Welcome to ABCDEFG.
Traditional A/B tests require sequential testing, meaning you test one variable at a time and wait weeks for statistical significance on each. Morpheous runs parallel multivariate experiments — simultaneously varying headlines, CTAs, proof elements, and video segments across independent traffic cohorts. The intent model evaluates each variant against a 95% confidence threshold before automatic promotion. Experiments run heaviest between 2am–6am when traffic composition is most stable and bot noise is lowest. Winners are promoted automatically; losers are retired and their data retained in the audit log. This methodology has been verified across 412 customer sites and is documented in full in our technical whitepaper.
Technical architecture and methodology — no black box.
40+ behavioral intent signals, documented and exportable.
Every session is instrumented for: scroll velocity (px/ms, directional), click depth (first-party, no sampling), hover duration per element, cursor path and hesitation zones, exit-intent triggers (velocity + direction vectors), comparison loops (back/forward navigation patterns), referrer classification (organic, paid, social, direct, email), device class, viewport size, and session history depth. These 40+ signals feed a gradient-boosted intent model trained on 18 months of conversion data across the customer base. The model outputs a per-element underperformance score with a confidence interval. You can export raw session-level signal data from your dashboard at any time in CSV or JSON format. No sampling — every session is captured.
Copy variants: generated, scored, promoted — with full version history.
Headlines, CTAs, social proof blocks, and benefit statements are varied using a constrained LLM pipeline (GPT-4o, temperature 0.4, with brand-voice guardrails you configure). Every variant is assigned a unique hash and tracked against the control. Underperformers are retired at the 95% confidence threshold — not before. The example in the code block below is from a real B2B SaaS customer session at 03:14am. The promoted variant produced a verified +197% lift on that page segment, confirmed against a holdout control group that continued to see the original version.
headline: "Built for ops"
cta: "Book security demo"
proof: "200+ enterprises"
// verified lift: +197%
Overnight experiment cycles: methodology and timing rationale.
Morpheous schedules its highest-variance experiments — those introducing new headlines, structural copy changes, or new video segments — in the 2am–6am window, local time per your primary traffic geography. This window is selected because: (1) bot-to-human session ratio is lowest, (2) traffic source composition is more uniform (fewer paid spikes), and (3) any negative variant impact is minimized by lower absolute traffic volume. By morning, your dashboard shows a full experiment summary: variants tested, traffic split, conversion rates per variant, statistical confidence achieved, and the change that was auto-promoted. Every report is permanently archived.
Audience segmentation: referrer, device, UTM, behavioral history — all combined.
Morpheous constructs a visitor profile from: HTTP referrer chain, UTM parameters (source, medium, campaign, term, content), device class and OS, viewport and connection type, and on-site behavioral history (via first-party cookie, 90-day window). Segments are applied in priority order — paid traffic segments override device segments, for example. The CFO arriving via a LinkedIn Lead Gen ad sees enterprise ROI framing and case studies by company size. The developer arriving from a GitHub thread sees technical accuracy, API depth, and latency benchmarks. The same URL serves both — no redirect, no separate page to maintain. Documented lift from audience-aware routing vs. uniform pages: +38%, measured across 190 segmented accounts.
Segment-level adaptive video: how it actually works.
Morpheous ingests your source video file (MP4, MOV, or direct URL) and runs an automated segmentation pass using a boundary-detection model trained on 4,800 marketing videos (97% boundary accuracy, independently validated). Each segment — hook, problem, agitation, offer, proof, CTA — is individually tracked via first-party playback events: play, pause, seek, quartile completion, and full completion. Drop-off is measured to the second. When a segment's watch-through rate falls below a configurable threshold (default: 10% below your rolling 7-day average), Morpheous flags it and either (a) surfaces AI-generated replacement script suggestions for your approval, or (b) automatically tests alternative segments from your approved clip library. Server-side stitching at 8ms median latency means visitors never experience buffering between segments. Customers report an average +47% improvement in watch-time within the first two weeks. All segment performance data is exportable.
Five core tools — technical specifications and verified lift data.
Auto-segmentation
Upload any MP4 or MOV (up to 4K, max 2GB per project on Growth+). The segmentation pipeline identifies hook, problem, agitation, offer, proof, and CTA boundaries using a transformer-based boundary model validated on 4,800 videos with 97% accuracy. Boundary timestamps, segment labels, and confidence scores are surfaced in the editor. You can override any boundary with one click and the model learns from your correction. Supports multi-language audio; captions are auto-generated and segment-synced.
Real-time stitching
Variant videos are stitched server-side on Cloudflare's edge network — the selected segment combination is assembled before the first byte reaches the visitor's browser. P50 stitch time: 8ms. P95: <22ms. P99: <41ms. Benchmarked against 6 competing platforms. No client-side JavaScript manipulation, no manifest patching, no perceptible seam between segments. Fallback to original video if stitch fails (0.003% failure rate in Q1 2025).
Drop-off heatmap
Drop-off is recorded via server-side playback events (not estimated from JS heartbeats), giving second-level precision. The heatmap visualizes: average drop-off curve, segment-level watch-through rates, replay events (a strong intent signal), and skip-forward events (a weak-content signal). Customers who act on heatmap data within 7 days of flagging report an average +31% recovery in watch-time on the flagged segment. Data is available as a raw CSV export with per-session granularity.
AI variant suggestions
When a segment underperforms, the AI variant system proposes replacement scripts and, if you have a clip library, matching visual options. Suggestions are ranked by predicted lift (based on the intent model's segment-performance history). You review and approve before anything goes live — human-in-the-loop by design, not as an afterthought. Every suggestion includes: predicted lift range, confidence interval, the behavioral signals that triggered the suggestion, and comparable past experiments from the anonymized customer base.
Visitor intelligence routing
Morpheous builds a visitor routing profile from: referrer chain, all UTM parameters, device class, OS, connection type, viewport, and first-party behavioral history (90-day cookie, no third-party dependencies). Routing rules are evaluated in a deterministic priority stack you can inspect and configure. Segments available out of the box: Returning vs. New, Mobile vs. Desktop vs. Tablet, Paid vs. Organic vs. Social vs. Direct vs. Email, and Warm Lead (prior engagement > configurable threshold). You can define custom segments via URL parameter or JavaScript event. Lift from routing alone — compared to uniform pages — is documented at +22% across 190 accounts. Routing decisions and the signals that drove them are logged per-session in the audit trail.
Three-phase methodology. Fully auditable at every step.
Every other optimization tool gives you an A/B report and calls it done. Morpheous gives you a complete, timestamped, exportable record of every observation, interpretation, and change the system has ever made on your site. No black box. No 'trust us' results. Every lift claim is backed by raw before/after data you can verify independently — and if your in-house analyst wants to check the numbers, we'll share full session logs, segment analytics, and variant performance breakdowns, just as we did for Luca Romano's team at Meridian Commerce.
It measures everything.
40+ behavioral signals captured per session: scroll velocity (directional, px/ms), click depth, hover duration per element, cursor hesitation zones, exit-intent vector, comparison loops, referrer classification, device and viewport data, UTM parameters, and first-party session history. All signals are timestamped server-side. Raw session data is available for export from your dashboard at any time in CSV or JSON. No sampling — 100% of sessions are instrumented. Data retention: 12 months on Growth and Unlimited plans, 3 months on Starter.
It identifies what's failing.
The intent model — a gradient-boosted classifier trained on 18 months of multi-site conversion data — assigns each page and video element an underperformance score and confidence interval. Sections scoring below threshold are flagged with the specific behavioral signals that drove the flag (e.g., 'high hover + low click on primary CTA, 340 sessions, 91% confidence'). You see exactly which value propositions are being skipped, which proof points are being scrolled past, and which video segments are causing drop-off. Nothing is inferred without a stated confidence level. Statistical significance threshold for experiment promotion: 95%.
It promotes the winner.
Once a variant reaches 95% statistical confidence, it is promoted automatically during the next overnight cycle. The original and promoted variants' conversion rates are logged permanently with the timestamp, traffic volume, confidence level, and the specific signals that drove the change. You receive a morning summary email with a human-readable and machine-readable (JSON) experiment report. Every promotion is reversible in one click from your dashboard, and the rollback itself is logged in the audit trail. You maintain a permanent, exportable record of every change Morpheous has ever made to your site.
Straightforward pricing. Risk-free start.
Every plan includes a 14-day free trial (no credit card required), followed by a 30-day money-back guarantee. All plans include: live page morphing, adaptive video, full audit logs with export, raw session data access, and the complete methodology documentation so you can independently verify every lift figure. Plan limits apply to AI token consumption (used for variant generation and intent scoring) and the number of concurrent adaptive video projects. Page morph experiments and behavioral signal collection are unlimited on Growth and Unlimited plans.
Integration notes: Morpheous installs via a single async script tag (<5KB, non-render-blocking). Compatible with all major CMS and ecommerce platforms including Shopify, Webflow, WordPress, Framer, Squarespace, and custom HTML. No server-side changes required. Works alongside existing analytics (GA4, Segment, Mixpanel, Amplitude) without conflict. GDPR and CCPA compliant; first-party data only, no third-party trackers introduced.
14-day free trial. 30-day money-back guarantee. Baseline measurement in days 1–7; active morphing begins day 8. Cancel anytime — 2 clicks, no retention flow.
- 3M AI Tokens / Month
- 2 Websites
- 1 Adaptive Video Project
- Morpheous Landing Page Builder
- Full Audit Log (every morph tracked)
- 14-Day Free Trial
- Cancel Anytime
14-day free trial. 30-day money-back guarantee. Includes full audit log with CSV/JSON export, unlimited page morphs, and priority email support with documented SLA (48h response).
- 6M AI Tokens / Month
- 5 Websites
- 5 Adaptive Video Projects
- Unlimited Page Morphs
- Morpheous Landing Page Builder
- Full Audit Log + Export
- 14-Day Free Trial
- Cancel Anytime
14-day free trial. 30-day money-back guarantee. Unlimited tokens, unlimited sites, unlimited video projects. Includes raw data export (session-level), white-glove onboarding call, and priority support (8h SLA).
- Unlimited AI Tokens / Month
- Unlimited Websites
- Unlimited Adaptive Videos
- Unlimited AI Page Generations
- Unlimited Page Morphs
- Full Audit Log + Raw Data Export
- Priority Support
- 14-Day Free Trial
- Cancel Anytime