How FlowAudit reviews your flows
Every FlowAudit report is a structured heuristic review: we evaluate one flow, against one goal, using a defined set of UX and conversion criteria — then rank what we find by how much it's likely costing you. This page lays out exactly which criteria we apply, how severity is assigned, where the model came from, and what it can't see.
This page is maintained by Poplab OÜ, the team behind FlowAudit. It describes our own method and is not an independent certification or third-party audit of our accuracy.
The four core lenses
Whatever flow you submit, every finding traces back to one of four questions. They're deliberately few — a long checklist produces long reports, not shippable ones.
Friction
Every extra step, field, decision, or moment of hesitation between the visitor and the outcome they came for.
- Form field count and whether each field earns its place
- Error handling, validation timing, and recovery paths
- Distractions and competing actions inside a committed flow
- Mobile-specific friction: tap targets, keyboard types, scroll depth
Clarity
Whether a first-time visitor can tell what this is, what it costs, and what happens next — without re-reading.
- Value proposition and headline hierarchy above the fold
- Plan differentiation and feature-comparison legibility on pricing
- Cognitive load, visual noise, and information hierarchy
- Progress indication and orientation across multi-step flows
Trust
Whether the page gives enough reason to believe before it asks for an email, a card, or a commitment.
- Trust signals near the point of commitment: guarantees, security, policies
- Social proof — logos, quotes, numbers — and whether it's specific or generic
- Objection handling: pricing surprises, hidden fees, cancellation terms
- Dark patterns and anything that reads as manipulative
Conversion alignment
Whether the page is actually built to produce the outcome you told us you want, or is quietly optimizing for something else.
- Primary CTA visibility, prominence, and copy
- Alignment of every element with the stated goal and success metric
- Feedback and confirmation after the user acts
- Time-to-first-value in onboarding and first-run experiences
Criteria change with the flow
A checkout page and an onboarding sequence fail in different ways, so they're not graded against the same list. When you pick a flow type, the review switches to the criteria set for that flow. These are the exact criteria in use today.
| Flow type | Criteria applied |
|---|---|
| Homepage / landing | Value proposition clarity, headline hierarchy, CTA visibility and copy, trust signals, social proof, cognitive load, visual noise, above-the-fold completeness |
| Pricing | Plan differentiation, CTA clarity per plan, anchoring logic, feature-comparison legibility, trust signals (guarantees, logos), objection handling, upgrade path clarity |
| Signup | Form field minimalism, friction reduction, value reinforcement near the form, trust signals, social proof, error handling and validation clarity, CTA copy, social-login options |
| Checkout | Form field minimalism, trust signals (SSL, payment logos), error recovery, CTA copy, distraction removal, mobile friction points |
| Onboarding | Time-to-first-value, progress indication, cognitive load per step, empty-state guidance, error messaging quality, goal alignment per screen |
| Dashboard / first run | First-run empty-state guidance, time-to-first-value, orientation and wayfinding, clear next action, information hierarchy, progressive disclosure, feedback and confirmation |
| Auto-detect / other | Value proposition clarity, primary CTA visibility and copy, trust signals, social proof, navigation and wayfinding, cognitive load, visual hierarchy, friction reduction, feedback and confirmation, goal alignment |
How the model was developed
The model is expert-authored. It starts from established usability and conversion principles — Nielsen-style usability heuristics, the long-standing research literature on form friction, trust, and cognitive load — and narrows them to the handful that actually move outcomes in SaaS acquisition and activation flows.
The specific criteria and severity rules come out of Poplab's hands-on audit practice: we ran this review manually for client after client before turning it into a product, and the flow-specific criteria above are the ones that repeatedly produced fixes teams actually shipped. Since launch, we review real FlowAudit output and tighten the model when it produces findings that are vague, duplicative, or not worth a sprint slot.
To be precise about the claim: this is a structured expert methodology, refined in practice. We have not run a controlled accuracy study, and we don't publish an accuracy percentage — anyone quoting one for a heuristic UX review is guessing. What we can promise is that the criteria are fixed, disclosed on this page, and applied the same way to every audit.
How severity is assigned
Severity is scored against your stated goal, not against a generic notion of good design. The same issue can be a P0 on a checkout and a P2 on a marketing page.
P0
Critical
Directly blocks or severely damages the stated outcome. If a visitor can plausibly abandon here, or the page actively misleads them, it's a P0. Fix these first.
P1
Important
Meaningfully hurts performance but isn't a hard blocker — unclear copy, weak trust placement, avoidable friction. Worth a slot in the next sprint.
P2
Polish
Optimizations, refinements, and experiments. Real improvements, but they shouldn't jump the queue ahead of a P0.
What this method can't tell you
- We don't use your analytics. No session data, funnel numbers, or conversion rates feed the review unless you tell us about them.
- Any percentage in a report is a heuristic estimate expressed as a range — an informed judgment about likely impact, never a measured result or a promised lift.
- For flows behind a login, we can only review the screenshots you upload. Anything you don't show us, we don't assess.
- This is expert-style judgment applied to what's visible. It's a strong starting hypothesis for your backlog, not a substitute for testing on your own traffic.
- The reliable way to validate a fix is to ship it and re-run the audit on the same flow — the run-over-run comparison shows what actually changed.
One more scope note: we audit for human conversion. AI search optimization is a different problem we don't currently solve.
See the method applied to your own flow
No credit card required. Takes about 90 seconds.