SMF·PAGEOPTIMIZER-PRO
Can a prompt replace PageOptimizer Pro?
SEO & marketing — technical audits and on-page optimization
Exhibit tracking slip
Verdict
POP's approach is deliberately granular: rather than one content score, it tells you the target phrase appears in the title on nine of ten competitors and in your H2s twice fewer times than the median. That comparison is arithmetic over pages you fetched, and it is the most actionable shape this kind of analysis takes. The unavoidable cost is the SERP call that tells you which pages to compare against, and the method itself — which elements matter and by how much — is what the subscription is really selling.
Exhibit A — The prompt
Received on31.07.2026Build an on-page comparison tool that reports per element, not as one score.
Input: a target query and the URL of your own page. Fetch the top 10 organic results through a commercial SERP API behind a swappable provider module, never by scraping. Exclude your own URL from the competitor set if it appears.
Fetching and parsing: fetch every competitor page and your own politely, honouring robots.txt. Parse each into distinct elements rather than one blob of text:
- title
- meta description
- H1
- H2 headings, as a list
- H3 headings, as a list
- body text, after boilerplate removal — navigation, footer, sidebar and cookie banners stripped, because counting terms in a competitor's menu is the most common way this analysis goes wrong
- image alt attributes
- anchor text of internal links
- the URL slug
Term handling: the target query plus its variants (stemmed forms, plural and singular, and a user-editable list of related phrases). Count occurrences of each term in each element, per page.
The comparison table, which is the output that matters. One row per element, showing: the median count across competitors, the interquartile range, how many of the ten use the term in that element at all, and your page's count. Then a concrete instruction derived from the gap — "add the phrase to one more H2", "your title uses it twice, the median is once" — with the numbers it came from beside it.
Over-optimisation, given equal weight to under-use: flag any element where your count exceeds the competitor maximum, not just the median. A tool that only ever says "add more" produces keyword-stuffed pages, and this one should say so explicitly in the interface.
Structural comparison beyond terms: word count against the competitor distribution, heading count and depth, number of images, and internal link count. Present these the same way — median, range, yours.
Audit trail: store the fetched competitor pages with the run, so a recommendation can be traced to the pages it came from and a run can be repeated later to see what changed.
Export: the comparison as CSV and a self-contained HTML report that opens with no server, since the report is what gets sent to whoever will make the edits.
State in the README that this counts what competitors do and does not know why they rank, that correlation is not the same as cause, and that the element weights the paid tools apply come from testing this build cannot reproduce.
Out of scope: generating or rewriting text, rank tracking, backlink data, and a browser extension.Opening prefills the prompt — press enter to run it.
Exhibit B — What you lose
- B.1 the SERP data identifying which pages to compare against
- B.2 the weighting method — which elements matter, tuned against real outcomes
- B.3 the browser extension that runs the comparison on a live page
- B.4 the ongoing research behind the recommendations as search changes
Exhibit C — Why people still pay: crawl scale, rule depth, and operational polish
Because the arithmetic is trivial and the method is not. Knowing which elements to count, and how much a difference in each one is worth, is the accumulated result of testing that a personal build has no way to reproduce.
Questions
Can I import my POP reports?
POP exports its recommendations as PDF and CSV, which are readable but not consumable — this build recomputes from its own fetch. Re-running an analysis for a page costs one SERP call, so there is little reason to import anything.
How is this different from a single content score?
A score tells you to try harder; an element table tells you which H2 to edit. The granularity is the whole point, and it is also why the tool has to parse pages into real elements rather than concatenating everything into one text blob.
What does it cost to run?
One SERP API call per analysis — a fraction of a cent — plus free page fetches. Analysing a hundred pages costs well under a dollar against a monthly subscription, which is the strongest financial case in the on-page category.
What is the one thing that does not survive the rebuild?
The weighting. POP's recommendations reflect a view about which elements matter and by how much, developed through testing. Yours reports raw differences with equal weight, which means it will sometimes send you to optimise something that never mattered.
Related tools
Receipt