What 9 SEO Metrics Mean Before You Change a Product Page

Brand promotion: this guide explains Guangsuan's SEO measurement notes for a fund analysis desk. It does not act as an outside audit.

A fund analyst about to revise a product detail page often opens Search Console and sees impressions, clicks, CTR and position side by side. Those four numbers feel like one verdict on the page. They are not. Each number is built under different rules, and GA4 arrival data is built under another set of rules again. Revise the page before fixing the reading rules, and the team can spend a sprint editing copy while the real block sits in recording, eligibility or segment mix. Method detail follows Guangsuan's 9 SEO metrics guide. Where repair touches templates or many URLs, whole-site scope is set out in the Guangsuan Google SEO service.

Short answer for a team about to touch a PDP

The nine groups describe separate stages: whether a result was presented, whether it was picked, whether arrival led to a contact the desk accepts, and whether the page was eligible to appear with supporting signals recorded correctly. An uplift in one stage is evidence for that stage only. Before any rewrite, freeze property, search type, one finished date span, country, device and brand split, then read the stages in that order. If Console holds data while analytics sits near zero, restore capture first. If presented volume grows while combined pick rate falls, inspect query mix before touching titles. If picks grow while accepted contacts stay flat, inspect page task, form path and qualification rules. If a priority page is outside the index, repair the proven block and log a recheck.

Freeze the reading rules first

Use one property, one search type and one completed span for every export. A partial last day cannot sit beside a completed month. Write down language, country, device, brand against non-brand and landing page on each file. Arrange queries by product line, use case and buying question rather than sorting one mixed download by clicks.

Search Console reports presentation and picks inside Google Search. GA4 reports observed activity after arrival. Zones, coverage, consent handling, tag firing and which canonical received credit can separate the two totals, so the team should not force them to match. For session-based acquisition, keep session grouping apart from first-visit grouping because each replies to a separate question. Average position is a presentation-weighted aggregate for the view picked. It is not a single live rank seen by each searcher. A third-party keyword tally rests on its own database, market and collection dates, so two tools can differ without either being wrong.

What the nine groups track on a product page

Think in four checks rather than nine isolated scores.

1. Was the page presented, and for which queries

Impressions, clicks, query spread and position belong here. Find the landing pages, countries and devices behind the shift. Confirm capture before judging seasonality, stock, ranking or wording. A smaller organic share can also reflect larger paid spend. For queries, keep conditions fixed. Look at representative queries that slipped or that open a useful gap. Ask whether current results lean to product, guide, video or local format, then ask whether the page serves that same task. Do not copy markup seen elsewhere only because it appears in source.

2. Was the result picked among similar results

Pick rate is picks divided by presentations, times 100. As a hypothetical teaching illustration only, with invented figures, 20 picks from 1,000 presentations equals 2%. That line shows arithmetic. It is not a pass mark, not a client result and not a target across intents. Compare only inside matching intent, position band, device, market and page. Added low-position presentations can drag a blended rate down while total picks climb. Where the same query, page, country, device and position band shows many presentations but few picks, read the title link and description actually shown before editing the body. State product type and task plainly and describe real page matter. Do not insert an unverified certificate, speed figure or benefit claim to win picks.

3. Did arrival lead to an accepted contact

A chat tap, finished form, download, basket action and paid order are separate outcomes. Test the live event, then separate tests, repeats, spam and genuine enquiries in the desk log. As a rule, queries cannot be tied one by one to a single order. Name the formula in use. One ratio uses sittings with a key event divided by sittings. Events divided by sittings is a separate ratio when one sitting holds several events. Review landing-page sittings, sittings with the wanted event and business-confirmed contacts together. Check the mobile form, contact route and de-duplication.

For bounce, apply the GA4 definition fixed in the property. It is not the old single-page-session definition. Default engagement covers lasting over 10 seconds, holding a key event, or at least two page or screen views, subject to property setup. A one-page stay can satisfy the reader need. Repeated page_view hits can inflate engagement, so verify capture first. For duration and depth, link focus time and later steps to task completion. Four pages to locate a spec is not necessarily better than one. Added length can mean study or confusion. Supply the missing spec, condition, worked check or next step rather than adding headings or images by quota.

4. Was the page eligible, and were supporting signals read correctly

Hold a defined list of canonical pages planned for indexing and handle priority pages first. Login areas, parameter copies and matter held out by choice remain in crawl space, so coverage of every discoverable URL is not a useful aim. Group Page indexing entries by reason and inspect samples. Asking for indexing does not force admission. A crawled-but-not-indexed state alone does not establish thin copy, and a correctly merged duplicate is not by itself a removal job. Repair proven blocks such as accidental noindex, blocking, wrong canonical target, unavailable page or missing rendered matter.

Total links and referring sites are separate counts. Coverage, refresh timing and de-duplication vary by crawler, so an absent row in one report does not establish that a live link vanished. Google's Links report is not a complete current indexing test for each source page. Where link help is used, reconcile source URL, target URL, anchor, attribute, availability and check time against the delivery list. Going live, entering the index and later movement are separate stages. Third-party authority scores come from separate vendor models and are not interchangeable. Compare peers inside the same tool and time window while reading real pages and outcomes.

Read country and device before picking the page

The desk should work from its own Search Console device split. For September 2026 worldwide, StatCounter reports mobile 60.04% and desktop 39.96% as a global traffic share. That dated figure describes worldwide traffic by device. It does not describe the fund desk industry, country or buyer, and it shifts by country and year.

Worked check the desk can repeat: fix one product group, one country and one device. Open Pages to locate a URL where picks fell. Open its query list with presentations, picks and position together. Tag new, existing and off-task queries. A row with very few presentations can swing on one pick, so do not sort by lowest rate alone. As a hypothetical teaching illustration only for one fixed scope, mobile 300 presentations with 9 picks equals 3%, desktop 300 presentations with 21 picks in the same scope does not by itself show desktop copy is stronger. Check whether mobile queries seek fast specs while desktop queries seek comparison, and whether the mobile title truncates or the spec table is hard to tap on a phone.

Pattern on matched scope	First check	Sensible next action
Console holds data, analytics near empty	Tags, consent, stream, filters	Restore reliable capture before judging demand
Presentations up, blended rate down, picks up	Query and position mix widened	Compare close segments before editing titles
Picks up, accepted contacts flat	Intent, offer detail, form ease, qualification	Review product path, shipping and reply route
Priority canonical outside index	Access, signals, matter, discovery path	Repair proven block and record recheck

What to assemble before asking Guangsuan

Bring matter another person could use to repeat the reading: site address, target country and language, affected URLs, one completed span with filters, what changed and when, who can approve code or copy, and how a contact is qualified. Note tags, consent handling and report filters where analytics sits near empty while Console holds data. If checks point past one page, such as template canonical, a robots rule for an exact path and user-agent, consent breaking capture, or section-wide eligibility, that is the point to ask about technical repair, matter supply and external resources together. Record dates, filters, edit made, owner and what still needs verification. Keep absent data marked absent rather than filling it with another benchmark.

FAQ

Should flat presentations trigger a rewrite? No. Flat presentations mean the page is seldom offered for the query. Confirm eligibility, entry links and whether page type matches what results favour for that task, then decide.

Picks are up but accepted contacts are flat. Is wording wrong? Not necessarily. That shape points past picking to page type, offer detail, form ease and qualification rules. Keep visits and business-confirmed contacts on separate lines while checking the mobile form and reply path.