Class.com sells virtual-classroom software at scale. Complex site. Legacy attribution. This is how iExcel rebuilt the search, analytics, and AI-visibility foundation underneath it.
Class.com sells virtual-classroom software to school districts, universities, and enterprise L&D teams — a Zoom-integrated learning environment used at scale, with a buyer set that spans K-12 procurement, higher-ed IT, and corporate training leads. The brand is established. The category is competitive. The organic-search surface, though, had grown the way most large sites grow: fast, layered over acquisitions and rebrands, with instrumentation that never quite caught up with the scale of what was shipping.
The engagement began with a diagnosis, not a strategy deck. Rankings had softened. AI Overviews were appearing above the fold on queries where Class historically owned the click — and the site wasn't showing up in them. GA4 events were firing inconsistently. Looker dashboards were showing traffic no one could source. HubSpot was inheriting leads with "not set" in the source field. The site was working. The measurement layer wasn't.
The brief we accepted was narrow and honest: rebuild the organic-search, analytics, and AI-search foundation from the ground up. Fix what the audit finds. Optimize what the audit ranks. Position Class.com to be answered in the AI-search environment that had already started to reshape the category.
Every serious SEO rebuild starts the same way: with an audit that isn't allowed to flatter anyone. The first phase of the Class engagement was a full technical and analytics diagnosis — SEMrush site audits (subdomain-included and subdomain-excluded), a Google Search Console review against sitemaps and indexation, and a GA4 property audit against events, key events, audiences, and platform links. Every finding routed into a shared dev-and-marketing checklist that had to be closed, not just noted.
The keyword universe was rebuilt from scratch. A master keyword planner sorted queries by funnel stage — TOFU discovery, MOFU evaluation, BOFU vendor comparison — so the editorial calendar had a defensible reason to prioritize anything. Backlink research ran alongside. Sitemap and page inventory were cross-referenced against actual crawl behavior to expose the pages Google was ignoring and the pages Google was mistranslating.
What the audit produced wasn't a recommendation. It was a punch list — with owners, deadlines, and downstream dependencies mapped. Nothing shipped until it earned its way into the roadmap.
The audit surface wasn't a single tool. It was every measurement layer stacked and cross-checked — SEMrush against GSC, GSC against GA4, GA4 against HubSpot, HubSpot against Looker. Where the layers disagreed, the layers got fixed.
The technical fixes compounded fast. Once the punch list closed, the site's SEMrush health score reached 94% — clearing the 92% benchmark Semrush publishes for the top decile of global websites — as critical crawl errors fell −99% against the audit's own baseline count.
impressions existed without clicks — the classic AI-Overview cannibalization signal.Class doesn't compete against a single peer set. The virtual-classroom category overlaps with video-first learning (Kaltura), interactive online classrooms (Engageli), enterprise talent platforms (Cornerstone), webinar infrastructure (BigMarker), sales-enablement content (Showpad), and academic learning management (Anthology). Each brand pulls different queries into its footprint. Each brand cannibalizes different Class keywords for different reasons.
The competitor research phase mapped all six on the same keyword universe. Where did their footprints overlap Class's? Where were they winning Class's category queries? Where were they showing up in AI Overviews that Class was not? Every answer landed on the editorial roadmap as a targeted rewrite, a new asset, or a technical fix — never as an aspiration.
The point wasn't to look like the competitors. The point was to understand what the AI systems were already treating as the canonical answer — and then earn the right to become the answer instead.
Diagnosis without execution is just an expensive PDF. The on-page optimization program ran as a rolling sprint against the audit's punch list — with named batches, named owners, and hard completion dates. Every rewrite was benchmarked against the master keyword planner, funnel-stage-mapped, and tied back to the GSC and SEMrush baselines that governed the audit.
The category didn't get harder over the last year. It changed shape. When roughly 42% of the tracked US keywords in the Class universe now trigger an AI Overview above the classic blue-link result, the SEO conversation stops being about position #1 and starts being about being cited. Every dot lit yellow in the grid below is a query where the answer is arriving through an AI system before the user ever scrolls.
Of roughly 3,740 US-tracked keywords across the Class.com surface, approximately 1,560 are now triggering AI Overviews in Google Search — around 42% of the mapped universe.
That's the number the GEO strategy is built against. Not "rank higher." Answer the question the AI has already decided to summarize — with content structured to be the source it cites.
You can't run a multi-channel EdTech engine on dashboards leadership has quietly stopped trusting. The analytics workstream was the one that made the SEO work legible — and the one that took the most persistent, unglamorous rebuilding.
We ran a Google Analytics 2025 audit by source, medium, landing page, and campaign. We flagged the pathologies every large site inherits: sessions arriving with source: (not set), HubSpot traffic mis-attributed to referral, product-support pages inflating engaged sessions for buyers who were actually already customers. Every finding routed to a fix — a UTM template, a redirect, a GA4 filter, or an event definition — and the fix landed with an owner.
Looker Studio came alive again. Dashboards were rebuilt against the corrected GA4 property. The Google Ads + GSC + GA4 link map was audited so the three surfaces agreed on what happened. When leadership opened the weekly view, they read one number, not three conflicting ones — and the CRM/MQL map to HubSpot finally reconciled against what the site was actually seeing.
None of the SEO or GEO work above holds up without a stack leadership can trust. This is what stayed on — every workstream instrumented for source integrity and cross-surface continuity.
Site audits (subdomain-included and excluded). Keyword universe tracked weekly. Competitor footprint benchmarked against the six named EdTech peers each cycle.
Sitemap validation, coverage audit, query-level performance. Cross-checked against SEMrush to expose impression-without-click patterns — the AI-Overview cannibalization signal.
Property rebuilt against a documented event schema. Key events promoted. Audiences rewritten. Google Ads + GSC links audited so the property acts as one measurement surface.
Legacy dashboards audited and rebuilt. Source, medium, landing page, and campaign as the primary read — with the corrected GA4 property as the single source of truth underneath.
Google Ads property linked cleanly to GA4 and GSC. Conversion definitions aligned. Traffic sources reconciled so paid clicks arrive on the CRM record with source, medium, and campaign intact.
CRM/MQL mapping reconciled against the corrected GA4 read. HubSpot traffic no longer mis-attributed to referral. Every marketing-sourced lead lands with source integrity intact.
Master keyword planner sorted by funnel stage — TOFU, MOFU, BOFU. Every rewrite in Batch 1 and Batch 2 mapped back to a specific funnel-stage query, not a generic term.
Backlink profile audited. Referring properties benchmarked against the six competitors. Highest-quality targets flagged for outreach and internal-link reinforcement.
Current State + Organic + AI Overview analysis run against the 3,740-keyword universe. Every AI-triggered query mapped to a content pattern engineered to be the cited source.
Traffic volume proves reach. Conversion proves fit. Once the analytics foundation was corrected, the primary demo-request page became the clearest read on which channel was actually working — and organic search led every other source into it.
Organic visitors landing on the demo-request page converted at 3.71%, comfortably ahead of the 2.35% cross-industry average WordStream publishes for landing pages. That's the number that reframed the roadmap: the audit and on-page program weren't just closing technical debt, they were compounding into the channel already carrying the pipeline.
If you run a large, complex EdTech site — one that grew faster than its measurement layer, that inherited legacy attribution, that competes in a category where 40%+ of queries are already returning AI Overviews — the Class.com engagement is the shape of what to do about it.
Audit every layer on the record. Rebuild the keyword universe against actual buyer intent. Map the six competitors who share your surface. Ship on-page rewrites in named batches with hard completion dates. Structure content to be the cited answer in AI Overviews, not just the ranked link below them. Reconcile GA4, Looker, GSC, HubSpot, and Google Ads until the dashboards agree.
Most agencies sell rankings. iExcel builds the search + analytics + AI-visibility foundation underneath them — and stays through the quarters that make the foundation compound.
A search foundation you can attribute against. An analytics stack leadership actually trusts. A content roadmap engineered for the AI-search environment, not the one that ended in 2023. Book a 15-minute sanity check on your setup — we'll tell you what's leaking and what to do about it.
Book a 15-min sanity check →