Does better presentation make a product more credible?
It is the assumption behind every design tool, including ours. We went looking for the evidence. Most of the statistics people quote for it turn out to be wrong.
A disclosure. Popshot makes screenshots look better. We have an obvious interest in the answer being yes. That is exactly why every figure here is traced to its original paper, why the sample sizes are printed next to the findings, and why the results that undercut our position are in the body rather than the footnotes.
The short answer
Presentation reliably changes what people believe about a product: how credible it seems, how easy it looks to use, how legitimate the team behind it feels. That effect is real, and it has been measured repeatedly for thirty years.
What the evidence does not support is the stronger claim usually bolted onto it: that better presentation makes a product actually work better, or that it reliably produces more sales. When researchers measured performance directly, the effect disappeared. And the best evidence on commercial outcomes suggests polish can backfire when it oversells what you are actually shipping.
So the honest version of the claim is narrower than the marketing version. It is still the version that matters, because credibility is a belief, and beliefs decide whether someone tries your product at all.
What the research actually found
Identical content, different presentation
The closest study to the question we care about is also the least famous. Robins and Holmes took the same written content and produced two versions of each page: one with a high visual treatment, one with a low one. Same words. Same information. Only the presentation changed.
Credibility ratings moved from −0.55 to +1.05 on a scale running −7 to +7. Participants were making these calls in about three and a half seconds.
Robins, D. & Holmes, J. (2008). “Aesthetics and credibility in web site design.” Information Processing & Management, 44(1), 386–399. Read the paperpaywalled
Caveat: 20 participants, all graduate students in one field, and only 7 of the 21 page pairs showed an individually significant difference. The authors describe the raw difference as “not large.”
Design is what people mention first
The most-cited study in this area asked 2,684 people to compare live websites and say why they trusted one over another. Across roughly 2,400 written comments, “design look” came up more than any other factor, appearing in 46.1% of them, ahead of information structure, accuracy, and company reputation.
That number is real. It is also routinely misquoted as “46% of people,” which is not what was measured, and the authors themselves asked readers not to treat it as precise. Their participants took part to earn a charity donation and had little reason to scrutinise the sites deeply. The authors predicted that a more motivated audience would lean on visual cues less.
Fogg, B.J., Soohoo, C., Danielson, D.R., Marable, L., Stanford, J. & Tauber, E.R. (2003). “How Do Users Evaluate the Credibility of Web Sites? A Study with Over 2,500 Participants.” Proceedings of DUX ’03. Stanford Persuasive Technology Lab / Consumer WebWatch. Read the paperfree
Now 23 years old, with no direct replication, and predating mobile entirely.
Professional photography moved a real market
The strongest causal evidence comes from outside software. Researchers studied 7,423 Airbnb properties over 16 months and found that listings which adopted verified professional photography saw occupancy rise 8.98%.
Zhang, S., Lee, D., Singh, P.V. & Srinivasan, K. (2022). “What Makes a Good Image? Airbnb Demand Analytics Leveraging Interpretable Image Features.” Management Science, 68(8), 5644–5666. Read the paperpaywalled
This is the peer-reviewed figure. Several larger numbers circulate from superseded drafts.
Where the case gets weaker
This is the part most articles on this subject leave out, and it is the part that changes what you should actually do.
Belief moved. Performance did not.
The best-known experiment on aesthetics and usability gave people an ATM interface at three levels of visual quality and two levels of actual usability, then had them complete eleven tasks. Aesthetics strongly predicted how usable people said the system was, both before and after using it.
It had no effect whatsoever on how long tasks took. At p = .987, that is about as flat as a result gets. Meanwhile the genuinely slower interface added roughly fourteen seconds per task, and people barely noticed. Their opinion of the system tracked how it looked, not how it behaved.
Tractinsky, N., Katz, A.S. & Ikar, D. (2000). “What is beautiful is usable.” Interacting with Computers, 13(2), 127–145. Read the paperpaywalledfree copy
And the effect did not replicate cleanly
Two well-controlled studies later found no aesthetic effect on perceived usability at all, and found the relationship running the other way: interfaces that actually worked well made people rate the design as more attractive. Another followed people for two weeks and watched the effect disappear entirely by the second week, concluding that single-session testing had been “grossly overestimating” it. A review of the field put it plainly: the relationship between beauty and usability “has been overplayed.”
The most recent meta-analysis pools 31 studies and nearly 19,000 people. It finds a modest positive effect overall. But when it isolates the studies that manipulated aesthetics deliberately rather than measuring it after the fact, the effect very nearly vanishes.
Polish can cost you
The Airbnb result has a sequel, from the same research team, and it is the finding we found hardest to sit with. High-quality images bring people in. But when the property does not live up to the impression the photographs created, reviews suffer, and that damage outweighs the initial gain.
Their conclusion is not “polish more.” It is that image quality should match the thing being sold. Presentation that runs ahead of the product borrows credibility it then has to pay back.
Zhang, S., Mehta, N., Singh, P.V. & Srinivasan, K. (2023). “Do Lower-Quality Images Lead to Higher Demand on Airbnb?” Working paper, not yet peer-reviewed. Read the paperfree
Changing a belief is not changing a behaviour
A randomised controlled trial gave 85 women the same web page about breast cancer risk from alcohol, dressed with either trust-signalling or trust-undermining design cues. The cues shifted how credible people rated the site. They had no measurable effect on how long anyone stayed, what they thought while reading, or what they said they would do next.
Harris, P.R., Sillence, E. & Briggs, P. (2009). “The Effect of Credibility-Related Design Cues on Responses to a Web-Based Message About the Breast Cancer Risks From Alcohol.” Journal of Medical Internet Research, 11(3), e37. Read the paperfree
Six numbers you should stop repeating
Every one of these appears in design and marketing writing constantly. None of them survives being traced to a source.
| The claim | What is wrong with it |
|---|---|
| “75% of users judge a company’s credibility on its website design.” Usually attributed to Stanford. | Not a Stanford finding. It traces back to a marketing agency’s blog post with no study behind it. |
| “46% of people judge credibility on visual design.” | Right study, wrong unit. It was 46.1% of the written comments participants left, not 46% of people. |
| “You have 50 milliseconds to establish credibility.” | That experiment measured visual appeal, not credibility. The study that did test trust at 50ms found people agreed with their own earlier judgement about half the time. A coin flip. |
| “Beautiful designs are more usable.” | The paper this comes from measured task completion time directly and found no effect at all (p = .987). It found a change in what people believed, not in what they achieved. |
| “67% of consumers say image quality is very important to their purchase decision.” | Traces to an infographic that discloses no sample size, no method, no date, and no author. It is the most-repeated number in this space and nobody can say where it came from. |
| “Professional photos earn Airbnb hosts $2,521 a year.” | From a superseded draft. The figure does not appear in the peer-reviewed paper that was eventually published. |
The 50-millisecond finding is Lindgaard, G., Fernandes, G., Dudek, C. & Brown, J. (2006), Behaviour & Information Technology, 25(2), 115–126. It measured visual appeal on 40 participants. Read the paperpaywalled
The trust-at-50ms result is Albert, W., Gribbons, W. & Almadas, J. (2009), Proceedings of the Human Factors and Ergonomics Society, 53(6). The other figures in this table are unsourceable by design: that is the point of listing them.
Nobody has studied this for software
Here is the thing we did not expect to find. There is no peer-reviewed research on the presentation of software screenshots at all. Not on app store listings, not on landing page visuals, not on product demos.
Everything credible in this literature is about photographs of physical things: rental properties, second-hand goods, retail products. And those all work through a mechanism that does not obviously carry over: a photograph reduces your uncertainty about an object you cannot touch. A screenshot is different. The interface is the product, and in most cases you can go try it yourself in thirty seconds.
Whether presentation matters more for software, less, or differently is an open question. Every statistic we found claiming to answer it came from a company selling screenshot tools, with no methodology attached. Including, we should say, companies rather like us.
What we take from this
A raw screenshot is not neutral. People form an impression of it fast, that impression attaches to the product behind it, and presentation is the loudest signal available when someone knows nothing else about you. That is well-supported, and it is the reason Popshot exists.
But we would rather make the modest claim accurately than the ambitious one loosely. Framing your screenshot will not make your product better. It will not, on its own, sell it. What it does is stop a good product from reading as an unfinished one. For an indie maker with no brand recognition to fall back on, that gap is worth closing.
The one thing we would take seriously from the research is the Airbnb result. Presentation that oversells what you have built does not just fail to help; it borrows credibility you then have to repay in reviews. Make the screenshot look like the product on its best honest day. Not like a different product.
Try it on your own screenshot
Free, no sign-in, and nothing leaves your browser.
Open Popshot — freeWhere people usually start: Product Hunt galleries, App Store screenshots, device mockups, or everything else.