On-Page SEO Checklist for a Small Business Website

Work the list in order

There is a version of this checklist that runs to a hundred items and produces an afternoon of activity with no change in traffic. The useful version is short, ordered by leverage, and honest about the fact that most small sites need a handful of things doing properly rather than everything done slightly.

The order matters. Technical access sits underneath everything, because a page that cannot be crawled cannot be improved by anything written on it. Then intent and content, since there is no technical fix for a page answering the wrong question. Then performance and markup, which help within a range rather than decisively. Then linking and consistency, which are mostly about how clearly the site signals what it is about and where it is.

One caution before the list. The fastest way to waste this work is to rewrite pages that already perform well. Established pages have usually accumulated links and engagement that a rewrite quietly discards. Change them when there is a specific reason, not as part of a clean-up.

Crawlability before anything else

Start with the mechanical question: can a search engine find, fetch and store every page you care about? Most problems here are configuration rather than design, and all of them are visible if you look.

The items worth checking, in rough order of how often they turn out to be the cause:

• Pages returning an error, or a redirect, where a page should be.
• Content behind a login, an internal search box, or a form. If it cannot be reached without submitting something, treat it as not indexed.
• Pages generated from a query string or session identifier, producing a near-infinite number of URLs from a small amount of content.
• Pages blocked by a robots directive or a header rule added years ago for a reason nobody can remember.
• Pages in the sitemap but not on the site, and pages on the site but missing from the sitemap.
• Default or duplicate pages — a paginated list that starts at page one again, a sort order nothing links to, a category page with no content of its own.

The test is straightforward: take the list of URLs that matter to the business and check each one directly for status, canonical tag and presence in the sitemap. Do it in a spreadsheet rather than trusting a single report.

On sites built with a page builder, a fair share of crawl problems come from templates that generate near-duplicates — tag archives, date archives, search results, author pages. Each is individually harmless and collectively they dilute the site. Deciding what to keep out of the index is a content decision as much as a technical one.

Titles, headings and matching the actual question

Once pages can be found, the work is whether each one answers something real. Three elements carry most of the weight: the title tag, the heading structure, and the opening of the page.

The title should state the subject plainly and be specific enough to be distinguishable from its neighbours. Titles written for internal colleagues rather than searchers are common in small organisations, where a page called “Services – Phase 2” tells a reader nothing and gets rewritten by nobody because it already looks tidy in a menu.

Headings should describe the content that follows them. A single h1 per page that matches the title’s intent, then sections that read as an outline. This is as much for the reader as for the machine: a page whose headings make sense when listed is easier to use, and the ones that do not usually indicate the structure is confused.

On matching intent, the method is unglamorous. Look at what people actually search for and what the pages they land on actually say. The mismatch is usually one of four kinds: the page promises something broader than it delivers, the language differs from the language people use, the page is thin where the query implies depth, or several near-identical pages are competing for the same thing and the search engine has to guess which is authoritative.

That last one is worth fixing rather than ignoring. Two pages targeting the same intent dilute each other. Merging them into one strong page, keeping the useful material from both and leaving a sensible internal link where the second used to sit, usually beats trying to differentiate them with a sentence.

Speed is an engineering task

Site speed is a development problem with a marketing symptom. It improves when the people writing the code change how they work, not when a plugin is installed.

The causes that recur: images uploaded at the size they were exported at, with no resizing and no modern format; render-blocking stylesheets and scripts loaded before the page needs them; third-party tags accumulated over years, each added by a different campaign; web fonts with a long visible fallback period; and long, unoptimised embeds sitting in the middle of content.

What actually moves the figure:

• Resizing and re-encoding images, and setting dimensions so the layout does not jump. Usually the largest single gain on a media-heavy site.
• Deferring scripts not needed for the first screen, including analytics.
• Reducing the tag pile to the ones with a current internal owner and a stated purpose. This needs somebody to decide, because nobody removes a tag without being sure it is unused.
• Self-hosting or carefully subsetting fonts.
• Moving embedded video out of the initial load.

Measure on real devices rather than a desktop on fast broadband, and measure the pages that matter rather than the home page. A site with a fast home page and a slow service page still loses the visit. When SmartEdge IT Solutions is asked to look at performance, the first fortnight is measurement, because the fix depends entirely on which of the causes above is actually present.

If the site runs on a managed platform, WordPress development work should start by finding out which of the above is actually happening, rather than installing a caching layer over it. On a custom build the same applies, and web development should treat this as an engineering task with a budget attached rather than as a setting somebody can toggle.

Structured data has to be true

Markup describing what a page contains is used to interpret that page. It is worth getting right and very easy to get wrong, because the failure is invisible: the markup validates, the page renders perfectly, and nothing at all appears.

The principle is simple. Markup must describe content that is visibly present on the page. Marking up a review that is not there, an FAQ the page does not visibly contain, or a product price that differs from what a visitor sees is not a technical optimisation. It is a false statement published to a third party, and it is the sort of thing that ends a relationship badly when discovered.

What is worth marking up is usually modest: organisation details, the specific services offered where each has its own page, articles with a clear author and date, and breadcrumbs. Where a page is a genuine question-and-answer page and the answers are visible to the reader, FAQ markup is defensible.

Validate the markup against the search engine’s own testing tool, and check the page renders as intended on a mobile device, because a great many of these sites have not been looked at on anything smaller than a large monitor for some time.

Internal linking is the cheapest work here

Internal links are entirely within your control, cost almost nothing, and are routinely ignored. They do three things: they tell a crawler what a page is about and which pages relate to it, they distribute attention across the site, and they move a reader from one relevant thing to the next.

The habit worth adopting is that when you publish a page, you go back and add links to it from the existing pages where it is genuinely relevant. Not from all of them — from the two or three where a reader who wanted the new page would plausibly have been. It takes twenty minutes and is more durable than anything done to the new page itself. It is also the item most often skipped when a deadline arrives, which is why new pages on small sites so often sit orphaned.

Link text should describe the destination. “Read more” tells a reader and a crawler nothing; the name of the service, the problem being described, or the question being answered does. Generic anchor text repeated across hundreds of links is a pattern worth avoiding.

On a small site, the pages with no inbound internal links are the place to look first. They are usually the newest, or the ones written for an audience who arrived from somewhere else entirely.

On a larger marketing site the same reasoning turns into search engine optimisation work argued with measurement rather than judgement. The underlying habit is identical.

Consistency across local listings

If the business serves a defined area, the details that appear about it in third-party listings need to agree with each other and with the website. Not identical everywhere, but not contradictory.

Name, address, telephone, hours and the services listed are the details to check. Differences in format are mostly harmless. Differences in substance are not. A listing showing a different telephone number from the website sends some callers to a building that is not the business, and it is one of the few problems adjacent to this work that causes an immediate and obvious loss.

Stale hours are the second common issue, particularly for businesses that are not open every day. A listing showing hours that were accurate two years ago causes the kind of wasted visit that ends a first impression permanently.

This is unglamorous maintenance and it needs an owner, because listings change precisely when nobody is watching. It fits naturally into the same review cycle as the content calendar rather than being treated as a one-off task.

Measurement, and what to leave alone

Set up the measurement before making changes, or you will never know which change did what. Two things cover most of it: search console, which shows what is being found and clicked, and an analytics tool configured to record which pages those visits actually land on.

The reports worth looking at monthly are short. Pages getting impressions but few clicks, which suggests the title is not matching the searcher. Pages getting clicks but leaving immediately, which suggests an intent mismatch. Search terms you did not target, which occasionally point at content you should have written. And the crawl errors, which is the one item on this list that quietly returns after you have dealt with it.

Then decide what to leave alone. Not every low-traffic page needs work; some exist for completeness and cause no harm. Not every underperforming page should be rewritten; some cover topics a small portion of your audience genuinely searches for. Chasing aggregate traffic across an entire site without looking at which pages matter produces a great deal of activity and very little clarity. SmartEdge IT Solutions writes these recommendations into the audit document with the reasoning attached, so the decision to leave something alone is recorded rather than merely apparent.

If you would rather have the whole picture done once than piecemeal, a full website audit produces the ordered version of the checklist above. Beyond that, the honest summary for a small business is that the first four items yield most of the available improvement, and everything after that is steady maintenance rather than a project.