How to Write About What You Do Without Inventing Statistics

"Studies show that 87 per cent of customers check reviews before buying." Sentences like that turn up in first drafts constantly: from freelancers, from AI tools, and from your own keyboard late in the evening. Almost none of them survive the obvious follow-up question. Which studies?

Learning how to write content without statistics is not about lowering your standards. It is about building authority from things you can actually stand behind: your process, worked examples, honest ranges, and the occasional properly cited figure. This piece covers all four, plus a quick fact-check routine for stripping unsupportable claims out of any draft before it goes live.

Why suspiciously round numbers keep appearing in drafts

Invented statistics happen for a boring reason: numbers feel authoritative and are easy to type. A writer under deadline pressure reaches for "70 per cent of small businesses" because it sounds researched, and an AI model does the same thing for the same underlying reason. It has seen thousands of articles shaped like that, so it produces one shaped like that.

The tells are consistent. Round percentages with no source. "Studies show" or "research suggests" with no study named. Figures that are oddly precise about things nobody measures, such as exactly how many minutes visitors spend deciding whether to trust a plumber. If a number could not plausibly have been counted by anyone, it was not.

The cost is higher than it used to be. Readers have become sceptical of unsourced claims, and AI assistants such as ChatGPT and Perplexity increasingly quote content directly, which means a made-up figure on your site can end up repeated as fact with your name attached. If part of your goal is being cited by those tools, the standards in this article matter twice over. There is more on that in our guide to answer engine optimisation.

How to write content without statistics

The good news is that most business content never needed statistics in the first place. It needed evidence, and evidence comes in several forms that you already own.

Show your process

Nothing signals expertise like describing exactly how you do the work. An accountant explaining the specific order in which they review a new client's books, an IT firm walking through what actually happens in the first hour after a ransomware call, a decorator listing what they check before quoting a hallway: these read as authoritative because only a practitioner could write them.

Process writing also happens to be what searchers want. Someone Googling "how does an EPC assessment work" wants the steps, not a percentage. Write the steps in order, note where things commonly go wrong, and say what the reader can prepare in advance. That is a genuinely useful article and it contains no data at all.

Use worked examples

A worked example is a specific, concrete scenario carried through from start to finish. Not a fabricated case study with an invented client name, which breaks the same honesty rule as an invented percentage, but a hypothetical clearly framed as one: "Suppose a ten-person office is paying for fifteen Microsoft 365 licences. Here is how you would find and remove the spares."

The difference between this and a fake case study is transparency. "Imagine", "suppose" and "say you run" tell the reader it is illustrative. "One of our clients, a Manchester law firm," when no such client exists, is a lie, and it is the kind of lie that gets noticed.

Give ranges and honest conditions

Real experts rarely give single figures, because real answers depend on circumstances. "A rewire on a three-bed semi usually lands somewhere between X and Y, and the biggest variable is access" is more credible than a false-precision quote, and it is the answer a genuine tradesperson would give across the kitchen table.

The same applies to "it depends" answers generally. Saying "it depends, and here are the three things it depends on" is not a dodge. It is the most trustworthy sentence structure in business writing, provided you actually list the three things. Readers can smell the difference between a hedge and a map.

Borrow authority honestly

You can also point at things that are verifiably true without owning the data yourself. "Google's own documentation says" is fine if you link to it. "HMRC's deadline for this is" is fine because anyone can check it. Verifiable public facts are not statistics you invented; they are references, and references make you look careful rather than clever.

When you do use a figure, cite it properly

Sometimes a number genuinely earns its place. When it does, three rules keep it defensible.

1. Name the source in the text, not just in a link. "According to the ONS" carries the weight; a bare hyperlink under the word "research" does not.
2. Link to the original, not to another blog quoting it. Second-hand statistics mutate. A figure about one country becomes global, a 2019 number becomes current, and by the fourth retelling nobody can find where it started.
3. Date it. "A 2023 survey found" ages honestly. An undated figure looks current forever, which eventually means it looks wrong.

If a figure fails any of these tests, you have two options: find the original and cite it, or delete it and describe the pattern instead. "Many small firms overpay for software licences they no longer use" needs no source because it claims nothing measurable. It is weaker than a real statistic, but infinitely stronger than a fake one.

A five-minute fact-check for any draft

Before publishing anything, human-written or AI-drafted, run this pass:

• Search the draft for "%", "per cent" and every digit. Each one either gets a named, linked, dated source or gets rewritten as a pattern or a range.
• Search for "studies", "research", "survey" and "experts". If no specific study, survey or expert follows, the sentence is decoration. Cut or rewrite it.
• Check every named example. Real clients need permission; hypotheticals need framing words that mark them as hypothetical.
• Check superlatives. "The most common cause" and "the fastest way" are claims. If you know them from doing the work, keep them. If they arrived in an AI draft, verify or soften them.
• Read the intro and conclusion twice. That is where drafts reach hardest for authority, and where fabrications cluster.

This matters more, not less, if you use AI tools for drafting. Google's guidelines focus on whether content is helpful and accurate rather than how it was produced, a point covered in more detail in what Google actually says about AI-written content, but accuracy is your responsibility either way. The byline on the page is yours.

A statistic you cannot defend is not evidence. It is a liability with a percentage sign on it.

Where Helio SEO fits in

Helio SEO writes articles that follow these rules by design: no invented figures, patterns and ranges where numbers cannot be verified, and reasoning you can inspect in the dashboard rather than take on trust. If you write your own content, the fact-check routine above is all you need, and it costs nothing but five careful minutes per draft.