Anyone working in content learned on-page SEO as a list of fields to fill in. Title with the keyword, meta description within the character limit, alt text on the images, and the page was ready to compete for the first results page.
That list still holds, and it stopped being enough. The results page changed format, because it now answers before it offers links.
What this demands from on-page SEO is not a new technique, it is greater rigour in the same fundamentals, with attention to how the text will be sliced and extracted.
What is on-page SEO and what does it solve in organic traffic?
On-page SEO is the optimization of a page's internal elements so search engines understand the subject, the intent served and the quality of the content. It covers the title tag, meta description, URL, heading hierarchy, body copy, images, video, internal linking and data markup.
What on-page SEO solves is eligibility. It puts the page in a position to compete for the right query.
Caption: The on-page SEO checklist feeds both the traditional results page and AI generated answers.
What it does not solve is authority. A page with flawless structure, published on a domain with no history and no external references, competes at a disadvantage against an average page on an established domain.
That distinction prevents the most common frustration among beginners. Optimizing the on-page elements of a piece and seeing no movement in the rankings rarely means the optimization was wrong, it usually means the bottleneck was somewhere else.
On-page SEO is therefore the first investment in search, not the only one. It is the cheapest, the fastest and the only one entirely under the publisher's control.
What is the difference between on-page and off-page SEO?
On-page SEO gathers everything inside the page and the site: content, structure, markup and performance. Off-page SEO gathers the signals that come from outside, mainly backlinks, brand mentions, citations in publications and presence on other search platforms.
The separation is didactic, and the operation is a single one.
Well structured content attracts links more easily, because it becomes a natural reference for anyone writing on the subject. External authority, in turn, makes that same content rank faster.
Working on-page and off-page SEO in parallel is what sustains organic traffic growth over time. Teams that pick only one side tend to stall on a plateau after the first gains.
It is worth knowing the off-page SEO tactics for organic traffic before deciding where to spend effort, because the answer depends on the domain's stage.
Why did on-page SEO change with AI Overviews and AI Mode?
On-page SEO changed because the results page started answering directly. Google launched AI Overviews in 2024, replacing the former Search Generative Experience name, and AI Mode in 2025. Both use the same index and the same ranking systems as traditional search.
The consequence for content producers is measurable.
A study by the Pew Research Center, covering 900 US adults and nearly 69,000 analysed searches in March 2025, found clicks on a traditional link in 8% of visits with an AI summary, against 15% of visits without one. Clicks on a link inside the summary itself came to 1%.
Session endings also rose, from 16% to 26% of pages when a generated summary was present.
This does not eliminate on-page SEO, it changes what on-page SEO has to deliver. Google's documentation on AI features in search states explicitly that there is no additional technical requirement to appear in these features.
What exists is an advantage for content that is easy to extract: one idea per paragraph, the answer in the first sentence, an explicit subject instead of a dangling pronoun, clean definitions and headings phrased as real questions.
In practice, the text is now read in pieces, so every piece has to make sense on its own. That is the central adjustment in the current checklist, and it runs through every item below.
How do you write the title tag and meta description of a page?
The title tag should stay within 60 characters, carry the primary keyword preferably at the start and reflect a real search. The meta description works best between 145 and 155 characters, with the keyword and a concrete reason to click, without promising what the page does not deliver.
The title tag is the highest return per unit of effort in all of on-page SEO. It defines how the page presents itself in the results and influences the click, and a generic title hands the query to a competitor even when the content is better.
Question format works well, because it mirrors how people search and speaks to the "People also ask" block. The remainder should carry a term from the content itself, never a generic filler such as "complete guide".
The meta description is not a ranking factor, and Google rewrites a good share of the ones it finds. It is still worth writing it: when it is used, it drives the click-through rate.
One detail that slips by: the title tag and the H1 do not have to match, and it is better if they do not. Two different phrasings cover two variations of search.
How do you optimize the URL and H1 without repeating the keyword?
The URL should be short, lowercase, without accents, with hyphens separating words and with the primary keyword. The H1 should stay within 60 characters, carry the keyword at the start and reinforce the topic with a term from the semantic cluster, without repeating the title tag's phrasing.
The URL is a one-time decision, so it pays to get it right first time. Changing it later requires a redirect, and every redirect adds a risk of losing traffic, which is why a lean slug of three to five words ages best.
Avoid dates and version numbers in the slug. A URL with "2026" forces a republish every year, or ages in front of the reader.
In the H1, the most common mistake is repeating the title tag word for word. That wastes the chance to cover a variation of intent with the same content.
How do you optimize blog content without keyword density?
Optimizing blog content today means covering the topic in depth and structuring the text for reading and extraction. The old rule of density and secondary keyword counts no longer holds: keep the keyword in the title tag, the H1, the first 100 words and one subheading, and write naturally.
Three habits inherited from an earlier phase of SEO are worth undoing.
The first is the density target. Forced repetition of a term is inert at best, and harmful when it makes the text artificial, something Google's guidance on helpful content addresses directly.
The second is the word count target. Long content ranks well when the depth justifies the length, and loses force when the volume comes from repetition and padding. Information density beats character count.
The third is the secondary keyword list as a mechanical checklist. The better path is to map the sub-questions a reader would ask and answer each one in its own section, which covers the topic's vocabulary without artificial effort.
On structure, three decisions change the outcome: each section opens with a direct answer in 40 to 60 words, each paragraph carries a single idea, and each subheading takes the form of a real question containing that section's term.
How do you optimize images, video and page performance?
Images need a descriptive file name, alt text that describes the actual content and a reduced weight, in a modern format such as WebP. Video increases time on page when it is contextual, and page performance is measured by Core Web Vitals, which affect the experience before they affect rankings.
Alt text exists first for accessibility, and the SEO gain is a consequence. Describing the image properly serves people using a screen reader and, at the same time, gives the search engine context about the page, while stacking keywords into the alt helps neither.
Image weight is the most frequent cause of a slow blog. A 2 MB cover image compromises loading on a mobile connection, and the problem is solved with compression in minutes.
With video, the criterion is relevance. A video embedded only because "video helps SEO" adds weight and retains nobody.
On Core Web Vitals, technical honesty is in order: they carry little weight as a direct ranking signal, and a lot of weight in the bounce rate. The gain shows up in conversion before it shows up in position.
How do you use internal linking and structured data in on-page SEO?
Internal linking distributes authority across pages on the same site and helps search engines understand how the content relates. Structured data, in JSON-LD, describes the content in a machine-readable way. The minimum kit for a blog is Article or BlogPosting, BreadcrumbList, Organization and Author.
Internal linking is the most underused item on the list. It costs nothing, is entirely controllable and solves two problems at once: it takes the reader to the next piece and signals to the search engine which pages are central to the topic, and a descriptive anchor works better than "click here" or "read also".
One practical rule: every new piece should receive links from relevant older pieces, not merely point to them. Publishing without revisiting the archive leaves the new page orphaned.
On structured data, an old expectation needs updating. The visual rich result for FAQPage was deprecated by Google, as its own structured data documentation records, and HowTo had been deprecated before that.
That does not mean the FAQ block lost its purpose. It remains useful for long-tail intent, for the related questions block and for AI citation, and schema.org applied to content still holds as semantic data.
What changed is the reason for doing it, and that matters when you have to justify the effort.
Which AI SEO techniques should you apply beyond on-page?
The AI SEO techniques with the most consistent effect are citing named sources, including verifiable statistics with a date and bringing in direct quotes from specialists. They increase the chance of a passage being selected in generated answers, and they work better combined than in isolation.
None of them is a new SEO technique, all of them are basic journalism.
The difference is that there is now a mechanical incentive. Answer engines retrieve passages and need support in order to cite, so a paragraph with a number, a source and a date has an advantage over a paragraph of generic opinion.
A second change is the breadth of the topic. Google unpacks one question into several sub-questions before composing the answer, a behaviour known as query fan-out, which favours content covering the whole cluster rather than a single query.
It is also worth following how Google's AI Mode displays sources, because the citation format changes which kind of content is favoured. The complementary reading on Answer Engine Optimization organises these fronts into a method.
Finally, authorship carries weight. A real byline, a bio with credentials and first-hand experience are the signals E-E-A-T describes, and they are hard to fabricate at scale.
How do you learn SEO and build an SEO training programme?
Learning SEO works better through guided practice than through theory courses. An effective internal SEO training programme starts from three things: an objective publishing checklist, review of real content with feedback, and a monthly read of the metrics that connects decisions to results.
The classic mistake is starting with algorithm theory. Understanding a core update is interesting and changes nothing in the routine of the person writing, whereas knowing how to build a title tag, structure a section with the answer at the top and choose an internal link with judgement changes it every day.
A sequence that tends to work in content teams has four stages, in order. Start with the first and only move on once it becomes a habit:
- Publishing checklist: title tag, H1, meta, slug, section structure, internal links and alt text, all checked before publishing.
- Peer review: one person reviews another's content with the checklist in hand, which teaches faster than any lesson.
- Reading the metrics: track impressions, clicks and average position per page in Search Console, and connect movement to the change that was made.
- Updating old content: revisit posts losing traffic, which teaches diagnosis and usually pays back faster than publishing from scratch.
Teams without the bandwidth for this normally save time with support from SEO specialists during the structuring phase, and then take over the routine. To track the effect, it helps to know how to check your site's position on Google and to review periodically how to optimize your blog.
Frequently asked questions about on-page SEO
So which on-page SEO checklist should you prioritise now?
The on-page SEO checklist is structurally the same, and its emphasis has shifted. Title tag, meta description, URL, H1, images and linking remain mandatory, and the structure of the text gained weight, because the answer on the results page is now assembled from passages extracted from the page.
If you have to prioritise, start with three fronts. They deliver the highest return in the shortest time.
First, rewrite the title tags of pages that already have impressions and few clicks. Second, reorganise the sections of your main content to open with a direct answer. Third, support your claims with named sources and dates.
Keyword density, secondary keyword counts and word count targets can leave the list. They consume time and do not move rankings.
If your operation needs to turn this into a routine, with a diagnosis of what is losing position and priorities set by potential, that is what mkt4edu's SEO work is about.
If AI Overviews and AI Mode are already part of your traffic, our SEO for LLM service extends this on-page checklist to what language models actually read and cite.
To review your site's situation and define the order of attack with your team, talk to our specialists.
.jpg)
Portugal