Skip to content

Read 9 SEO Metrics First, Then Decide What to Fix

For SaaS experiment leads reporting to management: how the nine SEO groups are counted, which layer to inspect when impressions are high and clicks are low, and how to keep before-after views comparable.

TWritten byTyrell Lab From theTyrell Lab journal

Brand promotion: this article introduces Guangsuan's SEO measurement guidance and services for SaaS experiment teams.

When an activation test owner must brief managers, the hard part is not the chart. It is deciding whether the next step is to rebuild a landing page or to repair how results are read. The nine SEO groups below record separate jobs, so a shift in one total stays evidence for that job and does not by itself prove the programme worked.

For definitions see Guangsuan's 9 SEO metrics guide. For full-site technical repair, content supply and outside resources see Guangsuan Google SEO service.

Group the nine numbers by the question they answer

Freeze one observation frame for every export: same property, same search type, same completed period, same query set, same URL, same country and same device split. Note language, brand against non-brand, and landing URL on the file. Leave an absent row marked absent.

  • Arrival from Search versus arrival on site. One report shows shows and picks inside Search; site analytics shows what could be observed after arrival. Zone offsets, scope, consent, tag firing and canonical assignment can pull the two totals apart, so do not force them to agree.
  • Query cover and mean position. Mean position is a show-weighted summary for the chosen view, not a single live rank. Outside keyword totals rest on their own lists and dates. Sort queries by product family, use and buying question, then check whether the ranking URL serves that job.
  • Pick share. Picks divided by shows times 100 gives the share. As a teaching illustration, 20 selections from 1,000 shows is 2%, which shows the maths rather than setting a target. Added low-position shows can drag a blended share down while total picks still climb.
  • Contacts and ratios. Test that the live event fires, then keep trials, repeats, spam and business-confirmed contacts on separate lines. Name the formula you use, because sessions holding an event divided by sessions answers a different question from events divided by sessions when one visit holds several events. Query lines as a rule cannot be tied one by one to a single order.
  • Bounce and focus signals. In GA4, bounce is the share of sessions without engagement, the reverse of engagement share. Confirm your own property setting. One answered question on a single view can satisfy the reader, while duplicate hits from a setup fault can inflate the figure.
  • Time with page in focus. Longer time can mean careful study or confusion. Finding a tolerance on one page is not worse than needing four pages. Read time with later pages and task completion together.
  • Eligibility to appear. Keep an explicit list of canonical URLs scheduled for indexing and rank important entries first. Sort indexing entries by reason and open samples. Requesting indexing starts a process and does not force entry. A crawled but currently not indexed note alone does not prove the copy was thin. Keep canonical for duplicate or highly similar material.
  • Outside links. Total links and referring sites are separate counts. Coverage, refresh timing and de-duplication vary by crawler. Google's Links report cannot serve as a live check that each source page is indexed now. Where outside help is used, match source, target, anchor, attribute, availability and check time against the delivery list. Going live, entering the index and later ranking better remain three separate stages.
  • Vendor scores. Scores from different vendors use different models and cannot be swapped. Compare close peers inside the same tool and time window while still reading real pages and outcomes.

High shows, few picks: inspect in this order

Do not rewrite the heading first. Work from capture to eligibility to wording.

  1. Capture. When Search holds steady and analytics sits near empty, check tags, consent handling, data stream and report filters before judging demand. Then note which URLs, countries and devices moved.
  2. Eligibility and page fit. Start from the Pages view to spot the URL, then drill to its queries with shows, picks and mean position side by side. Tag new, continuing and off-task queries. For one representative query open the live result and note product, guide, video or local favour, then ask whether your URL serves that same job. When the query asks for model, delivery or compatibility and the ranking URL only explains ideas, add product or comparison material and an enquiry entry first.
  3. Wording. Change the title or snippet-relevant passage only when the same narrow set shows many shows but few picks and the shown title misstates the answer. Describe only genuine page material. Do not add an unverified certificate, test claim or performance figure to win picks.

Keep the before and after view comparable

Use the same completed dates on both sides and leave the still-growing today out of the comparison. Write down the filters, the edit, the owner and the open items. This keeps the activation baseline usable for the next experiment review.

ObservationCheck firstNext action
Search steady, analytics near emptyTags, consent, streams, filtersRestore reliable capture before judging demand
Shows up, blended share down, picks upQuery and position mixCompare narrow sets before rewriting titles
Picks up, confirmed contacts flatIntent, market, forms, qualificationWalk the product path and read actual enquiry records
Important canonical outside the indexAccess, signals, material, discovery pathFix the proven block and log a recheck

What to prepare before asking Guangsuan for help

Guangsuan keeps quality-quantity, complementary and layered building as its backlink approach. Bring facts another person can repeat: site address, target country and language stated separately, affected URLs, one completed export with filters, what changed and when, who can approve copy or code, and how your team qualifies a contact.

For people-first checks, see Google guidance on creating helpful, reliable, people-first content.

FAQ

Do flat shows require a rewrite? No. Flat shows mean the page is seldom offered for the query. Confirm eligibility, entry links and whether the page type matches what results favour first.

Does a falling blended share force a title change? Not alone. Check whether new low-position queries widened the mix, then compare narrow sets before editing.

Does longer time prove success? No. Longer time can mean careful study or confusion. Add the missing condition or next step and check whether the task was completed.

Stop running experiments you can't defend.

Bring us a stuck funnel or a backlog of untested ideas. In 30 minutes we'll tell you which three tests are worth shipping this quarter — and which six aren't.