SEO Task | Jinnat Ul Hasan

Diagnosis · Rebuild · Scale

90% gone in
two months.
One campus.

Jinnat Ul Hasan

01 / 06

Part 1 · Diagnose

Before diagnosing the loss,
prove the loss.

GA4 data depends on consent and tags. Search Console data doesn't use them. So if GSC impressions and clicks both dropped, the loss is real. I would also use ranking tools to check for any ranking fluctuations, as further confirmation.

01

Real loss

Clicks and impressions both collapsed

A real ranking or indexing loss. The pages have stopped being served.

This is the cliff profile we are seeing here, so the next check is Index Coverage and URL Inspection on the campus path to see whether the pages are still eligible at all.

02

CTR issue

Impressions flat, clicks down

A SERP-level CTR issue, not a ranking loss. The pages still rank and still get served.

Usually an AI Overview or another zero-click feature answering the query above the result, a rewritten title or snippet, or a competitor winning the click with a richer listing.

SERP features are query-level though, so they would hit the sibling campuses too. That argues against this being our cause.

03

Not organic

GSC stable, GA4 down

A measurement problem, not an organic one. Nothing was actually lost.

The fix belongs to the tagging or consent setup rather than to SEO, and the checks below confirm which of those it is.

Ruling out measurement

Consent banner or CMP update blocking GA4
Tag removed or misfired on a redeploy
Property, filter or date configuration changed
Channel reclassification: organic reassigned to direct or referral
Bot filtering rules changed

02 / 06

Part 1 · Top Hypothesis

My top hypothesis isn't competition.
It's that the page stopped being eligible.

Branded search is the tell: nobody outranks a school for its own campus name. So I start from what could switch one campus off while its siblings carry on. If Marketing says nothing changed on their side, I believe them, because platform and template deploys rarely belong to the team.

01

Primary

Campus-scoped indexability failure

A templated change applied to this campus's section only: a noindex, a canonical pointing at a sibling, a robots.txt disallow, or a URL change shipped without redirects.

It is the only one that explains the whole pattern, so I would start here.

02

Secondary

Google Business Profile suspension or merge

This explains the local-intent half specifically, and it would not touch the siblings either.

I would run it in parallel rather than in sequence, because the dashboard answers it in two minutes.

03

Easy to rule out

Manual action or security issue

A partial-match manual action can hit one URL path while the rest of the domain carries on, and spam injected into a single campus section does the same.

I would not expect it on a school group, but Search Console answers it outright, so I rule it out before spending time elsewhere.

04

Not the cause

Backlink profile or negative SEO

New and lost referring domains against the drop date. Lost citations from a replatform would be a real finding.

Toxic links get devalued, not penalised, and they never touch the GBP. Neither branded nor local runs on links.

How I'd confirm or kill each

Index Coverage report and a site: query on the campus path
URL Inspection: indexability, declared vs Google-selected canonical
Server logs for a crawl-rate drop, and its exact date
Wayback and the platform deploy log against that date
GBP dashboard status and Maps visibility
Manual Actions and Security Issues tabs in Search Console

03 / 06

Part 2 · Keyword Targeting

Three buckets, ranked by
proximity to an enrolment.

Recovery is never instant. Pages have to be recrawled, reindexed and re-evaluated, and a full return was never guaranteed, so I would give it weeks before reading anything into it. If the terms are still missing after that, the page was never earning them on its own. It was coasting on the group's authority.

01

Highest intent

Brand terms

Campus name on its own, plus fees, admissions, open day, term dates, uniform, contact.

These convert hardest and should never be contested. If the campus page isn't winning them, nothing else matters yet.

02

Growth

Local intent

Independent, private or international school plus town, district, "near me", and catchment phrasing.

This is where new families who don't yet know the school arrive. It's the acquisition layer.

03

Qualifier

Curriculum terms

IB, A-Level, British curriculum, early years, sixth form, always paired with a location modifier.

Unpaired curriculum terms belong to the group, not the campus. That distinction prevents cannibalisation later.

How I'd build the list

Start from the campus's own pre-drop GSC impressions: terms it has already proven it can win
Add the local-intent gaps it never covered, sized with a keyword tool
Check who owns each SERP: local pack, organic, or a sibling campus
Pair every curriculum term with a location before targeting it
Sequence the work against the admissions cycle, ahead of the intake window
Benchmark recovery against the pre-drop query set, so new terms don't disguise the ones still missing
Judge it finally on enquiries and open-day registrations from organic, not on positions

04 / 06

Part 2 · The Rebuild

I'd rebuild it as the campus,
not as another template.

The campus's main admissions landing page runs the same template as every other campus in the group. I would rebuild it to earn these terms itself, for a parent and for a crawler.

01

The first read

Title, H1 and meta

All three carry the campus name, the town and the curriculum. Never a bare "Admissions", which is what the template gives every sibling.

Title Admissions | [Campus] School, [Town] | IB & A-Level
H1 Admissions at [Campus] School, [Town]

H2s then take the questions in order: entry points by age, process, fees, open days.

02

The substance

Content and next steps

Admissions process, fees, open days, catchment and outcomes, high on the page rather than buried. Length follows the content, never a word count.

Named head teacher and admissions staff with real bios. Parents choose on trust, so E-E-A-T earns its place here.

Local proof for parents: address, transport links, feeder areas, community ties.

The rebuild is the moment to fix signposting: open day, prospectus, or call. Fees published rather than gated, so nobody returns to the SERP for them.

03

The signals

Structured data

EducationalOrganization and LocalBusiness, with the name, address and phone identical to the Google Business Profile.

FAQPage on the admissions questions, and Event markup on the open days.

This is what re-establishes the campus as its own entity after a canonical or indexing fault, so search resolves the campus rather than the group.

Internal links, so the group points at the campus

Links from the group hub using the campus name as anchor text, not "read more"
Breadcrumbs that place the campus correctly inside the group hierarchy
Audit sibling cross-links: nothing should blur which page belongs to this campus
A self-referencing canonical on the campus page, re-checked after every deploy
Campus URLs in the sitemap, with redirects from anything that moved

05 / 06

Part 3a · Off-Page & Local Authority

Don't build authority from scratch.
Stop leaking what's already there.

The campus already sits on a domain with authority behind it, so I would not start by building more. Two separate jobs follow from that and they get conflated: off-page links that build authority, and local signals that build proximity and prominence.

01

Authority

Where school links actually come from

Local and regional press: results days, new facilities, head teacher commentary, community events
Sector directories and guides: Good Schools Guide, ISC, Which School?, and accreditation listings on COBIS and CIS
Feeder nurseries, sports clubs, charities and sponsorships. Real relationships that already exist and simply aren't linked
University destination pages and alumni mentions
Digital PR the group runs once and distributes: admissions-trend or results data stories

02

Proximity

Local signals that put the campus on the map

Google Business Profile: categories, photos, posts, Q&A, and a steady review velocity rather than a burst
NAP consistency across every citation and directory
Localised on-page content and schema, matching the profile exactly
The local council's "schools in [city]" page: one authoritative listing worth more than a long tail of thin directories

The strategic point

Consolidation, not acquisition: correct canonicals, internal linking, and locally-earned links pointing at the campus URL rather than the group homepage, so the authority the group already has resolves to this campus.

06 / 06

Part 3b · Detection At Scale

Don't check ten sites.
Query all ten at once.

If it happened here it can happen anywhere on the platform, so I would automate the detection rather than run the audit again. It has to distinguish a real fault from siblings competing with each other.

01

Detect

Two jobs I'd schedule, so nobody checks by hand

Daily Money URLs, sampled under load

Watch about twenty URLs a campus: status and final URL, canonical target, meta robots and X-Robots-Tag, robots.txt, H1 fingerprint
Fetch as Googlebot from a neutral IP and as a browser, comparing content rather than status. A bot block returns 200 with a challenge in the body

Weekly Full sweep, and the data behind it

Crawl every campus in full: indexability, sitemap reconciliation, redirect chains, group-hub links, hreflang, and AI crawler access agent by agent
Measure each campus against its own trailing baseline with the group trend as the control, so the admissions cycle doesn't trip it

02

Diagnose

Real problem, cannibalisation, or not SEO at all?

Signal	Real problem	Cannibalisation
Impressions	Disappear	Hold steady
Clicks	Disappear	Move to a sibling
Query to page	No page serving	Wrong campus serving
Local pack	Campus absent	Sibling shown instead

GSC query-to-page data settles the organic rows in one look. The local pack sits outside GSC and has to be checked separately. And read GSC against GA4 before either column: GSC is Google's view, GA4 is the site's, so sessions falling on steady clicks means tracking or delivery, not SEO.

The 08:00 report, exceptions only

Python owns the checks and the severity. An LLM correlates crawl, GSC and logs, ranks what matters, and can escalate but never downgrade.

The detection SLA

24 hours on key pages, seven days on everything else. The weekly report sends even when clean, so silence isn't a dead pipeline.

If it's cannibalisation

Fix by intent separation. Each campus owns its location and brand terms, group pages own unqualified curriculum terms.

Questions?

Jinnat Ul Hasan