Search advice has a short half-life. Open a guide from five years ago and most of it is either obsolete or actively harmful — keyword density targets, exact-match anchor text ratios, guest posting at volume, the whole apparatus of trying to reverse-engineer a system that changes several thousand times a year.
Underneath that churn, a small amount of the work has held up. It has held up because it is not really SEO. It is building a site that is genuinely easy to understand and genuinely worth arriving at, which is what every update has been imperfectly trying to reward.
One: say what the page is about, near the top, in ordinary words
This sounds too obvious to state. It is also the single most common failure I see on business websites, and it costs more traffic than anything technical.
A page selling gutter cleaning in Leeds should say, within the first hundred words, that it is about gutter cleaning in Leeds. Not “protecting your property investment” or “comprehensive exterior maintenance solutions.” The words a customer would type.
Search engines have become far better at understanding synonyms and intent, which has been widely misread as permission to be vague. The opposite is closer to true. Better language understanding means the system can tell more accurately that your page is about nothing in particular.
A practical test: read the first paragraph aloud to somebody outside your industry and ask what the page is selling. If they cannot say, no amount of technical work will save it.
Two: structure the page the way its content is actually shaped
Headings are not decoration and they are not font sizes. They are an outline that both people and machines use to understand how the page is organised. One h1 that states the subject, h2 for major sections, h3 for subsections, in order, without skipping levels because a smaller size looked better.
The same principle extends past headings. Lists marked up as lists. Tables as tables. Navigation in a nav. The main content in a main. This is unfashionable, boring work that also happens to make the page usable with a screen reader, which is a reasonable indication that it is the right thing to do independent of rankings.
Schema markup belongs here too. Telling a search engine explicitly that this is a local business with these hours in this area, or that this is an article by this person published on this date, removes ambiguity you would otherwise be hoping it resolves correctly.
Three: make it fast on a bad phone connection
Speed has been a ranking factor for over a decade and the emphasis has only sharpened. But the ranking effect is the less interesting half. The larger effect is that slow pages are abandoned, and abandoned pages do not sell anything.
Most slowness on small business sites comes from a short list of causes, in rough order of severity: enormous unoptimised images, a page builder loading a hundred kilobytes of unused CSS, four analytics and marketing scripts, and web fonts blocking the first render.
None of that requires deep expertise to fix. Compress the images and serve them at the size they are displayed. Remove the scripts nobody looks at the output of. Self-host two font files rather than pulling six weights from a third party. On a typical small business site that work alone often halves load time.
Test on a throttled connection, not your office fibre. A page that feels instant on a desktop in a city can be genuinely unusable on a phone on a rural road, which is where a good proportion of trade customers are when they search.
Four: be the best available answer to something specific
This is the one that outlasts everything else, and the hardest to fake.
Every update since roughly 2011 has moved in the same direction: rewarding pages that answer a question well and demoting pages that exist to occupy a keyword. The tactics that gamed this have all eventually stopped working. The underlying instruction has never changed.
In practice that means choosing narrower subjects than feels comfortable. A page about “roof repair” competes with everyone. A page about what actually happens when a tile roof leaks around a chimney flashing, with photographs of the repair, an honest range of what it costs, and a clear statement of when it becomes a full re-roof, competes with almost nobody — and the person reading it is much closer to buying.
Specific pages also age better. General pages are rewritten by competitors constantly. A detailed account of a particular problem, written by someone who has fixed it many times, stays useful for years.
What I would ignore
- Keyword density targets. There is no threshold. Write naturally about the subject and the words appear.
- Buying links. The risk has always outweighed the return and the detection has only improved.
- Publishing volume for its own sake. Twelve genuinely useful pages beat two hundred thin ones, and the thin ones can drag down the rest.
- Chasing every ranking factor announcement. Most are minor, many are misreported, and the reaction usually costs more than the factor was worth.
The uncomfortable part
This is slow. The four things above take months to show results and none of them produce a satisfying weekly report.
A worked example
A cabinetmaker I worked with had a page titled Services that listed nine things in a bulleted column and ran to about eighty words. It ranked for nothing. It also told a visitor almost nothing, which is the more important failure.
We replaced it with four pages, each about one thing he actually wanted more of: fitted wardrobes, kitchen carcassing, alcove units, and commercial shopfitting. Each page described the process, showed three completed jobs with photographs, gave an honest price range, and answered the two questions he was asked on every single enquiry call — how long it takes, and whether he does the installation himself.
No technical SEO was performed beyond correcting the heading structure and compressing the images. Within four months three of those pages were ranking on the first page for their terms, and the enquiries arriving through them were noticeably better qualified, because people had already read the price range and self-selected.
The lesson is not that four pages beat one. It is that the four pages contained things a competitor's page did not: real photographs, real numbers, and answers to questions only someone who does the work every day would know to answer.
On measuring the right thing
Rankings are a proxy, and a noisy one. They vary by location, by device, by whether the searcher is logged in, and increasingly by what the search engine decides to show above the results. Two people searching the same phrase from the same suburb can see meaningfully different pages.
Measure enquiries instead. Which pages were people on when they called or filled in the form? Which search terms brought them, according to Search Console rather than a rank tracker? Those two questions answer everything a small business actually needs to decide.
A page ranking third that produces four enquiries a month is worth more than a page ranking first that produces none, and the second situation is more common than most people expect — usually because the top-ranking page attracts a search intent that is nowhere near a purchase.
The technical minimum
Very little technical work is genuinely required. This is the whole list for most small business sites:
- One page per topic, with a URL that reads like a sentence rather than a database key.
- A title tag and meta description written for a human, unique on every page.
- A sitemap, submitted once to Search Console, and a robots file that does not accidentally block anything.
- Canonical tags, so the same content at two URLs does not compete with itself.
- Redirects from every old URL when anything moves.
- Schema markup for the things you are — a local business, an article, a product.
That is it. Everything beyond this list is optimisation of something already working, and most small sites are not yet at the point where it matters. If someone is selling you a monthly technical retainer and your five main pages still open with a slogan, the priorities are the wrong way round.
That is precisely why they keep working. Anything that produces results in a fortnight is a loophole, and loopholes get closed. Work that takes six months to pay off is work most competitors will not do, which is the whole of the advantage.
If you want one place to start: open your five most important pages on a phone, on a slow connection, and read the first paragraph of each aloud. Most sites fail that test, and most of what needs fixing becomes obvious in the process.
