Reference
Marketing claims, checked
Marketing runs on statistics that nobody has looked up. This is a short list of the ones we meet most often, each followed back to whatever is actually at the end of the citation trail, with the answer reported honestly including when the claim holds up.
In short
Several widely quoted marketing statistics cannot be traced to any published study. The claim that SEO leads close at 14.6% and outbound at 1.7% is attributed to at least four different publishers and no primary research. The marketing rule of seven has no original source; earlier authorities said twenty and the founding research said three. The widely cited five-minute lead response finding is real but dates from 2007, covered six companies, and reports odds ratios rather than probabilities.
| Claim | Verdict |
|---|---|
| SEO leads close at 14.6%, outbound leads at 1.7% Read the check | Unsourced. Widely repeated, circularly cited. |
| Respond within 5 minutes and you are 21x more likely to qualify a lead Read the check | Real study. Routinely misquoted. |
| The average UK cost per lead for [your trade] is X Read the check | No such published source exists. |
| It takes seven touchpoints before somebody buys Read the check | Unsourced. Older authorities said twenty; the research said three. |
| Businesses make two pounds for every one spent on Google Ads Read the check | Sourced, but the source is Google, about Google. |
| The average website conversion rate for your industry is X% Read the check | Mostly US data, and measuring the wrong event. |
Checked by following the citation trail by hand, most recently on 20 September 2026. Where a claim is described as unsourced, that means no primary study with stated methodology could be found behind it, not that the underlying number has been shown to be wrong.
The checks
SEO leads close at 14.6%, outbound leads at 1.7%
Unsourced. Widely repeated, circularly cited.
If you have read anything about SEO in the last decade you have met this pair of numbers. They are used to justify budgets, win pitches and end arguments. They are also, as far as we can est…
Respond within 5 minutes and you are 21x more likely to qualify a lead
Real study. Routinely misquoted.
This one is different from most marketing statistics, because there genuinely is a study underneath it. The problem is not invention. It is that almost nobody quoting it mentions the date, t…
The average UK cost per lead for [your trade] is X
No such published source exists.
This is the claim we get asked about most, usually in the form of a competitor's proposal with a confident table in it. The table is not necessarily dishonest. It is almost never what the re…
It takes seven touchpoints before somebody buys
Unsourced. Older authorities said twenty; the research said three.
This one decides how long your follow-up sequence is, how many times a prospect gets chased, and how much retargeting budget gets signed off. It is also, as far as anybody can establish, a n…
Businesses make two pounds for every one spent on Google Ads
Sourced, but the source is Google, about Google.
Unlike most figures on this list, this one has a traceable origin. The problem is what the origin is, and what an average of every advertiser on earth can tell a scaffolding firm in Hampshir…
The average website conversion rate for your industry is X%
Mostly US data, and measuring the wrong event.
Conversion rate benchmarks fail twice over for a trade business. They come from a different country, and they count an event most trade websites do not have.…
How we check
The method is unglamorous. Take the claim as it is usually worded, search for it, and open the pages that rank. For each one, find the attribution and follow it. Repeat until either a primary source appears or the chain starts going in circles.
Three outcomes are possible, and we try to keep them separate rather than collapsing everything into “myth”:
- Sourced and fairly quoted. The claim traces to a study, and what people say about it matches what the study found.
- Sourced and badly quoted. There is a real study, but the date, sample, geography or type of statistic gets dropped in the retelling. The five-minute rule is the clearest example.
- Unsourced. The chain leads from one secondary page to another and never arrives anywhere. This does not prove the number is false. It means nobody quoting it can tell you why it is true.
Where our own reading could be wrong, the page says so. If you can point us at a primary source we missed, we would genuinely rather know, and the page gets corrected with the date of the correction on it.
Holding a proposal full of numbers?Send it over. We will tell you which figures are sourced, which are American, and which are nobody's, including when the answer is that they are fine.
Call 07443 392243WhatsAppQuestions
Why does a marketing agency publish this?
Because we are asked to compete against these numbers. A prospect arrives holding a proposal built on statistics that cannot be sourced, and the choice is to either quote something equally unsupportable back or to explain where the figures come from. This is the second option, written down once so it does not have to be repeated on every call.
Do you check claims that turn out to be fine?
Yes, and we say so when they are. The five-minute lead response finding comes from a real study with a real sample; the problem is how it is quoted rather than whether it exists. Marking everything as false would be its own kind of dishonesty.
Can I suggest a claim to check?
Call or message the number on this site with the claim and where you saw it. If there is something substantial to say about it, it gets a page and you are welcome to disagree with the conclusion.
Is this fact-checking?
It is citation-checking, which is narrower and worth distinguishing. We follow a claim back to whatever is at the end of the trail and report what is there. Where nothing is there we say the claim is unsourced, which is not the same as saying the number is wrong.
Related
Also worth reading
- Which sources AI assistants actually citeWhere an unsourced statistic goes once an assistant repeats it.
Talk to a human
We would rather be checked than believed
Every number on this site names its source. Call and argue with any of them.