Skip to main content
The 2026 changelog covers Cited’s documentation launch and the platform changes that followed — the March and May 2026 Cited Index editions, expanded coverage of Google’s AI surfaces, plan-tier composition updates, the MCP server, the redefinition of Citation Rate, and the documentation of every dashboard surface.

August 2026

Citation Rate redefined, and Citation Share named — 14 August

Citation rate now means: of the answers that mention your brand, the share that also cite your own site. It previously meant “answers carrying at least one citation to any domain, out of all answers” — a property of the platform and the category rather than of your brand. Measured across tracked brands, the old metric sat at 83.5% platform-wide and spanned only 70–97%; the new one spans 0–70% on the same brands, because it is finally measuring something about you. Values from before and after 14 August are not comparable. Anything you exported or recorded under the old name measured the old formula. At the same time, the metric the dashboard had been showing as its headline citation number was given its own name — citation share, your domain’s share of all citations in your category. The two now sit side by side across Overview, AI Platforms, Prompts & Responses and Topics, each with its formula stated, because they differ only by denominator and are close to uncorrelated in practice. Full explanation: Citation rate.

Per-competitor Citation Rate — 14 August

Citation rate is now computed for each tracked competitor, so you can ask “when an AI names them, does it link them?” — a read on how much of a competitor’s position rests on their own site versus third-party coverage. A competitor with no declared domains shows an em dash rather than a zero. A confident 0% would read as “nobody links them” when it actually means “we do not know what they own.”

One canonical definition per metric — 17 August

Every metric now carries one definition everywhere it appears, in two parts: a plain-English line saying what it is, and the formula. Previously the same four metrics carried four to six different wordings across Overview, AI Platforms, Prompts, Competitors and the platform grids. The shape is borrowed from Google Search Console, which states a formula for the one metric that is a ratio rather than trying to carry a denominator inside a sentence. That also let us delete the “different denominator, not comparable” warnings three surfaces had grown — two formulas side by side make the difference visible without them.

Every dashboard surface documented — 15–17 August

Source Intelligence, Topics, AI Shelf, Tasks and Agents each have a documentation page for the first time, under a new Dashboard section. Each covers what the surface answers, what every number on it means, and how to act on it. The Source Intelligence page includes the drill reading model — why source and page tabs are compositions to read downward while platform, topic and prompt tabs are comparisons to read across. That distinction is the origin of most “these numbers don’t add up” questions.

Intent taxonomy corrected — 17 August

Two pages described a four-type intent taxonomy — informational, commercial, navigational, transactional — that Cited has never used. The real classification has six types describing how a customer framed the question: problem-first, context-specific, budget-anchored, comparison, recommendation-seeking, and feature-curious. Those are the tags in your dashboard, and they now match the docs. A strategic claim built on the wrong taxonomy was also removed. It asserted that being cited on informational prompts lifts share of voice on later commercial prompts. Tested against tracked brands, the gap between exploratory and decision intents runs in both directions and is brand-specific, so there is no general rule of that shape. Intent taxonomy now shows how to read your own split instead.

”Query” retired in favour of “prompt” — 17 August

Pages using “query” for something a customer asks an AI now say “prompt”, and four URLs changed to match: Old URLs redirect. “Query” is kept where it is genuinely the right word — Google’s query fan-out, traditional search, URL query strings, and “Query Deserves Freshness”.

July 2026

Cited MCP — 11 July

The Cited MCP server lets an AI client (Claude Desktop, Claude Code, Cursor, ChatGPT, Gemini CLI) query your AI-visibility data directly, over the open Model Context Protocol. Ten tools: resolvers for discovering your brands and dimensions, conformed metric tools over Cited’s metrics cube that return the same numbers as the dashboard, and read tools for AI Shelf and the Cited Index. Claude Desktop connects by signing in — no key to manage. See Capabilities for the tool list and Recipes for prompts to start from.

June 2026

Docs site refresh — June 2026

A full refresh of the Cited documentation site landed in June 2026, aligning every non-signals page with the platform state as of mid-June: expanded platform coverage (Google AI Overviews and AI Mode tracked alongside ChatGPT, Perplexity, Gemini, Claude, and Grok), updated plan-tier compositions, current AI Narrative cadence, refreshed dashboard tab structure, and a subscriber-audience voice pass across Get Started, Concepts, Methodology, Playbooks, and Glossary sections.

May 2026

Cited Index — May 2026 edition published

The May 2026 Cited Index edition shipped with refreshed category composition. Three categories were marked inactive (Digital Payments, Credit Cards, Online Learning) and three new categories were added (Health & Wellness, Personal Grooming, Conversational AI Platforms). The edition covers 200+ Indian brands across 8 active categories, benchmarked using a refreshed prompt set across the four Index platforms (ChatGPT, Perplexity, Gemini, Claude).

Plan tier composition update — May 17

Anthropic (Claude) and xAI (Grok) repositioned to the Scale plan and above. Pro plan composition is now ChatGPT, Perplexity, Gemini, Google AI Overviews, and Google AI Mode (5 platforms). Starter retains ChatGPT, Perplexity, and Gemini (3 platforms). Scale and Custom plans continue to cover all 7 platforms including Claude and Grok.

AI Narrative cadence — May 5

All plans moved to a single weekly AI Narrative refresh cadence (was previously twice-weekly on Pro and above, weekly on Starter). The change simplifies the refresh rhythm and aligns narrative regeneration with the rest of the weekly insight surfaces (Source Intelligence, Tasks, Recommendations).

April 2026

Brand Intent editing refinements — April 27

Removed Regenerate button from Brand Intent — positioning is now fully user-controlled after initial AI seeding during onboarding. Added persistent recompute status banner, last-edited timestamp display, and discard changes button. See Configure positioning.

Target attributes self-serve editing — April 21

Brand owners can now view and edit their target attributes directly in Settings > Brand Intent. Changes trigger automatic recomputation of recommendation Impact Scores. A “Regenerate from AI” button re-extracts attributes from the brand intelligence brief, with a 24-hour cooldown per brand. Target attribute limit raised from 10 to 20. Full guide: Configure positioning

Brand-relative Impact Scoring (v3.0) — April 20

Impact Score semantics changed from universal to brand-relative. A score of 60 now means “60% of the maximum achievable impact for THIS brand’s current data maturity” rather than a universal 0-100 scale. Changes:
  • Brand-relative denominator ensures meaningful prioritization signal for brands at any data maturity level
  • Defensive Competitive Gap logic: dominant brands scored by proximity-to-vulnerability rather than distance-to-catch-up
  • Strategic alignment multiplier (1.5x) when recommendation matches declared brand positioning
  • Low-prompt-count safeguard prevents single-prompt gaps reaching HIGH on strategic boost alone
  • QV floor raised from 1 to 3 — zero-state brands no longer hit a structural ceiling
  • Visibility-based score floor removed (was artificially flattening tier distribution)
What to expect:
  • Existing recommendations recomputed once, no ongoing churn
  • Dominant brands may see most recommendations scored LOW — this reflects genuine low competitive urgency
  • Brands with populated brand_intent.target_attributes see strategic alignment boosts on approximately 75% of recommendations
Full methodology: How we score recommendation impact

Narrative competitor accuracy — April 19

Your AI Narrative now only references competitors you have explicitly declared in your brand profile. Previously, the AI summary could reference competitors inferred from claim content, occasionally surfacing inaccurate or irrelevant names. The new behavior anchors on your declared competitor list, with a clean fallback when no competitors are declared. We have also added an admin force-refresh capability to regenerate narrative profiles on demand — useful after major claim cleanups or before sales presentations.

Docs site launch — April 15

Cited’s documentation site launched at docs.getcited.in with 74 pages of substantive content covering five sections: Get Started (5 pages), Concepts (15 pages across Metrics, Foundations, and Platforms), Methodology (8 pages), Playbooks (5 pages), and Glossary (40 terms plus index). The site is built on Mintlify with auto-generated llms.txt and llms-full.txt for AI discoverability. All content passes a 24-check quality assurance process covering citability, docs structure, MDX validity, and voice accuracy.

FAQ rendering migrated to native Accordion — April 15

All FAQ sections across the docs site were migrated from a custom FAQ MDX component to Mintlify’s native Accordion and AccordionGroup components. This provides proper expand/collapse interactivity and aligns with Mintlify’s supported component patterns.

March 2026

Cited Index — March 2026 edition published

The first edition of the Cited Index was published, covering 253 Indian brands across 8 categories (Travel and Luggage, Skincare and Beauty, Audio and Wearables, HR and Payroll, Credit Cards, Online Learning, CRM and Sales, Digital Payments). Benchmarks were computed from 740 LLM responses across 4 platforms (ChatGPT, Perplexity, Gemini, Claude) using 185 non-branded category prompts. Key finding: the median Indian brand has an 8% mention rate on non-branded category prompts. Top-quartile brands reach 16%.

Grok integration — March 2026

Grok was added as the fifth tracked platform via xAI’s Responses API with web_search tool. Grok is tracked on the main Cited platform for pilot brands but is not yet included in the Cited Index pipeline.