On-Page SEO Checklist: the page-level fixes that matter

What is on-page SEO, exactly?

On-page SEO is everything about a single page that you can change yourself: its title, its headings, its words, its images and its links. It is called "on-page" because none of it depends on other websites — unlike backlinks, which you cannot directly control.

Think of it as labelling a jar clearly. Google's ranking tutorial in this series explained that Google judges pages on relevance and usefulness. On-page SEO is how you *show* relevance: the page states, in the words searchers use, what it is about and who it is for. Nothing here will rescue a page nobody wants — but a good page with bad labelling stays invisible, and labelling is fixable in an afternoon.

Does the page have one clear keyword?

Before touching anything, answer one question: what single search should this page win? Take the research table from the keyword research tutorial — each row already has one main keyword and one intent. If you skipped that tutorial, do its 30-minute exercise first; this checklist cannot start without its answer.

The test is simple: can you say the page's keyword in one breath, and does everything on the page serve it? A page targeting "birthday cake delivery" should not also try to rank for "wedding photography prices". Two jobs means two pages.

Expected result: one written-down keyword for the page you are about to fix, copied from your research table — for example, "birthday cake delivery".

Is the title written for a human who is searching?

The title (the <title> tag) is the single most visible thing you control: it becomes the blue clickable headline in Google's results. Write it for the person scanning those results, not for a robot:

1. Put the keyword near the start. "Birthday Cake Delivery in Dera Ismail Khan — Order Fresh" beats "Welcome to Our Website".
2. Say what happens next. A title that promises an action ("Order Fresh", "See Prices", "Step-by-Step Guide") gets more clicks than a title that just names a topic.
3. Keep it short enough to display in full. As a rule of thumb, around 60 characters — longer titles get cut off with "…" in the results, and the cut usually eats the interesting part.
4. One title per page, and no duplicates. Every page on a site needs its own title. Two pages with the same title are telling Google they are the same page.

Expected result: a title you could read aloud to someone and they would know exactly what the page offers and what to do there.

Does the meta description earn the click?

The meta description is the short paragraph under the title in the search results. Be honest about what it does: it does not directly push a page higher in the rankings — but it decides whether the searcher clicks *your* result or the one below it. A page that ranks fifth with a tempting description can out-click a page ranking third with a dull one.

Write it like a two-sentence advert for the page:

• Include the keyword naturally, because Google bolds the searcher's words in the description — a bolded phrase catches the eye.
• Say the benefit and the next step: what the visitor gets, and what to do ("See today's designs and prices, then order by WhatsApp in two minutes").
• Keep it to roughly 150–160 characters so it is not cut off mid-sentence.

If you write no meta description, Google invents one from your page text — sometimes fine, usually bland. Two minutes of writing beats leaving it to chance.

Expected result: a description that makes *you* want to click, reading the results page as a stranger would.

Do the headings tell the story on their own?

Headings (<h1>, <h2>, <h3>) are the page's skeleton. A searcher — and Google — should be able to read just the headings and understand the whole page. Three rules:

1. Exactly one <h1>, and it matches the search. The H1 is the page's headline: "Birthday Cake Delivery in Dera Ismail Khan". Not "Welcome", not the business name alone.
2. <h2>s break the page into the questions the searcher has. For an order page: "Today's designs", "Prices", "How to order". Each H2 is a promise the section underneath must keep.
3. Headings go in order — never skip levels. H1, then H2, then H3 under an H2. Skipping from H1 to H3 is like a book with chapters but no sections; it confuses screen readers and Google alike. (This is the HTML side of the rule the semantic HTML tutorial in the frontend series teaches.)

A quick self-test: copy only the headings into a blank document. If a stranger could tell what the page is about and what to do there, the structure works.

Expected result: an outline of headings that reads like a table of contents for the searcher's questions.

Does the first paragraph answer the search?

Most visitors decide in seconds whether a page is for them, and Google pays extra attention to the opening text. So the first paragraph must do its job immediately:

• Use the searcher's words early. If the keyword is "birthday cake delivery", those words should appear naturally in the first paragraph — not repeated five times, just present, the way Day 12's bakery example does it.
• Answer the question, then elaborate. "We deliver fresh birthday cakes across Dera Ismail Khan, ordered by WhatsApp in about two minutes." The details — designs, prices, areas — come after.
• No throat-clearing. Delete openers like "In today's fast-paced world…" and "Welcome to our website". The searcher arrived with a question; start with the answer.

Expected result: a first paragraph that satisfies the search on its own, so everything below it is welcome detail rather than a rescue mission.

Is the content genuinely worth ranking?

This is the item no checklist can fake. Google's whole job — per the ranking tutorial — is returning the most relevant, useful page. Ask these questions honestly:

• Does the page fully answer the search? A page targeting "birthday cake delivery" needs designs, prices, delivery areas and how to order — not two sentences and a phone number.
• Is any of it specific to this page? Generic text that could sit on any bakery's site ("we bake with love and the finest ingredients") tells Google nothing. Specifics — actual designs, actual prices, actual delivery areas — tell it everything.
• Could a competitor's page obviously beat it? Open the pages currently ranking for your keyword. If yours would clearly lose a side-by-side comparison, the fix is better content, not better tags.

Thin pages lose to complete pages. There is no minimum word count that matters — there is only the question: after reading this, does the searcher need to go back to Google?

Expected result: a page you would genuinely recommend to a friend with that question, not just a page that ticks the technical boxes.

Are the images helping or hurting?

Images are part of on-page SEO, and beginners get them wrong in two predictable ways: meaningless filenames and missing descriptions.

1. Rename the file before uploading. IMG_20260112.jpg tells Google nothing. chocolate-birthday-cake-delivery-dera-ismail-khan.jpg tells it exactly what the image shows — and descriptive filenames are also how your images can be found in image search.
2. Write real alt text. Alt text describes the image for people who cannot see it (and for Google): "Chocolate birthday cake with 'Happy Birthday' icing, ready for delivery". Not "image1", not a stuffed keyword list.
3. Keep files small. A 4MB photo slows the page, and a slow page loses visitors — especially on phones. Resize and compress images before uploading; nobody needs a 4000-pixel-wide cake photo on a web page.

Expected result: every image on the page has a descriptive filename, honest alt text, and a file size measured in kilobytes, not megabytes.

Do the links on the page make sense?

Links tell Google how your pages relate to each other, and they tell visitors where to go next. Check three things:

1. Link to your related pages with descriptive words. "See our birthday cake designs" as the link text beats "click here" — the words of the link tell Google what the linked page is about. (This is the same honest-labelling idea as the headings.)
2. Every important page should be reachable. If the order page is the page you want to rank, link to it from the homepage and from related pages — like the guide and gallery pages from the keyword research table. A page no other page links to is a page Google struggles to find and trust.
3. Link out when it genuinely helps. Linking to a useful external page (a map, a supplier, a source) is fine and normal. Hoarding all links for yourself looks — and is — unhelpful.

Expected result: a page connected to the rest of the site by links whose text describes where they go.

Does the page work properly on a phone?

Most searches now happen on phones, and Google judges your page mainly by how it looks and works on a phone — not on a desktop. So check the page on an actual phone, not just by shrinking your desktop browser:

• Can you read the text without pinching to zoom?
• Do buttons and links tap easily, without hitting the wrong one?
• Does the page load in a few seconds on mobile data?
• Does anything overlap, or slide off the side of the screen?

If the answer to any of these is no, fix the page layout before worrying about keywords — a page people immediately leave cannot rank well, whatever its title says.

Expected result: a page you would happily use yourself, one-handed, on a mid-range phone.

Worked example: auditing a practice bakery page

The practice business is the one from the keyword research tutorial: an imaginary home bakery taking cake orders through Instagram and WhatsApp (a practice example, not a client). Its research table gave the order page one main keyword — "birthday cake delivery" — with "birthday cake price" as a supporting phrase. Here is that page before and after the checklist.

Before — what the checklist finds wrong:

• Title: "Home" — says nothing, matches no search.
• No meta description — Google shows a random sentence from the page.
• H1: "Welcome to our website".
• First paragraph: "We are a small bakery. We make cakes. Contact us."
• The keyword "birthday cake delivery" appears nowhere on the page.
• One photo named IMG_20260112.jpg, no alt text, 3.8MB.
• No links to any other page; no page links to it.

After — the fixed page head and headings:

<!DOCTYPE html>
<html lang="en">
<head>
  <meta charset="UTF-8">
  <meta name="viewport" content="width=device-width, initial-scale=1.0">
  <title>Birthday Cake Delivery in Dera Ismail Khan — Order on WhatsApp</title>
  <meta name="description" content="Order a fresh birthday cake for delivery in Dera Ismail Khan. See today's designs and prices, then order by WhatsApp in about two minutes.">
</head>
<body>
  <h1>Birthday Cake Delivery in Dera Ismail Khan</h1>
  <p>We deliver fresh birthday cakes across Dera Ismail Khan — order by
  WhatsApp in about two minutes, and your cake arrives the same day.</p>

  <h2>Today's birthday cake designs</h2>
  <p>...</p>

  <h2>Birthday cake prices</h2>
  <p>...</p>

  <h2>How to order</h2>
  <p>...</p>
</body>
</html>

Expected result: paste this into a file named practice-page.html, open it in a browser, and the tab shows the full title, the page shows one clear H1, the headings read as a table of contents, and the first paragraph answers the search in the searcher's words. View Source (right-click → View page source) shows the meta description sitting in the <head> where Google reads it.

The remaining fixes, as a checklist pass:

Expected result: a page that a stranger searching "birthday cake delivery" would recognise as the answer in the first ten seconds — and that Google can read, label and file correctly.

Common mistakes

• Repeating the keyword until the text sounds broken. "Birthday cake delivery in Dera Ismail Khan for the best birthday cake delivery" reads like a robot wrote it — because one did. Use the keyword naturally, once in the title, once in the H1, once early in the text, and then write for the human.
• Writing the title for Google instead of the searcher. "Best Cheap Birthday Cake Delivery Dera Ismail Khan Buy Now" is keyword soup. The title from the worked example wins the click because a person would click it.
• Fixing ten pages halfway instead of one page fully. On-page SEO compounds per page. One page that passes all ten checks teaches you the method; ten pages with new titles teach you nothing.
• Changing everything, then changing it back a week later. Search engines need time to re-read a page. Make the changes, note the date, and leave the page alone for a few weeks before judging.
• Ignoring the phone check. A page that is unreadable on a phone fails the visitor before Google ever gets a vote. The mobile check is not item 10 by importance — it is item 10 because you check it last, after the content is right.

Frequently asked questions