Behavioral conversion optimization

Stop spending A/B tests on weak ideas.

Your conversion pages have thousands of plausible variants. You can only test a few. We identify the ones most likely to move conversion before you build them, then give you a ranked shortlist to test first.

Request free diagnostic → See how it works
Peer-reviewed science Built by professors Calibrated on your buyers
Why us

We study the exact mistakes decision surfaces create.

Thomas Graeber

Thomas Graeber

Co-Founder · Professor, University of Zurich

I study how people choose when they are uncertain, cognitively overloaded, or forced to rely on shortcuts. That is what pricing pages and product decisions often trigger. Previously Professor at Harvard (2020–2025); now Director of the Center for the Cognitive Foundations of Economics in Zurich.

Research →
Christopher Roth

Christopher Roth

Co-Founder · Professor, University of Cologne

I study how framing, presentation, and narratives shift judgments and choices at scale. That is the backbone of how we generate and rank variants. Large-scale experiments on belief formation; previously at Oxford and Warwick. Named Top 40 Under 40 by Capital Magazine.

Research →
What you get

A ranked list of tests worth running next.

Not a brainstorm. Not a slide deck full of generic CRO advice. A practical test roadmap built around your actual decision surface.

01 / Ranked variants

Variants, not advice.

3–10 concrete variants to test first, prioritized by expected impact and confidence. Each one arrives with a behavioral rationale attached — you can defend the pick to your team, not just ship it and hope.

02 / Test order

A sequence, not a wishlist.

Which variants to run first, which to hold back, and where interaction effects matter. A ready-to-brief testing plan — specific enough to hand a designer or engineer directly.

03 / Turnaround

Fast, with an engine you keep.

First diagnostic in ~2 weeks. A full calibrated engagement in 6–8 weeks — and at the end you keep the Behavior Sandbox, our calibrated model you use on every future variant.

You send one screen. We return a sharper testing agenda.
The problem

Most teams are testing from a tiny corner of the space.

Take a pricing page. Change the tier names, the highlighted plan, the CTA, the anchor, the order, the copy, the default billing view — the page already explodes into thousands of plausible versions. Most teams test two or three.

— Your design space
every dot is a plausible version of your page
Our simulated candidates
we evaluate every variant in this space
Your current designThe version your customers see today.
What you A/B testA handful of variants, clustered around the current design.
Our ranked shortlistSpread across the whole space. Where the lift actually lives.
Rare lucky overlapSometimes what you test and what we'd rank coincide. Mostly it doesn't.
— Illustration · even a handful of simplified decisions already makes the design space too big to A/B test through
Tier structure count & naming pattern 8
Recommended tier badge, card size, segment cue 6
Price anchoring enterprise anchor, per-unit, discount frame 5
CTA wording generic vs. outcome-led, trial vs. paid 6
Trust-signal placement next to CTA, top of page, below fold 4
8 × 6 × 5 × 6 × 4 = 5,760 versions

Add a few more axes — headline copy, feature ordering, billing defaults — and you're already past anything an A/B testing program can cover in a year.

We narrow the space using behavioral economics and rank the variants worth testing before you A/B test them — so every test slot you spend goes to a variant worth running.
How it works

Research-backed, transparent at every step.

Every recommendation traces back to peer-reviewed science and calibration data we collected ourselves — and you can audit how we got there at every stage. You don't need to tell us what to test. We find it, and we show our work.

01 / Problem definition

What are we optimizing?

You tell us the decision your customers are making, what they're seeing, and the objective: conversion, revenue, upsell, retention. That's it — we don't need a list of variants from you.

02 / Pre-optimization

A free diagnostic of your touchpoint.

Send us a customer touchpoint and we'll run our diagnostic on it — a real demo of the method calibrated on comparable products. No commitment, no card.

Request free diagnostic → Free
03 / Calibration

Calibrated consumer twins.

We run our own behavioral data collection on your target customers — surveys and incentivized experiments we design and field ourselves. From that, we build calibrated simulated buyers specific to your product. You don't need to share user data.

04 / Perception measurement

How visitors actually experience your touchpoint.

We deploy a scientifically validated battery of perceptions — overwhelm, clarity, trust, attention paths, confusion, unmet needs and many more — built from our own published research. These perceptions are what predict behavior.

05 / Variant generation + simulation

Thousands of variants, ranked.

From the measured perceptions, we generate interventions individually targeted at the binding constraints on conversion — then rank them through cross-model simulation and targeted human experiments wherever simulation alone leaves real uncertainty.

06 / Delivery

Ranked roadmap + Behavior Sandbox.

A prioritized A/B-test roadmap with predicted uplift, confidence bounds, and behavioral rationale — plus your own Behavior Sandbox, the calibrated environment you keep using for every future variant.

The science behind

Most visitors leave your page unsure what's best for them.

>80%
of consumers report not knowing which option is right for them after visiting a product or pricing page. The problem isn't your product. It's the difficulty of the decision.

Difficulty comes in three forms — and each triggers a different shortcut. When you don't know which type is binding, you're optimizing blind.

— Three types of difficulty
01 — Understanding

Do they grasp what each feature means?

The customer doesn't understand the attributes well enough to evaluate them. When comprehension breaks down, people fall back on the simplest available shortcut — typically price, brand familiarity, or whatever requires the least interpretation.

02 — Self-knowledge

Do they know what they like, want and need?

The customer understands the options but can't map them to their own situation. They don't know which features they'll actually use, how much capacity they need, or which tier fits. This uncertainty drives systematic under-purchasing and deferral.

03 — Integration

Can they hold it all in their head?

The customer understands each attribute and knows what they want — but there are too many dimensions to compare at once. The comparison exceeds working memory. People stop weighing all the attributes and start relying on a single feature, a default, or a gut feeling.

What you keep

Your first engagement delivers the shortlist. The Behavior Sandbox is what you keep.

The calibrated model we build for your customers doesn't disappear when the first engagement ends. It becomes a Behavior Sandbox — a reusable environment your team feeds any future variant into, before you build, before you ship, before you burn an A/B slot.

New pricing ideas, onboarding changes, upgrade prompts, fresh copy directions. Same calibrated customer model, new inputs — for every future change.

Included in every full engagement
→

Test your own hypotheses

Drop in a screenshot, a mockup, or a text description. Your variant is scored alongside ours — on the same calibrated customer model, through the same pipeline. No more arguing over which idea deserves the next A/B slot.

→

Orders of magnitude more variants

Where a one-off diagnostic covers dozens, the sandbox evaluates thousands — tier framings, anchor architectures, segment cues, copy variants, and the interaction effects between them. Ship the top few.

→

Reuse it for every future change

New pricing structure next quarter. Onboarding rework. Feature launch. Run any of them through the same sandbox. The calibration stays valid; only the variants change.

Where it works

Any customer decision you want to convert on.

We focus on the high-stakes decision surfaces where behavioral friction most often explains conversion loss.

Pricing pages

Plan structure, tier framing, anchor design, highlighted tier, decoys, per-unit framing — conversion and revenue as separate objectives.

Upgrade flows

Free-to-paid conversion, feature bundling, commitment framing, upgrade triggers, CTA and social-proof placement.

Checkout & onboarding

Friction identification, trust-signal placement, defaults, permission requests, activation milestones, first-action framing.

We also work on landing pages, email flows, and feature announcements — any decision surface where behavioral friction costs you conversion.

Get started

Stop guessing. Test the right variants.

Send us a customer touchpoint. We'll run a free diagnostic and send back concrete changes worth testing first.

No commitment. Concrete suggestions in a few days.

✓

Request received.

We'll review your touchpoint and get back to you within a few days with a diagnostic.