Making income verification understandable
As a Senior Content Designer in ClearScore, I worked on our income verification launch, including conducting the user testing, synthesised it and turned the findings into a multi-variant experiment on the loans results page.
The team's working hypothesis was a simple 'income confirmed' tag would work for users. Commercial pressure wanted it to ship quickly. I made the case for testing with users first, and the findings changed the experiment: we tested a plain English subtitle alongside the tag, and the subtitle won on both metrics.
It's now live for 100% of UK loans users, and the research I conducted set the direction not only debunked our hypothesis, but it set the next phase of certainty work in the marketplace.
+8.7%
+2.2%
+£80k
Apply clicks on income verified offers
Apply clicks across all UK loans offers
Incremental revenue increase per month
The problem
Income verification was one of the main reasons lenders declined a loan application after the user had already applied, and roughly 1 in 10 users were declined downstream even on pre-approved offers because the lenders had deemed they failed an income check. This is because the lender couldn't prove the user's income is what they stated on the application.
However some lenders could already verify a user's income before they applied, through open banking or through their own technology. On those offers, the chance of acceptance was far higher. But our results page didn't show it.
The commercial objective was clear: prove that labelling verified offers increased clicks and conversion on those specific loans, so we'd have a case to persuade more lenders to verify income up front. However the risk is that if we explain this badly we would confuse people, clutter an already dense results page, or make pre-approval and Triple Lock look weaker by comparison.
What we knew and what we assumed
We knew which loan offers were income verified: a flag in the backend data told us. We also knew from earlier user research that certainty was a major factor when users chose a loan - users wanted to know they would be approved for a loan they applied for.
What we didn't know was how to say it on the page. Our squad's working hypothesis was a tag, 'Income confirmed', on the specific loan offer card. It was quick to build and fitted the existing component library. I was sceptical. A tag tells users something has happened. It doesn't tell them why it matters to them and the loans results page already carried a lot of tags competing for attention.
That idea is what the user research aimed to answer.
My approach: test understanding before testing conversion
I ran an unmoderated study on Dscout in January 2026. Around 12 participants, all actively looking for a loan, on and off ClearScore. Each person saw the loans results page with a different income verification treatment: a tag, a descriptive subtitle or an explainer at the top of the page with a banner on the offer detail page.
They then worked through tasks where I varied price and income verification to see which one won.
"His user testing on income verification was also a strong piece of work. It gave us valuable insight into how users think about certainty, helped support the decision to roll out the subtitle to all users, and has helped shape how we're thinking about the next algo piece."
Alice Pinch, Senior Product Manager
An example of the content ideation rounds that I designed in Figma, exploring different UI components and copy.
Drafting the user research objective brief and setting up the usability test in Dscout.
Four findings came out of it. Two overturned our squad's hypothesis.
-
92% of participants understood what 'income confirmed' meant, but around a third couldn't pinpoint exactly how verification worked. Understanding a label isn't the same as trusting it. The treatment had to explain the benefit, not just name the status.
-
Only 1 in 4 noticed any treatment without being prompted. Visibility was a bigger problem than wording. A subtle tag was always going to struggle.
-
Nearly everyone scrolled past the top of page explainer. Banner blindness meant people did not stop to read it.
-
Certainty is still the most important factor for loans. Price wins everything else. When two offers were priced the same, every participant who understood the label picked the income confirmed one. When the verified offer was more expensive, 5 of 6 took the cheaper one.
I wrote the recommendation that shaped the experiment afterwards. We should test the tag and the descriptive subtitle against control, drop the explainer, and keep algo ranking changes out of scope so we could isolate the content effect.
Key decisions and what I ruled out
Ruled out by research: a top-of-page explainer
While it would have given us room to explain income verification, three in four participants scrolled straight past it.
Ruled out in ideation: an explanatory paragraph on the tile
It explained the benefit but pushed the APR and monthly repayment down the page, which are the two things users scan first.
Tested and lost: the tag
This was our squad's lead hypothesis. I argued not to lead with this, and pushed for user testing. It finished at +1.7% overall and +1.0% on verified offers against the subtitle's +2.2% and +8.7%.
Decided: a descriptive subtitle
One line, in plain English, saying why this offer was more certain. It explains the benefit at the point of decision without adding a new component to the loan offer tile.
What we shipped
We went with a multi-variant test between the tag and subtitle, with the control as the measurable baseline, although we were certain that the subtitle treatment would perform best as it did in the user testing. It sits under the lender name and tells the user, in one line, that this offer is more certain and why.
The research shaped every part of it. Users understood the label but not the mechanism, so the line explains the benefit rather than naming a status. Only a quarter noticed the tag treatment, so it sits in the reading path, directly under the lender name, rather than as a badge in the corner.
Control variant
This also overturned our initial hypothesis that the simple tag would perform best, showing the value in doing user research.
Variant A (Tag treatment)
The team's original hypothesis: 'Income confirmed' as a badge on the offer tile.
Variant B (Subtitle treatment)
The variant the research pointed to. It overturned our starting assumption that the tag would win.
Results
The subtitle beat both the control and the tag, on both metrics.
The experiment ran for four weeks, between March and April 2026, on 20% of UK loans traffic. We then proceeded to roll out the subtitle variant to 100% of UK loans traffic when an offer is verified.
This gave us our first clear evidence of how users weigh certainty against price, and that shaped the follow on algorithm rollout on how verified offers are ranked on the results page for users.
It became the foundation for the Future of Certainty work: a higher-certainty tier above Triple Lock, aiming for over 95% acceptance. I led the design thinking on that, and ran the unmoderated testing in August 2026 that found 'Boosted' meant nothing to users on first contact while 'higher chance of approval' was understood by everyone who saw it
The designs were featured at ClearScore's mid-year partner event since it was one of the flagship features we rolled out in 2026.
"From the initial research into income confirmed you ensured this was a 'content first' design which is exactly what it needed to be. Off the back of that your designs/research influenced the direction of the experimentation, which in turn was adopted in our marketplace for a 2.2 increase in apply clicks - one of the most innovative things that's happened on the marketplace this year"
Peter Astbury
Head of Design (line manager)
What I learned
The most useful thing the research did was kill a design option. The explainer looked like the obvious way to explain verification but three in four users never saw it.
Finding that out in a cheap Dscout study rather than a four week experiment in production test is the whole argument that we should testing comprehension before conversion. As a content designer, this was instinctive for me.
I went in thinking certainty was the top priority. It isn't when it costs more. That changed how I write for the marketplace: the certainty signal has to support the offer, not compete with the rate.