They are not reading your website. Not at first, and possibly not at all.
Research published in Behaviour & Information Technology found that people form a reliable judgement of a web page's visual appeal in about 50 milliseconds — and that the judgement made in 50 milliseconds correlates closely with the one made in 500. That is a twentieth of a second, which is less time than it takes to read one word. Through the halo effect, that snap verdict then colours how everything afterwards is interpreted.
Stanford's web credibility work found roughly 75% of users judge a company's credibility on the basis of its website design.
Those two findings get quoted a great deal, usually to sell design work, and usually with the wrong conclusion attached. The conclusion is not that your site needs to be beautiful.
Here is what is actually happening in those eight seconds.
0.05 seconds — is this a real business
Before any content is processed, one thing gets decided: does this look like somewhere that is still operating.
This is not an aesthetic judgement, it is an inference. A visitor has no information about you, so they read the site as evidence of the business behind it. Text overlapping an image, a layout that jumps on a phone, a stock photo of an unrelated office, a copyright date three years old — none of those are opinions about design. They are read as signals that nobody is minding this, and by extension, that nobody may mind your job either.
The bar is much lower than designers imply. It is not impressive. It is maintained.
2 seconds — am I in the right place
Now the visitor looks for themselves in your page. Two questions, both binary, both answered from the top of the screen:
Do you do the specific thing I need? Not "outdoor living solutions". Composite decking. Cedar fencing. The words they typed.
Do you do it where I am? This is the one small business sites get wrong most often, and it is the cheapest to fix. "Serving the greater metro area" fails, because the visitor cannot tell whether their town is in it, and a visitor who cannot tell assumes not. Name the towns.
Fail either and they are gone, and you will never know they were there. This accounts for far more lost enquiries than any design problem, and it costs nothing to fix — it is a sentence.
5 seconds — is any of this true
Assuming they are still reading, the visitor starts looking for evidence, and they are surprisingly good at it. They have been lied to by websites before.
What reads as real: photographs of your work, not stock. A named person. An address, or a clearly defined service area. Testimonials with full attribution — a name, a business, a town — rather than "great service! — J.M." Licence or certification numbers, which almost nobody bothers to display and which are checkable. Specific numbers instead of adjectives: fourteen projects, since 2009, most jobs in four days.
What reads as false, immediately: stock photography of models in hard hats. "Trusted by hundreds of satisfied customers." Five-star testimonials with no attribution. Any claim to be the best.
The rule underneath all of it: a specific claim is evidence and a general claim is noise. "We pride ourselves on quality workmanship" is not a weaker version of a good claim. It is not a claim at all — every competitor's site says it, so it carries no information, and a visitor's eyes slide off it entirely.
Missing contact information is the sharpest red flag of the set. A business that is hard to contact reads as a business that does not want to be found, and nobody investigates further to find out why.
8 seconds — what happens if I get in touch
This is the decision, and it is the one nearly every website ignores.
Nobody is deciding whether you are the best. They are deciding whether contacting you is going to be a mistake.
Think about what the visitor is actually weighing. If they fill in that form or ring that number, what happens? Do they get a call back within the hour, or nothing for four days? Do they end up in a sales conversation they cannot easily leave? Will they be asked to explain their whole situation before finding out whether you even handle this kind of work? Will they be quoted a number that turns into a different number later?
Every one of those is a cost, and none of them are money. They are the reasons a person who has already decided they need you closes the tab and thinks I'll deal with this later. Which they usually do — with whoever made it easiest.
Which means the highest-return content on a small business website is almost never another paragraph of persuasion. It is the paragraph that says what happens next:
Send us what you have — even a photo and a rough idea is enough. You will get a reply within a day with a straight answer about whether this is something we do, and roughly what it costs. If it is not a fit, we will say so and point you somewhere better.
That paragraph removes four separate fears in three sentences, and it costs nothing to write. It also does something persuasion cannot: it makes the next step feel small.
What this means for how much you spend
The awkward implication of all of the above is that the expensive parts of a website are frequently not the parts doing the work.
Almost everything described here — naming your towns, using your own photographs, attributing testimonials properly, stating what happens after someone gets in touch, showing your contact details — is writing and honesty, not design or engineering. A modest page that does all of it will out-convert an elaborate one that does none of it, and it will do so on the same traffic.
Which is not an argument that design does not matter. The 50-millisecond judgement is real and it happens before any of your writing is read. But the bar it sets is coherent and cared-for, and that bar is met well below the price point most businesses assume.
So the honest sequence is: clear the visual bar, then spend everything else on specificity and on making contact feel safe. If you are trying to work out what that should cost, the real price bands are here — and none of the four questions above are in the expensive tier.
Worth saying plainly: all of this assumes someone reached your site at all. If enquiries are thin rather than unconverted, the problem may be further upstream — whether search and AI assistants can find you is a different diagnosis entirely.
A test worth doing
Open your own site on your phone, and read only what is visible before scrolling. Then answer, as a stranger:
- What does this business do, exactly?
- Would they come to my town?
- Is there any evidence here that is checkable?
- If I get in touch, what happens, and how long will it take?
Most sites answer one of the four. If yours answers three, you are ahead of nearly all of your competitors, and no amount of redesign will beat that.
Our Starter Website is $695, one-time: one designed page written from facts you confirm, with your services and service area named properly, your proof placed next to the claims it supports, a contact path that says what happens next, and the search groundwork done at build time. Launched on a domain registered in your name.
It exists precisely because of the argument above — most small businesses do not need a bigger website. They need the four questions answered on one page that loads quickly and looks like somebody is home.
Sources
- Lindgaard, Fernandes, Dudek and Brown (2006) — Attention web designers: You have 50 milliseconds to make a good first impression!, Behaviour & Information Technology 25(2), 115–126. Publisher listing. Source of the 50-millisecond finding and of the correlation between judgements made at 50ms and at 500ms.
- Stanford Web Credibility Research (Fogg et al.) is the origin of the widely quoted finding that around 75% of users judge a company's credibility on the basis of its website design. It is cited far more often than it is linked; treat the figure as a well-attested finding from that programme rather than a precisely reproduced statistic.
- The trust-signal observations — real photographs over stock, attributed testimonials, visible contact details, missing contact information as a red flag — are consistent across web-credibility literature and current practitioner guidance rather than drawn from one study.
Checked 2026-07-31.