Ask ChatGPT, Gemini or Google’s AI Mode a question and you get an answer, not a list of links. Behind that answer, the AI has read a lot of web pages, pulled out the few passages that actually answered the question, and quietly ignored the rest.
That is the shift. Your website is no longer only competing to rank. It is competing to be the passage an AI picks up, trusts and repeats, and most business websites were never written with that in mind.
Two simple writing ideas make a surprising difference here. Neither is new, neither is a trick, and both come from the world of plain, clear communication rather than from search engines. They are called BLUF and chunking, and this article explains what they are, why AI search rewards them, and why putting them into practice across a whole website is harder than it sounds.
Why the way you write now matters
An AI answer engine does not read your website the way a customer does. It does not land on your homepage, admire the design and work its way through the menu. It breaks pages into passages, works out what each passage is about, and keeps the ones that look like clear, direct, trustworthy answers to the question in front of it.
Google describes its own AI features in similar terms: its systems can identify the different topics on a page and surface the part that is relevant, and it recommends content that is organised into clear paragraphs and sections with headings that make the structure easy to follow. (Google Search Central)
The practical consequence is that a page can be accurate, well designed and genuinely helpful, and still be passed over, because the answer is buried in paragraph six, wrapped in a preamble, or split across three sections that only make sense together. The AI is not being unfair. It is doing what any busy reader does: looking for the bit that answers the question and moving on.
BLUF: bottom line up front
BLUF stands for “bottom line up front”. It is a writing rule from the US Army, whose correspondence regulation tells officers to put the main point at the beginning of a message rather than building up to it. (Wikipedia) The idea has since spread to business writing, intelligence briefings and medical handovers, anywhere a reader needs the answer before the reasoning.
On a web page, BLUF means the first sentence or two under a heading give the answer. The explanation, the caveats and the background come afterwards, for the reader who wants them.
Most business websites do the opposite. A page headed “How much does a boiler service cost?” opens with two paragraphs about why servicing matters, a note about the company’s twenty years of experience, and only then, if at all, a price. A human reader skims past the preamble with mild irritation. An AI answer engine, looking for a passage that directly answers the question, may simply not find one.
Why BLUF works for AI search:
- The answer and the question sit together. When the first sentence under a heading answers that heading, the passage is self-evidently relevant. There is nothing for the AI to infer.
- It survives being quoted. AI tools lift sentences out of context. A sentence that states the point plainly still makes sense on its own. A sentence that says “as we explained above, this is why it matters” does not.
- It reads as confident. Hedged, meandering openings look uncertain. Direct ones look like expertise, and AI systems are weighing which sources to trust.
BLUF is not about being blunt or short. The full explanation still belongs on the page. It is about order: conclusion first, then support.
Chunking: every section has to stand on its own
When an AI system processes a web page, it does not hold the whole page in mind. It splits the page into smaller pieces, often called chunks, and works with those. Different AI tools split pages in different ways, and you have no control over where the cuts fall. What you can control is whether each piece still makes sense after it has been cut out.
That is what chunking means as a writing discipline: organising a page so that every section is self-contained. Each section covers one topic, sits under a heading that says plainly what it is about, and can be read cold, without the paragraphs before or after it, and still be understood.
A simple test: pick any section of a page on your website and read it on its own. If it opens with “this”, “these” or “as mentioned above”, if it never says what product, service or place it is talking about, or if it only makes sense because of something explained two sections earlier, it will not travel well. An AI that lifts that section gets a fragment, and fragments do not get cited.
Three habits make sections self-contained:
- One heading, one topic. A heading like “Boiler servicing in Brentwood: cost, timing and what is included” tells both a reader and a machine exactly what the section answers. “More information” tells nobody anything.
- Name things. Write “our emergency plumbing service covers Brentwood, Billericay and Shenfield”, not “we cover the local area”. Specific names of services, places, products and people are what an AI uses to work out what a passage is about and who it belongs to.
- Finish the thought. If a section needs a price, a timescale or a condition to be useful, put it in that section, even if it also appears elsewhere on the page.
A word of caution. Chunking is not the same as chopping. Google is clear that there is no need to break content into tiny pieces for AI, and that its systems can find the relevant part of a longer page on their own. (Google Search Central) The goal is not lots of short fragments. It is a page whose sections are each complete, clearly labelled and written for the person reading them. That is good writing by any standard, and it happens to be exactly what AI search can use.
What it looks like on a real page
Here is a section from a typical service page, before and after. The business and the figures are invented, but the pattern will be familiar.
Before
Our approach
We know that choosing the right people to look after your heating is a big decision. With over twenty years in the trade, our friendly, fully qualified team take pride in doing the job properly, and we have built our reputation on word of mouth. Whatever you need, we are here to help, so get in touch today to find out more about how we can assist.
It is pleasant and it is probably true. But what question does it answer? There is no service named, no place, no price, no timescale, and nothing an AI could quote in reply to “how much is a boiler service near me?” or “who services Worcester boilers in Brentwood?”.
After
Boiler servicing in Brentwood: cost and what is included
A standard gas boiler service from Smith Heating costs £85 and takes about an hour. It includes a flue gas test, a check of all seals and controls, a clean of the main components and a written safety report. We are Gas Safe registered and service Worcester Bosch, Vaillant and Ideal boilers across Brentwood, Shenfield and Billericay, usually within five working days of booking.
Same business, same tone, roughly the same length. The difference is that the heading says what the section is about, the first sentence answers the obvious question, and every name a customer or an AI might search for is on the page. Read it on its own and it still makes complete sense. That is BLUF and chunking working together.
Notice what has not happened. The page is not shorter, not stuffed with keywords and not written for a robot. It is simply clearer, which is why the real customer reading it is better served too.
Why it is harder than it sounds
Rewriting one section is easy. Rewriting a website is not, and this is where most good intentions stall.
- You have to know which questions to answer. BLUF only works if the bottom line is the one your customers are actually asking for. That means finding out what people ask search engines and AI tools about your kind of business, in your area, in their words, and checking which of those questions your competitors already answer better than you do.
- It touches every page, not just the blog. Service pages, location pages, FAQs, the about page and the pricing page all need the same treatment. A site with one well-structured article and forty vague pages is still, on balance, vague.
- Consistency is the hard part. The names you use for your services, the places you cover and the people who do the work need to match everywhere: on the website, in Google Business Profile, in directories and in reviews. AI systems cross-check. Three different descriptions of the same service read as three different businesses.
- Structure has to survive the page builder. A section that is self-contained in a Word document can be scattered across accordions, tabs, sliders and pop-ups by the time it reaches the website, some of it invisible to crawlers altogether. The words and the build have to be planned together.
- None of it matters if the AI cannot get in. Hosting, security services and robots settings can block AI crawlers without anyone noticing. Well-written content behind a closed door is still invisible.
- It is easy to overdo. Chase the technique too hard and you end up with pages chopped into fragments, stuffed with place names, written for machines. Google says plainly that it ignores that kind of thing and treats the worst of it as spam. The skill is in writing clearly for people in a way that machines can also use.
The businesses that get this right usually have someone who understands all three parts at once: what customers ask, how to write the answer, and how the website is built and served. That is rarely one person inside a small business, and it is not a job a page builder or an AI writing tool can do on its own.
How DotScott helps
DotScott works with small and medium-sized businesses to join the dots between what their customers ask, what their website says, and how AI search reads it. For most clients that runs in three stages.
- Find out how AI sees you now. We check whether AI tools can reach and read your site, ask them the questions your customers ask, and record what they say about you and which competitors they recommend instead. That gives you a baseline and, usually, a short list of the pages that matter most.
- Rewrite the pages that matter. We restructure each priority page section by section: a clear heading that states the topic, the answer first, specific names of services and places, and every section able to stand on its own. We work in your voice, with your real prices, timescales and examples, because that first-hand detail is what AI tools cannot invent and what makes a passage worth citing.
- Make sure the build matches the words. We check that the structure survives the page builder, that nothing important is hidden in tabs or scripts, and that hosting and security settings are not quietly turning AI crawlers away. Then we measure what changes.
None of this replaces ordinary SEO. It is SEO done to a higher standard of clarity, which is what AI search now demands.
Where to start
The quickest first step is to take one important page on your website, pick any section, and read it on its own. If the heading does not say what the section is about, or the first sentence does not answer the question a customer would ask, you have found the place to begin.
If you would rather have an honest, outside view of how your whole website reads to AI, get in touch and we will talk it through.
Sources
- Optimizing your website for generative AI features on Google Search, Google Search Central
- BLUF (communication), Wikipedia, on the origin of “bottom line up front” in US Army Regulation 25-50