On-Page SEO Checklist You Can Run in a Browser

The invoice says on-page optimisation. The monthly report describes activity. The page looks the same as it did in March, and you have no way of knowing whether it is the same, because the changes that matter for search are mostly invisible from the front of a page. That leaves two bad options: keep paying on trust, or cancel a service that might have been doing exactly what it said. There is a third. Seven of the things that count as on-page work are visible in any page's source, and checking them takes ten minutes with a browser, a right-click, and no software at all. What follows is that checklist, what each item proves, and what Google says each one does and does not do.

Open the page in a browser, right-click, choose View Source, and use the browser's find function. Seven searches tell you most of what an on-page audit would tell you about setup.

Save the URL, the date and the excerpts. That record is the whole basis for the rest of this piece, and it costs nothing.

The checklist most people find is a list of places to put a keyword, followed by a ranking expectation. It skips the question that comes first, which is whether the page can be discovered, crawled and understood at all, and it treats a title and a description as a finish line when Google treats them as suggestions. Google's title link documentation says it may generate a title from several sources, and its snippet documentation says it may use page content instead of the meta description when it judges that more useful. A keyword in the title is a reasonable thing to do. It is evidence of setup rather than of anything else.

Real work is legible as a deliverable. A defined list of URLs that were worked on. A before record: what the title, description, heading, canonical and robots signals were. The proposed change for each page. Then a live after record showing the same signals in source, an excerpt proving the main copy is in the delivered HTML, examples of real internal anchors, and the line in the sitemap containing the preferred URL. A provider who can show that record has done on-page work you can verify. A report that describes optimisation without a URL and a source excerpt has described something you cannot.

The first three checks take a minute each. Search for `<title` and read what is between the tags: it should describe this page, in words a person would recognise, and it should differ from every other page's. Search for `name="description"` and read the content attribute: one or two sentences about this page, again unique to it. Then search for `<h1` and confirm there is one, that it is the heading you can see on the page, and that it matches what the page is for.

What stops this being a keyword-stuffing exercise is discipline on two points. The page needs one clear primary topic, and the three signals should all describe it the same way. And Google's own SEO starter guide frames these as descriptors that help people and crawlers understand the page rather than as levers. A title that reads as a sentence about the page is doing its job. A title that reads as a list of search terms is a title Google is more likely to rewrite.

A page that cannot be reached cannot be optimised, and the two checks that matter most for reach are the ones the keyword lists leave out. Google's link guidance says it can generally crawl a link only when it is an HTML anchor element with an href attribute, and that it cannot reliably extract a URL from anchors without one or from other elements driven by script events. So, a menu item that works when you click it may still be invisible to a crawler. Right-click the menu item, inspect it, and look for `<a href="...">` with a real URL. If the element is a `<div>` or a `<button>` with an onclick and nothing else, the page it leads to is being found some other way or not at all.

The sitemap is the second check. Open `/sitemap.xml`, or the sitemap location named in `/robots.txt`, and search the file for the page's preferred URL. Google's sitemap documentation says the sitemap should contain the preferred canonical URLs, fully qualified, and its crawling FAQ says a sitemap helps Google learn about a site and guarantees neither indexing nor a ranking. So, a page present in the sitemap is eligible for discovery, and that is the whole claim.

The delivered-copy check sits between those two. Search the source for a distinctive sentence from the page's main text. If it is there, the copy is in the HTML the server sends. If it is absent, it is being injected by JavaScript after load, and Google's JavaScript SEO basics say that needs care: Google queues pages for rendering, a `noindex` can cause it to skip rendering and script execution entirely, and JavaScript changes to meta tags are fragile. When the copy is script-injected, the browser check reaches its limit and the URL Inspection tool in Search Console, which shows what Google actually rendered, is the next step.

The professional sequence starts with an inventory and a short list of priority URLs rather than with editing everything at once. For each page: its purpose, its intended visitor, the current title, description, heading, canonical and robots signals, the rendered copy, the internal links, and whether it is in the sitemap. A before snapshot is saved. Access and indexability are fixed before any copy is polished, because a beautifully written page with a `noindex` on it is a page nobody will read. Then the page itself: one primary topic, a unique title and description, a visible heading, useful body copy, descriptive links to relevant pages, alternative text on meaningful images, and a canonical that names the preferred version. Then it is published, the live source is verified, the preferred URL is confirmed in the sitemap, and the change is recorded.

What takes time after that belongs to Google, and Google is specific about it. Its recrawl page says a recrawl request can take from a few days to a few weeks, that it guarantees nothing about inclusion, and that repeated requests for the same URL do not speed it up. The crawling FAQ says most sites see three days or more before a page is indexed, with no same-day expectation outside unusual, time-sensitive cases. None of those is a ranking clock, and the earlier piece on how long before a new site shows up on Google covers why a changed page and a changed position are different events.

Clutch's SEO agency pricing guide, updated 2 September 2026, reports monthly SEO fees of $2,000 to $20,000 across its listed agencies, most of them charging $100 to $149 an hour, with its category listing showing on-page SEO at $25 to $49 an hour. Its reviewed projects average $37,158.66 in value and $3,199.19 a month over a typical 12-month timeline. Those are Clutch's aggregates for work far broader than a page review, from one marketplace, and they describe that marketplace rather than any studio or any one-page scope.

What moves a real scope upward is the number and condition of the pages, research and copywriting, custom development, JavaScript rendering problems, migration or redirect work, multiple locations or languages, approval cycles, and ongoing measurement. What makes a scope readable is a written list of the URLs, the deliverables, the exclusions, the fee and billing schedule, the term and how it ends, who supplies approvals and access, what happens to credentials and files at the end, and the report format, with a plain statement of what counts as a completed monthly deliverable. Ask for the after record described earlier as the report format, and the question of whether anything was done answers itself every month.

Two clauses deserve a second read. The claim clause: no promise of a rank, a traffic level, leads or an AI Overview citation, because Google guarantees none of those to anyone, and the FTC's advertising guidance requires that any express or implied claim, including claims added to your pages during the work, be truthful and supported before it runs. And the exit clause: the domain, the hosting, the CMS administrator account, the analytics, the Search Console property and the business listings stay in your name, with the provider as a user you can remove. This is general information, not legal advice.

Everything in the seven-item table is safe for an owner to do on a page they control, because it changes nothing. Open the page, view the source, search for each signal, inspect a link, open the sitemap, copy the excerpts into a note with the date. Repeat it on the pages that matter most to the business, which for most local firms is the home page, each service page and each town page.

The stopping line is editing. Changing templates, code, canonical tags, robots directives, redirects, structured data or large amounts of copy without a backup and a change record is how a page that ranked stops ranking, and Google's JavaScript guidance adds a specific trap: a directive changed only through script may never be seen, and a `noindex` can stop rendering before the script runs. Escalate anything where the source and the visible page disagree, where two pages look like duplicates of each other, where redirects are involved, where the page fails to render, or where the copy carries a claim that needs substantiation.

A browser view is useful and not conclusive. Google recommends URL Inspection in Search Console for seeing what it saw when JavaScript is involved, which is one reason the Search Console property should be in your name and not only your provider's.

Prevention here is a release checklist and a change record, and they take the guesswork out of every future report. Before any page is published, and again after any redesign, check the live URL and its source for a unique title, a usable description, one visible heading, the main copy in the delivered HTML, the preferred canonical, no unintended `noindex`, crawlable internal anchors, and the sitemap line. The record for each change carries the page, the date, who edited it, what changed, who approved it, the source evidence, and any follow-up.

You know the process works when the same check, repeated on the published page, produces a before and an after that agree with what was invoiced, rather than when somebody asserts a result. For deeper verification, keep owner or full-user access to Search Console, where the URL Inspection and Sitemaps reports show what Google crawled and indexed, and remember that even there Google promises no crawl date and no inclusion.

One local note on the pages that matter most. A business that genuinely serves several towns is right to give each town its own page, and the thing that makes those pages work is that each carries its own researched copy about that place. Google's canonicalisation is what collapses the other kind, a template with the town name swapped, into one page it picks for you, which is the reason the checklist above should be run on every town page separately. A town page that passes it is a real page. The earlier piece on why a competitor outranks you covers what happens when it does not.

What does on-page SEO include?

The signals on the page itself: a unique title element and meta description, one visible main heading, useful copy in the delivered HTML, descriptive internal links that are real anchors, alternative text on meaningful images, a canonical naming the preferred URL, no unintended robots directive, and the page's presence in the sitemap.

Is there an on-page SEO checker I can use for free?

Your browser. View Source and the find function show every signal in the table above, and Search Console's URL Inspection tool shows what Google rendered. Neither costs anything.

What is the difference between on-page and off-page SEO?

On-page work changes the page: its title, heading, copy, links and technical signals. Off-page work concerns things outside it, such as links from other sites and business listings. This checklist covers only the first, which is the part you can verify yourself.

If the keyword is in the title, H1 and description, is the job done?

Those are three of the signals, and they prove setup rather than crawlability, indexing, accuracy or any result. Google may write its own title and snippet, and a sitemap guarantees no indexing.

How long after the changes will the page move in search?

Google says a recrawl can take days to weeks and most sites wait three days or more for indexing, and it guarantees neither. Google names no clock for a change in position.

My provider says the work was done. How do I check?

Ask for the URL list, the before record and the after record, then run the seven checks on two or three of the URLs yourself. Work that was done is visible in the source.

The invoice and the report described work you could not see, and most of it was sitting in the page source the whole time. Seven searches in View Source turn a monthly statement into something you can check against the actual page, and a provider doing real on-page work will welcome the question because the answer is in their favour.

Run the seven checks on the pages that matter, keep the record, hold the Search Console property in your own name, and every future report has an after record you can inspect rather than a description you have to trust.

If the checks turn up a service page whose copy is not in the HTML or a menu built from click handlers, Arlington Website Designer fixes exactly that on sites for businesses in Arlington, TX, with a before and after record you can read. Send the three URLs that matter most and what View Source showed for each through the contact page, and you will get back which signals are missing on each and which to fix first.