Every on page SEO services proposal promises the same activities. None says what finished work looks like. Could you tell if none of it happened?
Every on page seo services proposal promises the same activities. None says what finished work looks like. Could you tell if none of it happened?
That gap matters, because on-page work is the one discipline in SEO where every item leaves a mark you can inspect yourself, without your provider's dashboard or their permission. A title rewrite that was billed and never made shows up in Search Console as flat impressions. A schema block claiming a FAQ the page does not contain is a five-second view-source check. The uncomfortable part is not that vendors overstate. It is that the buyer had a way to check and did not know it.
Three pages define this category. seo.co splits it into three layers and hands you a score from a tool you would have to buy to reproduce. ClearLead Digital, updated July 27, 2026, publishes a price table and nothing on verifying delivery. Semrush's definitional guide, published March 20, 2026, stops at the character limits. This specification is for the person paying for it.
Why On Page SEO Services Are Not Interchangeable With "SEO"
On-page work covers what a single page controls: title, headings, content coverage, internal links, images, structured data and its share of the page-speed budget. Three things sit outside that boundary and get billed as if they were inside it.
Technical SEO decides whether a page can be crawled, indexed and rendered at all. A proposal listing "crawlability fixes" under an on-page scope is usually a technical retainer in disguise.
Off-page SEO covers links and brand signals from other domains. ClearLead's exclusion list names the boundary plainly: link building and outreach are off-page, paid search is a separate service, social content is separate, and a website redesign is separate too. Adopt it, because it is the only one published.
The ambiguity nobody settles. Victorious files schema optimisation under technical SEO. seo.co files it under on-page. The split matters to your invoice, because an on-page engagement should not charge twice for one implementation. Implementation sits with whoever owns your templates, verification with you.
A general retainer under-serves this discipline because a retainer rewards ongoing activity while on-page work is finite: you audit a page set, fix it, measure, then move on. That shape is a project, which is why per-page pricing exists. Where the purchase is broader, the same logic applies to SEO audit services, content marketing services and PPC management services.
The Deliverable List, Item by Item
Eight items: what each is, what competent implementation looks like, and who should own it. If a proposal is missing the middle column, you are being sold activities rather than outcomes. For the do-it-yourself version, the on-page SEO checklist runs the procedure.
Deliverable | What it actually is | Acceptance criterion | In-house or outsource |
|---|---|---|---|
Title, meta and URL set | A unique title tag, meta description and slug per indexable page, as a URL-mapped sheet delivered before anything is published | Titles 50 to 60 characters, unique, primary term in the first half; descriptions that agree with the page; slugs of 3 to 5 words, hyphenated (Semrush, March 20, 2026) | In-house |
Heading hierarchy restructure | Exactly one H1, logical H2 and H3 nesting, sections written to stand alone | Each section answers its own heading in the first sentence; no skipped levels | In-house |
Content gap and entity analysis | Your page against the pages ranking for the target query, listing topics and questions you do not cover | Every gap named as a specific section to write, with the query it serves | Outsource above roughly 50 pages |
Internal-link map with anchor text | A source-to-target list of links with the anchor text for each | No page more than 4 clicks from the root (seo.co's rule); every target gets a contextual link | In-house |
Image optimisation | Alt text, descriptive filenames, modern formats, compressed files | Alt text 125 characters or fewer, no "image of"; WebP for photos, PNG for transparency, SVG for logos (Semrush, March 20, 2026) | In-house |
Schema markup implementation | JSON-LD on the page types where it applies, written to match the page | Rich Results Test and Schema Markup Validator return no errors, and the markup describes content a visitor can see | Outsource, then verify |
Core Web Vitals items on the page | Image weight, render-blocking assets, layout stability, font loading | Mobile CrUX p75 over a rolling 28 days: LCP at or under 2.5 s, INP at or under 200 ms, CLS at or under 0.1 | Whoever owns your templates |
Post-change measurement | A dated before-and-after baseline per page changed | Search Console Performance report, same queries and dates, position and CTR recorded before and again at 28 days | In-house |
Core Web Vitals is the row most often faked. A passing Lighthouse run in a lab is not a passing page. Google's thresholds sit at the 75th percentile of real visits, so a page passes only when 75% of real page views hit the good value, from the Chrome User Experience Report over a rolling 28-day window. Mobile is almost always the failing report, because the thresholds were calibrated around mid-range hardware. A report still showing FID uses a metric retired in March 2024.
What the AI-Search Era Added to the Deliverable
Three of the pages ranking here mention AI. ewrdigital.com puts "AI Optimization" in its title and never delivers it: no citation method, no AI Overviews, no measurement. GoBlueMedia sells AI work as a separate service line, and ClearLead's July 2026 update contains no AI, LLM or citation content at all.
Plan by query type, because the audience is uneven. Roughly 48% of Google queries triggered an AI Overview by March 2026, up from 31% in February 2025. Prevalence runs from 39% to 71% on informational queries, 62% on comparisons, 48% on commercial investigation and 34% on transactional. Because detection methods differ, plan against a 20% to 50% range rather than one figure (Omnibound, 2026). Almost 80% of AI-Overview-triggering keywords sit in the 0 to 40 difficulty range (Semrush, "26 AI SEO Statistics for 2026").
Citation is no longer a by-product of ranking. Only 17% of AI Overview citations come from pages in the organic top 10, down from 76% in mid-2024 (BrightEdge, February 2026). You can be cited without ranking and rank without being cited, which makes page-level extractability a separate asset. Cited brands earn 35% more organic clicks than uncited ones (Seer Interactive, 2026).
What extractable means in the markup. Sections that answer their own heading in the first sentence, claims with names and dates attached, and tables. AI-cited pages average 2,290 words, roughly three times the typical page, and 40% of the most-cited pages include a table against 28.2% in the bottom half (authoritytech.io, 2026). Treat the direction as sound and the decimals as soft: parts of that corpus are industry-funded and not peer reviewed.
The schema correction two competitors need. seo.co and ClearLead still list FAQPage and HowTo among the types they implement to qualify for rich results. FAQ rich results stopped appearing in Google Search for every site on May 7, 2026, and HowTo rich results were fully deprecated on desktop in September 2023.
The honest answer has a second half no page here states. FAQPage still correlates with AI citation: a +45% citation lift against pages with no FAQ signal, and 68% of AI-cited pages carry schema against 38.5% of the general web (authoritytech.io, 2026). Implement FAQPage for extraction, not for the dropdown. Schema on weak content produces no lift; on well-structured content it compounds what works, which is why schema is an amplifier rather than the driver two vendors here sell it as.
The concept layer sits in on-page GEO, and the procurement layer for buying AI-search work as its own engagement is generative engine optimization services. Read both before letting an on-page retainer absorb AI work at retainer rates. The organic sibling purchase is AI SEO services.
Pricing and Scope: Why the Unit Matters More Than the Number
Only one page ranking here publishes a per-page price, and it is one agency's rate card rather than a market average. ClearLead Digital, as of July 27, 2026: $150 to $500 per page for one-time optimisation, $300 to $1,500 for an audit, $500 to $3,000 per month for a retainer, $2,000 to $10,000+ for a full-site overhaul, and $1,500 to $5,000+ per month for bundled work. Its own warning is the most useful line in that table: be cautious of providers under $300 per month, because at that price the scope is usually too limited to produce meaningful results.
Everything else is denominated in the wrong unit. Loopex Digital's June 1, 2026 listicle, which ranks itself first, publishes Clutch hourly bands of $25 to $49, $50 to $99 and $100 to $149, and Clutch's directory lists on-page SEO at $25 to $49 per hour across 65,550 companies.
You are not choosing a price, you are choosing what the price is denominated in, and therefore what you can dispute. An hourly rate means you bought time. A retainer means you bought effort. A per-page rate with a defined page means you bought an artefact, and an artefact can be inspected and rejected.
So if the quote says "per page", ask what counts as a page before you sign.
- A paginated category showing 200 products across four URLs: one page or four?
- A product template applied to 200 products: one page or 200?
- A page that exists in staging and is not live yet: billable or not?
- Archived pages, tag pages and author archives: in scope or excluded?
A $150 rate applied to a page count you never agreed to is not a cheap rate.
How to Verify the Work, Deliverable by Deliverable
This is the section no competitor publishes, and it costs you an afternoon.
Titles, meta and URLs. Filter the Search Console Performance report to the pages that changed, 28 days before against 28 days after. You want CTR movement at a stable average position. If position moved and CTR did not, the title changed less than the report claims.
Headings and coverage. Crawl the site and extract headings. Check for exactly one H1, and that each section answers its own heading in the first sentence.
Internal links. Export the link graph and compare it against the source-to-target sheet you were handed, then find anything deeper than four clicks from the root. Screaming Frog or an equivalent is enough.
Images. Sort your crawl output by pages with missing alt text, then spot-check five images against the 125-character limit.
Schema. View source, find the JSON-LD, and read it against the rendered page. Then run the URL through the Rich Results Test and the Schema Markup Validator. A template block describing content the visitor cannot see fails exactly the test AI systems apply.
Core Web Vitals. Read the mobile numbers in the Core Web Vitals report in Search Console, or the field data section at the top of PageSpeed Insights. A provider reporting a lab score is reporting on their tool, not your page.
The baseline. Ask for the dated before-and-after file before work starts. One dated example of what a real on-page change does, from the definitional guide itself: Semrush reports that its own backlinks article moved from position 5 to position 2 after a February 2026 update and now appears in what it describes as dozens of AI Overviews (Semrush, March 20, 2026).
One caution. ClearLead also claims a 40% sales lift and doubled traffic, with no company names, baselines or dates. That is the shape of claim to discount, not to measure a provider against.
What You Can Do In-House
Most of the table above belongs to your team, and the reason is not cost. It is that the acceptance criteria are checkable with tools you already have or can use free.
Write the title and meta set yourself under roughly 50 pages: you have the character limits and the Search Console report that proves the change worked. Build the internal-link map yourself, because a crawl export plus a spreadsheet beats a retainer line item. Own the baseline, because that is what makes any outsourced engagement auditable.
Outsource three things: gap analysis at scale, because comparing hundreds of pages against their ranking competitors is genuine labour; schema implementation, because a mistake in the template layer is site-wide; and Core Web Vitals work, because it is engineering and belongs with whoever owns your build.
The tools that make the in-house half practical are compared in the best on-page SEO tools, and SEO automation covers where the routine audit work can run itself. For a small site, see SEO services for small business; choosing an agency instead of a project is covered in SaaS SEO agency. One product note: the on-page SEO optimizer runs the audit side of that list in the same project as your content and analytics, turning the baseline into a by-product. Allable is Free forever on 300 credits a month, Pro at €37/month (€31/month billed annually) and Business at €107/month (€91/month billed annually).
Bottom Line
On page SEO services are the one purchase in this category you can fully inspect: eight deliverables, an acceptance criterion and a verification step for each, and the dated baseline file requested before work starts. Name the unit of pricing, define what a page is, and set the measurement window at 28 days for Core Web Vitals and longer for conversion. A proposal that comes back with all of that is evidence; one that comes back with a list of activities and a monthly number is vocabulary.
Which four of these eight deliverables will your next vendor's report show you proof of, and how will you check it?
Frequently Asked Questions
What are on page SEO services?
The work done to the elements a single page controls: title tags, meta descriptions, heading structure, content coverage, internal links, images, structured data and page-level Core Web Vitals. They exclude link building, paid search, social content and site redesign. The label describes a scope, not a standard, so a proposal tells you nothing until you see the deliverable list.
How much do on page SEO services cost?
The only published per-page range in the top results comes from one agency's rate card, ClearLead Digital as of July 27, 2026: $150 to $500 per page for one-time optimisation, $300 to $1,500 for an audit, $500 to $3,000 per month for a retainer and $2,000 to $10,000+ for a full-site overhaul. Treat it as one agency's figures, not a market average. Other pages quote Clutch hourly bands of $25 to $49 per hour, which is the wrong unit.
What is included in on-page optimisation?
Eight things: a title, meta and URL set, a heading hierarchy restructure, a content coverage and entity gap analysis, an internal-link map with anchor text, image optimisation, schema markup implementation, the page-level Core Web Vitals work, and a post-change measurement with a dated baseline. Ask for all eight by name, and ask for the acceptance criterion on each.
Is on-page SEO enough on its own?
No. On-page work cannot fix a page blocked from crawling or rendering, and it cannot build the external authority that makes a competitive query winnable. What it does is turn a page with a fair chance of ranking into one that ranks, and it is the only discipline in SEO where every deliverable is inspectable. On-page is genuinely sufficient on queries in the 0 to 40 difficulty range, where almost 80% of AI-Overview-triggering keywords sit (Semrush, 26 AI SEO Statistics for 2026).
How long does on page SEO services take to work?
The only on-page-specific timeline in the top results comes from ClearLead Digital as of July 27, 2026: most sites begin seeing measurable results within 4 to 12 weeks, quick wins on a page already ranking on page two can show movement in 2 to 4 weeks after Google re-crawls, and site-wide improvements take 3 to 6 months.
