Skip to content
SEO与数字营销

What 9 SEO Metrics Mean Before You Change a Product Page

For ecommerce teams: what impressions, clicks, CTR, position, sessions, conversions, engagement, indexing and links each answer, how to lock country/device scope, and which PDP fix to do first.

By GetPhoto In SEO与数字营销
GetPhoto
About the author

Brand promotion: this article introduces Guangsuan's SEO measurement guidance and Google SEO service. It is a vendor introduction, not an independent review.

No single SEO number tells an ecommerce team whether to rewrite a product detail page. These nine groups of metrics answer different questions, so a change in one number is evidence about that number only. Before you change a PDP, lock the observation scope, read discovery, selection, arrival and eligibility in order, then fix the title and description, the page body, or recording and eligibility first.

For full method definitions, see Guangsuan's 9 SEO metrics guide. Where the block affects templates or many URLs, scope and acceptance for whole-site work are described in the Guangsuan Google SEO service.

What impressions, clicks, CTR and position actually mean

Search Console reports impressions and clicks on Google Search, while GA4 reports observable activity after people reach the site. A Console click is not the same as a measured session, because time zones, scope, consent, tag firing and canonical credit can differ. Do not force the two tools to agree.

Impressions count when a normal web result enters the results page the user opened. The user does not need to scroll to the link, but an unopened later page does not earn an impression only because the query happened. Carousels and expanded areas have their own rules. Impressions are therefore not search volume and not proof the title was read carefully.

Average position is an impression-based aggregate for the selected scope, not a fixed live position seen by every searcher. It is affected by result type and aggregation, so an average of 12 cannot be read as always on page two.

CTR is clicks divided by impressions multiplied by 100%. As a hypothetical teaching illustration only, 20 clicks from 1,000 impressions is 2%. That line shows the arithmetic and is not a pass mark, a client result, or a target for every country, intent or layout. Brand, procurement and questions already answered in results need separate judgment.

Lock the scope before you compare pages

Use one search type and one complete date range for every export. Do not compare an incomplete recent day with a complete historical day. Filter by target country and device, and keep brand and non-brand queries apart.

Record language, country, device, branded versus non-branded queries, and the relevant landing pages. Group queries by product, application and purchasing question. Use a repeatable order:

  1. Open Pages to find a URL where clicks or CTR fell.
  2. Open the query list for that page with impressions, clicks and average position together.
  3. Mark new, existing and irrelevant queries instead of sorting by lowest CTR alone, because a row with very few impressions can swing on one click.
  4. For important queries, check the live result with time, language, region, device and result type: normal web, product, map, video or other answer form.
  5. Leave missing data marked as missing rather than filling it with another benchmark.

As a hypothetical teaching illustration only for one fixed scope, mobile 400 impressions with 12 clicks is 12 divided by 400, or 3%. Desktop 400 impressions with 28 clicks in the same scope does not by itself prove desktop content is better. Check whether mobile queries ask for quick specs while desktop queries ask for comparison, and whether the mobile title is truncated or the spec table is hard to tap.

Read the nine groups in task order

1. Discovery: traffic, queries and position

Search Console shows visibility and selections. GA4 shows what can be observed after arrival. For session-based acquisition, use the session grouping and keep first-visit grouping separate, because they answer different questions. Locate the landing pages, countries and devices driving the change, verify recording, then check seasonality, availability, rankings or content. A falling share of organic traffic can also mean paid campaigns expanded.

For queries, keep the observation conditions fixed. A third-party ranking-keyword count depends on its database, market and collection dates, so two tools do not need to report identical counts. Inspect representative declining or promising queries and ask whether the current result page favours a product, tutorial, video or local service, then check whether your page answers that same task.

2. Selection: CTR compared like with like

Compare only within matching intent, position band, device, market and page. More low-position impressions can pull the combined rate down while total clicks rise, so inspect similar segments before rewriting a title.

If impressions are high but clicks are low on the same query, page, country, device and position band, read the actual title and summary shown in results before changing the body. State the product type and task honestly and describe genuine page content in the description. Describe what the page actually provides rather than stacking synonyms or repeating the title. Do not add an unverified certification, speed figure or benefit claim to win clicks.

If the query asks for model, delivery, price or compatibility but the ranking page only explains concepts, the wrong page is ranking. Add product or comparison material to the correct page first. If definition queries are already answered in results, add a deeper judgment, tool, exception or next step, because a bolder title alone will not help.

3. Arrival and qualification: sessions, events and real enquiries

A chat tap, completed form, download, basket move and paid order are different outcomes. Test the event on the live page, then separate tests, repeats, spam and genuine enquiries in your own log. Queries as a rule cannot be joined one by one to a single order.

Label the ratio you actually use. One common rate uses sessions with a key event divided by sessions. Counting events divided by sessions is a different ratio when one session holds several events. Check the mobile form, contact route and de-duplication, and review organic landing-page sessions, sessions with the intended event, and business-confirmed contacts together.

For bounce and engagement, use the GA4 definition for your property. A one-page visit can still answer the question. Check recording first, because duplicated page views can make engagement look artificially strong. There is no universal threshold where a rate above 70% triggers a penalty.

For time and depth, relate focus time and later steps to task completion. Needing four pages to find a specification is not necessarily better than finding it on one. Longer time can mean useful study or confusion. Add the missing specification, condition, example or next step rather than adding headings or images by quota.

4. Eligibility and trust: indexing, links and third-party scores

Maintain a defined set of canonical pages intended for indexing and prioritise important pages within it. Sites contain login pages, duplicate parameter copies and content deliberately kept out, so covering every discoverable URL is not a useful aim. Group Page indexing report entries by their reasons and inspect representative URLs. Requesting indexing does not force acceptance. A crawled-but-not-indexed state alone does not prove thin copy, and a correctly consolidated duplicate is not by itself a removal task. Fix evidenced blocks such as accidental noindex, blocking, wrong canonical targets, unavailable pages or missing rendered material.

For links, total links and referring sites are different counts. Discovery coverage, refresh timing and de-duplication differ across crawlers, so a missing row in one report does not prove a live link vanished. Where a link service is used, reconcile source URL, target URL, anchor, attribute, availability and check time against the delivery list. Going live, getting indexed and moving the target page are separate milestones, so keep delivery, indexing and later performance on separate lines.

Moz Domain Authority is a vendor predictive metric, not a public score assigned by Google. Scores from different vendors come from different models and are not interchangeable. Compare peers inside the same tool and time window while also reading actual pages and outcomes.

Use the device and country split to pick the first fix

Compare mobile and desktop from the same property, dates and market, then combine with site analytics and order or enquiry records. A hypothetical share inside one fixed scope, such as 60% mobile selections, describes only that scope and does not prove mobile is 60% of world search or 60% of orders. For a PDP where evening browsing is on phones but checkout happens on desktop, that separation stops the team from rebuilding desktop pages when the task is readability and tap targets on phones.

What you see on matched segmentsCheck firstSensible next move
Console steady, analytics almost emptyTags, consent, streams and filtersRestore reliable recording before judging demand
Impressions up, combined CTR down, clicks upShift in query and position mixCompare close segments before rewriting titles
Clicks up, checked enquiries flatIntent, market, forms and lead reviewReview product path, shipping and contact route
Key PDP not indexedAccess, signals, content and discoveryFix the proven block and log follow-up

What to bring before you ask Guangsuan for help

Bring facts that let someone else reproduce your reading: site address, target country and language, affected URLs, one complete date range with filters, what changed and when, who can approve code or copy, and how you qualify an enquiry. Note tags, consent handling, data stream and report filters if analytics looks almost empty while Console shows data.

If checks point past one page — template canonical, robots rule for an exact path and user-agent, consent breaking capture, or whole-section eligibility — that is the point to ask about technical repair, content supply and external resources together. Going live, getting indexed and moving the target page stay separate, so record dates, filters, the change made, its owner and what remains unverified. Missing data stays labelled as missing.

Frequently asked questions

Should flat impressions trigger a rewrite? Hold the rewrite. Flat impressions mean the page is seldom offered for the query. Confirm eligibility, entry links and whether the page type matches what results favour for that task, then decide.

Clicks are up but confirmed enquiries are flat — is wording wrong? Not always. That shape points past selection to page type, offer detail, form ease and qualification rules. Keep visits and business-confirmed contacts on separate lines while you check the mobile form and reply path.

Ship your first image in 24 hours.

Flat $9 per asset. No shoot fees. No retouching upcharges.

Get your first photo free