Why published web design statistics are usually unsourced

A handful of statistics — a percentage for how fast users judge a website, a multiplier for how much slow pages cost in lost revenue, a figure for how many visitors leave after one bad experience, an ecommerce cart-abandonment percentage — appear on thousands of web design and marketing agency sites, phrased almost identically, with the citation quietly dropped somewhere along the way. Understanding why this happens, and how to tell a sourced figure from an unsourced one, is worth more than any individual number.

How a figure becomes untraceable

A statistic typically starts from something real — a study, a survey, an internal analytics finding from one company. It gets cited once, correctly, with attribution. A second site quotes the first site rather than the original source, and drops a detail in doing so — the sample size, the year, the specific claim being measured. A third site quotes the second, and the figure is now several citations removed from anything checkable. Repeated across years and thousands of sites, the number survives; the source does not. By the time it appears on an agency’s homepage as “studies show,” there is frequently no way to verify it was ever measured the way it’s now being used, or that the original context still applies.

Why the web design industry keeps repeating unsourced statistics

Precise-sounding statistics are persuasive regardless of whether they’re sourced — “73%” reads as more credible than “most,” even when neither is actually verifiable in context. There is also no real cost to using an unsourced figure: nobody checks, and a competitor using the same number provides cover, because a claim that everyone repeats starts to feel established rather than unverified. This is a structural incentive problem in marketing content generally, not something specific to web design, but it shows up with particular density in how confidently designers describe user behaviour to prospective clients.

What this site does instead, and why it’s slower

Every figure published on this site states its source, the tool that produced it, and the date it was measured, recorded in the research files documented on this site’s methodology. Where a claim could not be traced to something checkable, it is not published — including plausible-sounding claims that would be easy to include and hard for a reader to challenge. This is measurably slower to write than repeating an industry-standard figure, and it is the only version of this content worth publishing, because an unsourced number a reader cannot check is not information, however confidently it’s phrased.

How to spot an unsourced web design statistic yourself

Ask, of any number in marketing content: what was actually measured, by whom, in what year, on what sample? If the answer traces to another marketing page rather than a named, dated, checkable source, treat the figure as unverified regardless of how many sites repeat it or how precise it sounds. A number stated to two decimal places is not more credible than a round one — precision is not the same thing as sourcing.

Why this matters more in the conversion and performance clusters than elsewhere

Claims about how website speed, design or trust signals affect visitor behaviour are the single densest concentration of unsourced statistics in this industry, because they’re the numbers agencies most want to use to justify a sale. This site’s own content on those subjects deliberately avoids invented percentages for exactly that reason — a discipline documented in the conversion guides and performance guides.

What to do when a supplier quotes you a statistic

Ask them directly where it comes from. A supplier who can point to a specific, checkable source is demonstrating real rigour. A supplier who cannot, or who is visibly surprised by the question, is repeating something they picked up from elsewhere without checking it themselves — which is useful information about how carefully the rest of their claims have been checked too.

Why AI assistants make this problem worse, not better

An AI assistant summarising or answering a question about web design draws on the same pool of published content this problem describes, including the unsourced statistics circulating across thousands of sites. If an assistant’s training or retrieval process doesn’t distinguish a sourced claim from an unsourced one repeated widely, it can present a fabricated or long-untraceable figure with the same apparent confidence as a genuinely verified one — potentially amplifying an unsourced claim’s apparent authority rather than correcting it. This is a further reason sourcing discipline matters more, not less, as AI-generated content becomes a larger share of what people read.

A specific historical example: the web design statistics behind the first-impression claim

Widely circulated claims about exactly how many milliseconds it takes to form a first impression of a website trace back, when investigated, to a single small academic study measuring something more specific and narrower than the claim as commonly repeated — and the specific number attached to it has drifted across years of citation. This is not to say the underlying research area is worthless; it is to say the number as it circulates in marketing content today is a poor representation of what was actually studied, which is the pattern this entire page describes.

What responsible use of a real statistic looks like

Where this site does cite a number — a measured search volume, a legislative date, a specific WCAG requirement — it states the source, the date, and the method, precisely so a reader can independently verify or challenge it. This is the standard any statistic should be held to, on this site or anywhere else, and it is a genuinely low bar that an enormous share of published web design content fails to clear.

What to do next

Treat any unsourced statistic in web design marketing — including, if you ever find one, on this site — as a claim to verify rather than a fact to accept. The full discipline behind every figure this site does publish is set out on how this site measures things.

The same standard applies to every claim made about a specific business type across the industries section — a fact stated there is either checkable or it is not published.