Short answer: the sources reviewed do not independently reproduce an 80% loss of HubSpot blog traffic. Public reporting does indicate a substantial decline, but the figures describe different scopes and measurement systems. A precise percentage is only meaningful when the page set, period, metric, geography, and source system match.
Search Engine Land reported Semrush estimates falling from approximately 13.5 million organic visits in November to 8.6 million in December, a decrease of about 36%. The report also discussed a roughly 70% blog-related decline attributed to a public social post. Those figures are not interchangeable, and neither establishes an 80% loss in HubSpot’s first-party analytics.
HubSpot’s own explanation presents the decline as a combination of a deliberate move toward influence-focused content, changing informational search behavior, and continued strength in transactional search. That is a first-party executive interpretation, not a published audit. The article does not provide the query groups, Search Console exports, conversion data, page inventory, or calculation needed to reproduce the 80% claim.
Did HubSpot really lose 80% of its blog traffic?
The defensible conclusion is narrower: public sources report a substantial decline in HubSpot’s organic and blog-related traffic, but the frequently repeated 80% figure is not independently reproducible from the sources reviewed. Semrush estimates should be attributed to Semrush as reported by Search Engine Land, which discloses its ownership by Semrush. They should not be presented as HubSpot analytics data.
HubSpot says it began emphasizing influence and non-search channels as early as 2020, including YouTube, podcasts, social media, HubSpot Academy, and The Hustle. It also says informational traffic declined while transactional performance remained strong. Those statements describe HubSpot’s reported experience. The public explanation does not disclose the internal definitions, date ranges, query classifications, or conversion calculations behind them.
A large reported decline is not a verified 80% loss. Match the source, page set, period, metric, and denominator before repeating a precise percentage.
Why traffic-loss numbers disagree
First define what the number measures. A third-party estimated visit, a Google Search Console click, an analytics session, and a pageview are different observations. Search Console reports Google Search performance. An analytics platform records activity under its own session and tracking rules. Third-party tools model traffic from external data. Each can be useful without being directly comparable.
Before calculating a change, record the source system, property or account, blog URL scope, metric definition, requested dates, timezone, geography, device scope, search type, filters, dimensions, aggregation, extraction time, and data state. Also distinguish clicks from impressions, click-through rate, average position, sessions, and conversions. A lower click count identifies a change in clicks, not its cause.
| Measure | What it represents | Best use |
|---|---|---|
| Estimated visits | A third-party model of traffic | Directional historical or competitive context |
| Search clicks | Clicks from Google Search in Search Console | Google search performance |
| Sessions | Visits recorded by an analytics platform | On-site engagement and conversion analysis |
| Conversions | Defined actions such as a demo request | Business outcomes, subject to attribution limits |
Google’s Search Analytics API supports dimensions such as date, page, query, country, device, and search appearance. Aggregation choices change the meaning of a result, and detailed results may be limited or incomplete. Google’s Search Console data guidance should be checked before treating an export as complete.
Diagnose the decline before naming a cause
Start with an authorized, verified Search Console property. Compare equivalent finalized periods for clicks, impressions, CTR, and average position. Segment the result by page, query, country, device, and search appearance where useful. Then determine whether the change is sitewide or concentrated in a page group, query group, country, device, or search feature.
Use the pattern as a diagnostic lead rather than proof of causation:
- Impressions and clicks both fall: check search demand, seasonality, indexing, ranking changes, and reporting anomalies.
- Impressions remain stable while clicks fall: investigate CTR, result presentation, competing features, ads, and changes in search behavior.
- Average position falls across affected queries: review ranking-update timing, technical changes, page relevance, and competing results.
- Clicks fall while conversions hold steady: investigate whether lower-intent traffic declined or whether tracking and attribution changed.
Google recommends checking Google Trends, ranking-update timing, technical issues, indexing, and the Search Console data-anomaly log. Its traffic-drop diagnostic guidance supports a multi-cause investigation. Do not attribute a decline to AI Overviews or another AI feature from a clicks chart alone.
A repeatable measurement workflow for an SEO decline
Use the Search Console API only after authorizing access to a verified property. Submit explicit dates, search type, dimensions, filters, and aggregation. Record whether the property is a domain property or URL-prefix property, whether dates are finalized, and whether the result is grouped by page or property.
Keep each report at its declared grain. A date-and-page observation is not the same as a date-page-query observation. A property-level total is not interchangeable with a page-level result. For recurring loads, use a database-enforced unique constraint and an atomic upsert or transaction. A lookup followed by an insert is not race-safe when concurrent workers process the same data.
The following record is illustrative and is not HubSpot data. It describes one Search Console observation at a date, page, and query grain:
{
"property_id": "sc-domain:example.com",
"date": "2026-01-15",
"page": "https://example.com/blog/example/",
"query": "example search question",
"search_type": "web",
"clicks": 12,
"impressions": 240,
"ctr": 0.05,
"average_position": 7.2,
"aggregation_type": "byPage",
"data_state": "finalized",
"extracted_at": "2026-02-05T12:00:00Z"
}If the selected dimensions include device, country, or search appearance, those fields must also be part of the declared grain and uniqueness key. If a taxonomy is added later for intent analysis, store it in a derived classification table or include its version in the derived record key. Do not imply that taxonomy fields are supplied by the Search Console API.
Search Console applies row limits and may omit detailed page-query combinations. Treat an API extract as a limited report at the requested grain, not a complete census of every query.
Classify informational and transactional performance explicitly
HubSpot says informational traffic declined while transactional performance remained strong. The public explanation does not publish the query groups, periods, conversion figures, or calculations behind that comparison. Present it as HubSpot’s reported experience, not as an independently verified result.
Google does not provide a universal informational-versus-transactional intent field in the Search Console API. A publisher must create and version its own taxonomy. Use page templates for clear product, pricing, demo, comparison, integration, or signup pages. Apply deterministic query rules when an explicit product or action marker makes intent clear. Send ambiguous phrases such as “CRM” to human review because the same query can express research, navigation, or purchase intent.
Store the original query, intent class, classification method, taxonomy version, confidence, reviewer status, and relevant conversion event. Compare clicks and impressions using the same taxonomy version, then join to separately defined conversion records. Keep raw Search Console observations, period-level summaries, and CRM contact or deal events in separate tables. A CRM touchpoint can document an observed journey, but it does not by itself prove that a page caused a sale.
Treat “transactional” as a versioned analytical label, not a performance guarantee. If the rules change, retain the old classification and compare periods only within a declared taxonomy version.
Measure Google AI search and external AI citations separately
Google says AI Overviews and AI Mode use content from its Search index. A page must be indexed and eligible to appear with a snippet to be eligible as a supporting link, but eligibility does not guarantee display or traffic. Google also says there are no additional technical requirements or special optimizations for these features beyond Search fundamentals. Its AI features documentation says these features are included in overall Search Console Web reporting.
Google’s June 2026 announcement describes dedicated reporting for impressions from generative AI features in Search. Availability may depend on rollout and property access. The documentation reviewed does not establish a generally available public API endpoint for that dedicated report.
External-model research is a different measurement system. For each prompt execution, store the prompt and version, engine and model, locale, timestamp, complete response capture or permitted reference, and collection method. Store each citation as a separate record linked to that run, including its normalized URL and ordinal. A prompt run is a snapshot, not a trend or share-of-voice metric. Define the prompt set, model set, schedule, and denominator before aggregating.
Keep one row per Search Console observation, one row per prompt run, and one row per citation. A citation key must distinguish multiple citations in the same response, for example by using prompt_run_id plus citation_ordinal. Do not merge a citation record with daily traffic or attribute an aggregate AI visibility measure to an individual analytics user. OpenAI’s documentation describes citation-bearing search answers in ChatGPT, but it is not a general citation-monitoring API specification.
Turn a measured decline into an editorial or business decision
Once the affected pages and query groups are clear, compare the reader’s task with the page’s purpose. Google’s people-first content guidance emphasizes an intended audience, first-hand expertise, useful original value, satisfying coverage, and a clear site purpose. This is editorial guidance, not a ranking guarantee, and Google states that E-E-A-T is not a single ranking factor.
Depth and differentiation are practical editorial choices, not proof that generic content cannot rank. Generic pages may face greater differentiation pressure when search interfaces answer basic questions directly, while original research, first-hand experience, and useful specificity give readers stronger reasons to choose a page. Transactional queries can also be affected by demand, competition, ranking changes, SERP features, and technical issues, so they should be measured rather than assumed to be insulated.
Use the observed failure mode to choose the next action:
- Demand and impressions fall together: investigate seasonality and market demand before rewriting pages.
- Impressions hold but CTR falls: review relevance, titles, snippets, SERP features, ads, and competing results.
- Position weakens: inspect affected pages, queries, technical changes, site focus, and update timing.
- Traffic holds but conversions fall: check intent match, landing-page experience, offer quality, qualification, and tracking.
Connect organic landing pages to defined conversion events and, where useful, CRM outcomes. Keep attribution definitions explicit. A recorded content touchpoint can be evidence of influence within a defined journey, but a single anecdote cannot establish causation, pipeline impact, or return on investment.
- Confirm the affected audience, page set, query group, and comparable period.
- Name the observed failure mode: demand, impressions, CTR, position, sessions, or conversion.
- Check that the page satisfies the reader’s task and adds useful, original value.
- Assign an owner and define the post-change metric, attribution rule, and review period.
Bottom line: the public record supports a substantial reported decline, but it does not settle the exact 80% figure or establish HubSpot’s total business impact. A credible diagnosis begins with comparable first-party measurements, preserves each record’s actual grain, separates observed changes from possible explanations, and connects editorial action to a defined follow-up metric.