Product growth
Design for conversion without designing against the user
A conversion is meaningful when a person understands the choice, can complete it and is still glad they did. Five design levers help teams improve that outcome responsibly.
Define the conversion as a valuable state change
“Increase conversion” is incomplete until the action and its consequence are named. A purchase, trial start, account opening and recurring investment are not interchangeable. Each has a different commitment, uncertainty, recovery path and business value.
Write the target as a state change: a qualified person moves from this starting state to this valuable completed state, with these conditions understood. Then name the failure states around it. This stops the team from optimising a click that merely moves confusion to the next screen.
Lever 1: Friction
Friction is effort that stands between intent and action. Remove repeated fields, unexplained prerequisites, unnecessary account creation and dead-end errors. Keep useful friction when it prevents harm: review before payment, confirmation before an irreversible change or a clear suitability boundary before a consequential financial action.
The question is not “Can this take fewer steps?” It is “Does each step help the person decide, prepare, act or recover?” A longer flow with visible progress and meaningful context can feel easier than a compressed flow full of surprise.
Lever 2: Comprehension
People cannot confidently choose what they cannot understand. Give the offer a clear hierarchy: what it is, who it is for, what it costs, what happens next and what could change. Move essential explanation next to the decision instead of hiding it in a generic help centre.
Comprehension is especially important when products use unfamiliar categories or domain language. Plain language does not mean removing necessary precision. It means introducing precision in the order a person needs it.
Lever 3: Trust
Trust comes from alignment between promise and behaviour. Show total cost before commitment. Explain why information is requested. Make cancellation, support and recovery findable. Use authentic evidence with source, timeframe and permission rather than decorative claims.
In fintech, explicit state and consequence are part of trust: which account, instrument, amount and order is selected; when money moves; what is pending; what can still be changed. In SaaS, trust includes data handling, permissions and a credible path out. In Ecommerce, it includes availability, delivery, returns and payment state.
Lever 4: Motivation
Motivation connects the action to progress the person already values. Show the useful outcome, not manufactured urgency. Demonstrate the next meaningful state and reduce uncertainty about reaching it.
A strong call to action is specific enough to predict the result: “Review order,” “Create workspace,” or “Choose delivery” usually supports a better decision than a generic “Continue.” Motivation should clarify value, never conceal cost or make refusal harder than acceptance.
Lever 5: Measurement
Instrument the full path before declaring a change successful. Record meaningful events and the context needed to interpret them. Check data quality, consent, duplicate firing and whether the event represents an actual state change rather than a rendered screen.
Measurement also needs qualitative evidence. Analytics can show where a pattern changed; observation and support evidence help explain why. A clean dashboard cannot repair an ambiguous event definition.
| Lever | Ecommerce | SaaS | Fintech |
|---|---|---|---|
| Friction | Guest checkout, useful defaults, recoverable payment | Progressive setup, relevant permissions, resumable onboarding | Prepared documents, visible progress, explicit review |
| Comprehension | Total price, delivery and returns before purchase | Value, limits and plan differences in working language | Instrument, risk, fees, timing and state in context |
| Trust | Authentic stock and fulfilment information | Data use, security boundaries and exit path | Source, consequence, consent and regulatory context |
| Motivation | Connect the item to a real use and available fulfilment | Show the first meaningful outcome, not an empty dashboard | Connect action to an understood goal without promising returns |
| Measurement | Item view through fulfilled order and return | Qualified signup through activation and retained value | Started journey through verified, completed and exception states |
Build a metric map, not a single target
| Measure | What it asks | Useful guardrail |
|---|---|---|
| Task completion | Can qualified people complete the intended job? | Completion without moderator or support intervention |
| Activation | Did the person reach the first meaningful product outcome? | Exclude empty account creation from value |
| Conversion | Did the defined state change occur? | Track cancellation, refund or reversal downstream |
| Error rate | Where does preventable failure interrupt progress? | Separate user correction from system failure |
| Time to value | How long until the promised usefulness becomes real? | Do not reward speed that removes necessary understanding |
| Assisted conversion | How often does completion require human help? | Watch support effort and unresolved contacts |
| Guardrail | What must not worsen as the target improves? | Regret, complaint, accessibility, failure, risk or unwanted commitment |
Diagnose the point of loss before redesigning it
From funnel symptom to verified improvement
Define the state change
Name the person, starting state, completed value and required understanding.
Map the journey and data
Connect interface moments, operational states, events, support contacts and known exits.
Identify the likely lever
Use behavioural and qualitative evidence to locate friction, comprehension, trust or motivation problems.
Change one coherent part
Design the smallest experience change capable of testing the hypothesis without introducing deception.
Read target and guardrails
Assess completion and value alongside errors, assistance, reversals and longer-term consequence.
Three examples of better questions
Ecommerce
Instead of “How do we make checkout shorter?”, ask: “Which information or interaction prevents a ready customer from understanding total cost and completing payment?” The answer may be address effort, delivery ambiguity, payment failure or a trust gap created earlier on the product page.
SaaS
Instead of “How do we increase signups?”, ask: “Can the right person reach the first useful shared outcome with the permissions and data they actually have?” A large signup number paired with empty workspaces is acquisition without activation.
Fintech
Instead of “How do we make investing feel effortless?”, ask: “Can a qualified person understand the instrument, consequence and selected state well enough to act without preventable error?” Consequential decisions should feel clear, not trivial.
The best conversion design preserves agency
Dark patterns may move a local metric by hiding cost, preselecting consent, manufacturing urgency or making refusal difficult. They also change the nature of the conversion: the system records an action that does not represent informed intent.
Good conversion design reduces avoidable effort while increasing clarity and control. It helps the right person take the right action, understand what happened and recover when reality changes. That is a stronger foundation for growth than a funnel that succeeds only by narrowing what the user can see.
Sources and further reading
- Google Analytics: Measure ecommerceOfficial implementation reference for measuring ecommerce actions and item-level context.
- GOV.UK Service Manual: Usability benchmarkingGuidance for measuring task success, time, error and experience across a service.
- U.S. Federal Trade Commission: Bringing Dark Patterns to LightExamples of interface practices that can impair choice or manipulate action.