Sunrise Senior Living’s AEO Dashboard for September 2026
Welcome, Sunrise Senior Living. This is your September 2026 report on how visible your brand is to AI search engines: ChatGPT, Google Gemini, Copilot, Perplexity, Claude, and Grok. It shows where you stand today and how you compare to the four other companies in your cohort: Brookdale, Atria, Erickson, and Belmont Village. September is your first measured month, so every chart shows one point per company; trend lines start building in October. The tiles below and the links throughout take you deeper.
Your AI SuperScore and the 10 Categories That Build It
Your SuperScore sits at the center, surrounded by the ten categories that feed it. Each category shows this month’s score plus how much impact and effort improving it carries. Click any category to open its full breakdown and recommendations.
Impact reflects how much headroom a category has to lift your SuperScore; effort reflects the typical work involved. Higher scores are always better.
Where Sunrise Stands in Its First Measured Month
Sunrise Senior Living's SuperScore is 53.8 in September, 2nd of 5 behind Brookdale (69.2) and ahead of Erickson (49.0), Atria (48.0), and Belmont Village (42.0). Sunrise leads Accessibility (58.8) and Originality (62.7); is 2nd in Authority (70.1), Freshness (63.6), and Relevance (43.0); 3rd in Citability (66.8); 4th in Clarity (48.2), Structure (34.4), and Trustworthiness (73.6); and 5th in Coverage (33.4). Most of the 15.4-point gap to Brookdale sits in Relevance (8.6 points), Coverage (3.6), and Structure (2.7).
Where Sunrise Leads
Measured September values where Sunrise ranks 1st or 2nd in the cohort.
- Readability 75.15, 1st of 4 scored
- Accessibility Scores 50, 1st of 5
- Page Load Speed (LCP) 1.6s, tied 1st
- Featured Snippet Rate 9%, 2nd, 1 point off the lead
- Keyword Coverage 73,000 keywords, 2nd
- Branded Search Volume 22,200 a month, 2nd
- Internal Link Depth 2.63 clicks, 2nd
Where Sunrise Trails
Measured September values where Sunrise sits 4th or 5th, or far behind the leader.
- Rich Media Depth 19 assets vs Erickson's 309 (5th)
- CTR 2.5% vs Brookdale's 3.3% (5th)
- Schema Markup 2 types vs Atria's 7 (tied last)
- Customer Support content 3 pages vs Atria's 71
- Pages per Topic 199 vs Brookdale's 1,444
- No move-in (onboarding) content
- Sentiment 50% positive vs Belmont Village's 90%
- Return Visitor % 43.66% (4th)
The Pattern Behind the Scores
Sunrise is a well-known brand with thin depth. Where a metric measures reputation or reach, Sunrise is at or near the top: 22,200 branded searches, 63,903 ranking keywords, 100 news mentions, the cohort's most readable writing, and a shallow, well-linked site. Where a metric measures how deeply it answers families' questions or how machine-readable it is, Sunrise trails: 199 care-topic pages to Brookdale's 1,444, 3 support pages to Atria's 71, no move-in guide, 19 rich-media assets, 2 schema types, and a 2.5% click-through rate. AI engines answer ‘what is memory care’ or ‘assisted living near me’ from the source with the deepest, best-structured answers, which today is usually Brookdale.
Reputation is the second theme: Sentiment is 50% positive against Belmont Village's 90%, the Yelp listing is unclaimed at 1.0 stars, and the X account has been silent since 2025-09-30. ChatGPT and Perplexity quote review text and Reddit threads when families ask whether Sunrise is any good.
Your Biggest Win and Biggest Risk
The two categories that matter most this month: where you are furthest ahead of the cohort and should protect that lead, and where the largest gap to the cohort leader is open and worth closing first.
What’s New in AI Search This Month
These are the most recent developments across the six major AI search engines that decide who to surface and cite. Engine-level changes can shift which sources get cited even when your own signals stay flat.
Top News This Month From the AI Search Engines
Here are the most recent developments in each of the six major AI search engines for the month. To see each platform’s full strengths, weaknesses, and detailed update history, visit the Platforms page. Open Platforms ›
Your September 2026 Performance
Two quick reads on your month: how your category strengths and weaknesses compare to the cohort, and where your overall score ranks against the other companies.
Your Strengths and Weaknesses at a Glance
Your 10 category scores layered over the cohort average and over the cohort leader. Where your blue shape extends past the dashed line you are above the cohort average; where the leader’s line (shown in that company’s color) sits outside yours, the leader is ahead in that area.
How You Compare to the Cohort
Three views of how your September scores stack up against the four other companies in your cohort: the full cohort grid, your win and loss record, and where you sit in each category’s field.
How Every Company Performs in Each Area
This is a bird’s-eye view of all 5 companies across all 10 AEO categories for September 2026. Darker cells mean stronger scores. Your row sits at the top so your position is the first thing you see. Hover any cell for the exact score.
Your Win and Loss Record Against the Cohort
Across all your scored metrics this month, this is how often your score beats, matches, or trails the cohort average. It is the simplest read on how many individual battles you are winning.
Where You Sit in Each Category’s Field
For every category, your score (blue dot) is placed on the 0–100 field against the full cohort range, the cohort average, and the cohort leader. The closer your dot is to the right, and the further it sits past the average mark, the stronger your position.
Turning Scores Into Priorities
Three views that translate your scores into a plan: which actions give the most return for the least effort, where the gap to the leader is widest, and how much overall score you can still recover.
Impact vs Effort: What to Do First
Each dot is one of this month’s 30 recommended actions, placed by how many SuperScore points it adds when it reaches its target (higher is better) against how much effort it takes (further left is easier). Supporting actions that share a target with another action sit at zero. The actions in the upper-left corner are your quick wins. Color shows the resource area: Coder, Content, or Coordinator.
How Far Behind the Leader You Are, by Category
For each category, this is the number of points between your score and the cohort leader’s, sorted widest gap first. The longer the bar, the more ground there is to make up.
How Much Score You Can Still Recover
Starting from your current SuperScore, this shows the lift from reaching every target in this month’s 30 recommended actions, category by category, assuming competitors hold their September values.
September 2026 Recommended Actions
Every action below passes at least one competitor on one metric, and each card names that competitor, the target, and the SuperScore gain. Six actions take hours to days and add about 9.8 points together, moving Sunrise from 53.8 to about 63.6: a ‘Moving In’ guide (+2.1), rewritten titles and meta descriptions with star-rating schema for a 3.01% click-through rate (+2.6), Organization, LocalBusiness, and FAQPage schema (+2.0), a quarterly community-page review (+1.3), four comparison pages (+1.1), and a named expert for journalists (+0.7). If all 30 actions reach their targets and competitors hold their September values, Sunrise reaches about 72.9 and passes Brookdale (69.2). The actions are grouped by the resource type best suited to deliver each one. Open the SuperScore to read every action with its reason, or download the full list as a CSV.
Explore Your Dashboard
There are three places to go deeper from here. The first two break down the scores that build your AEO performance; the third covers the AI platforms that read them.
SuperScore
This page shows the overall AEO score for your cohort, built from 10 category scores that map how well each brand will appear in AI engines and content. The score covers 10 areas of digital marketing that compose your profile for ChatGPT (OpenAI), Google Gemini (including AI Overviews), Microsoft Copilot (including Bing AI), Perplexity, Claude (Anthropic), and Grok (xAI).
Data analyzed: 5 companies · 10 score areas · September 2026 is the first captured month
AEO SuperScore
Hover over a dot to see the category breakdown for that company. Click any category to drill in. Higher is always better. A score of 0 means the brand is absent on that dimension; 100 means the brand has won what is achievable.
Explore Each Category
Open any of the 10 categories that make up your SuperScore for the full breakdown.Cross-Category Priorities for Sunrise Senior Living
September 2026 Recommended Actions
Below are the 30 highest-leverage recommended actions for the month, grouped by the resource type best suited to deliver each one: coder, content, and coordinator. Each item names the metric it targets and explains the reason, with a link to the underlying category where you can read the supporting analysis. You can also download the full list as a CSV.
AEO Platforms: What's Happening in AI Search?
This screen tracks the latest developments across the AI engines that power AEO. Each platform has its own priorities, signals, and update cadence, and each one represents a channel where your brand may or may not appear.
Platform Intelligence · September 2026
6 platforms tracked
Platform Updates
Below is the latest intelligence we have on each AI engine as of September 2026: what each one is, what drives placement in its answers, where it is strong and weak as a search tool, and what changed this month.
Overview
OpenAI's AI assistant with web search, agentic browsing, and task execution, spanning Free, Plus, Pro, Business, and Enterprise tiers. Hybrid: AI assistant with web search, agentic browsing, and task execution.
- Largest user base of the six at 1.2B weekly users
- GPT-6 Astra and GPT-6.1 Sol improve multi-step research and agentic execution in ChatGPT Work
- Browses and cites without being asked to when search auto-triggers
- New always-on Dots agents carry their own browser for ongoing tasks
- Web search is conditional, not always-on, so citation behavior differs sharply by query type
- Bing indexing is necessary but not sufficient for citation
- Shopping results now favor connected product feeds over the open web, concentrating visibility among large merchants
- Expanding ads (60+ markets, Sponsored Agents test) add commercial bias to results
What Drives Placement
Documented: Bing API supplies URL discovery for ChatGPT's web search, and OpenAI's own ChatGPT-User crawler must separately be able to fetch the page; for shopping queries, OpenAI now ranks results from connected merchant product feeds (Shopify and Etsy by default, other merchants by application) on availability, price, quality, and whether the merchant is the maker or primary seller. Observed: search activates on a minority of definitional queries (~10%) but most discovery-type queries (~73%) per third-party analysis; Bing's top-3 ranking only weakly predicts what ChatGPT cites (~7% alignment); Profound's tracking shows feed-based retrieval rose from ~8% to ~65% of shopping recommendations between July and early September, with the top 10 merchants' share of references nearly doubling. Inferred: brand mentions on Reddit and other high-authority third-party sites likely aid inclusion for non-shopping queries, but OpenAI has not documented this; for product queries, open-web content is now secondary to a connected feed.
What's New with ChatGPT
- Began rolling out GPT-6 Astra to paid tiers with improvements to coding, research, and multi-step workflows, plus document, spreadsheet, and presentation creation; enterprise access disabled by default due to its cybersecurity rating. (Sep 3)
- Released ChatGPT Images 2.5 to all tiers. (Sep 8)
- Launched the Data plugin for Work and Codex; added Box, Dropbox, and SharePoint to Library. (Sep 10)
- Announced planned retirement of custom GPTs with a migration path to plugins. (Sep 11)
- Began testing Sponsored Agents in the US and integrated ChatGPT Ads with Shopify and HubSpot. (Sep 16)
Overview
Google's AI assistant, also embedded in Search as AI Overviews and AI Mode, and across Android and Workspace. Hybrid: search engine layer (AI Overviews/AI Mode) plus standalone AI assistant.
- Sits directly inside the world's dominant general search engine, with AI Mode reported at over 1 billion monthly users and AI Overviews on roughly 40% of US searches
- Largest device/platform reach across Search, Android (default assistant after Google Assistant's Sep 4 shutdown), Windows desktop, and Workspace
- Strong multimodal handling backed by Gemini 3.8 and Gemini 4 Argon
- AI Overviews are associated with a 58% drop in clickthrough to top-ranking results (Ahrefs)
- Some AI Overview links now route users into AI Mode rather than to source sites
- AI Mode users rarely exit to external websites
- No published AEO-specific ranking method exists separate from core Search
What Drives Placement
Documented: inclusion in Google's core Search index, structured/schema data, and standard E-E-A-T ranking signals per Google Search Central documentation; a September 2026 Spam Update and multimodal Search performance reporting in Search Console both operate inside core Search rather than as AI-specific systems. Observed: AI Overviews and AI Mode draw from top-ranking organic and Knowledge Graph sources; Google stated in September it has increased the number of links in AI result content trays; SEO observers recorded on Sep 21 that some links inside AI Overviews now open AI Mode follow-up queries instead of external sites. Inferred: overlap with existing SEO signals is a reasonable inference, not a documented fact; Google has not published a distinct AEO ranking system.
What's New with Gemini
- Launched Gemini 3.8 Flash and Gemini 3.8 Flash Cyber in the Gemini app, AI Mode in Search, Sheets, and the API. (Sep 2)
- Discontinued Google Assistant on Android phones, tablets, Wear OS, and Android Auto, with Gemini as the default assistant. (Sep 4)
- Launched the Gemini app for Windows. (Sep 10)
- Announced Gemini 3.8 Live audio models plus Gemini Notebook study tools. (Sep 16)
- Began experimental rollout of CC, a collaborative household agent in Google Labs. (Sep 18)
Overview
Microsoft's AI assistant embedded in Bing, Edge, Windows, and Microsoft 365, now offering a choice of underlying models including OpenAI's and Anthropic's. Hybrid: AI assistant spanning a search engine (Bing), a browser (Edge), and a productivity suite (Microsoft 365).
- Only major assistant offering a choice of frontier models per surface (OpenAI, Anthropic, and others) with an included default model
- First-party grounding in a user's or organization's own Microsoft 365 content via Work IQ, now joined by Fabric IQ for Power BI and Dynamics data
- Broad distribution across Windows, Edge, Office, and the unified Copilot app with Cowork, Code, and Autopilot agents
- Retrieval and citation behavior are not uniform across surfaces: Bing, Edge, Copilot Chat, and Microsoft 365 Copilot each draw from different sources, and admins can now block up to 1,000 web domains from enterprise answers while news results bypass those blocks
- Smaller consumer search share than Google
- The Sep 25 move to credit-based billing puts the most capable retrieval behind metered usage
What Drives Placement
Documented: for Bing/Edge, Bing index and Bing Webmaster Tools signals apply; for Microsoft 365 Copilot and Copilot Chat, retrieval also draws on an organization's own Microsoft Graph/tenant content (Work IQ), which is not public-web ranking at all; Domain Exclusion returned Sep 16, letting tenants block up to 1,000 external domains from web grounding, and it applies only to web page results, so news-vertical results can still be cited from an excluded domain. Observed: model choice is selectable per surface, now with an included "Auto" model and credit-billed frontier models (GPT-6 Astra, Claude Fable 5.1); Bing is testing an "AI Mode" tab routing to bing.com/copilotsearch; Copilot is rolling out Bing-style rich answer cards for weather, sports, finance, places, and news. Inferred: content structure, freshness, and eligibility for Bing news/vertical results likely help on the Bing/Edge and web-grounded Copilot side by analogy with Bing search, but this is not separately documented for Copilot.
What's New with Copilot
- Made Anthropic's Claude Fable 5.1 available in Copilot Cowork and Copilot Studio, replacing Fable 5. (Sep 2)
- Began rolling out OpenAI's GPT-6 Astra as a selectable model in Copilot Cowork and Copilot Studio. (Sep 4)
- Launched natural-language app building in Copilot Cowork (Frontier) and Copilot Studio (preview). (Sep 10)
- Restored Domain Exclusion for Microsoft 365 Copilot web grounding, up to 1,000 domains, off by default. (Sep 16)
- Shipped Federated Copilot Connectors GA (MCP-based), Copilot Cowork across Android, iOS, Windows, Mac, and Web, Vision for desktop screen and mobile camera in voice sessions, CarPlay support, and a redesigned Copilot Notebooks. (Sep 23)
Overview
An AI assistant built as a citation-first answer engine, paired with the Comet browser and the Perplexity Computer agent. Answer engine.
- Most citation-transparent of the six by design
- Fastest-moving on real-time/current-events retrieval
- Publishes its own search-pipeline metrics
- Now routes sensitive work to on-device models (Hybrid Compute on Mac, Portable Computer on Windows/AMD) without changing its cited-answer format
- Smaller user base than the majors, about 45 million core MAUs, with Statcounter AI referral share falling from 7.73% in April to 4.31% in August 2026, so lower baseline visibility despite strong citation transparency
- September product investment went almost entirely to Computer/agentic features rather than the answer engine
- Amazon's Sep 21 amended complaint over Comet's server-side crawling adds legal risk to its agentic browsing
What Drives Placement
Documented: Perplexity operates its own crawler/index and a "Search as Code" retrieval pipeline, with a public Search SDK for custom search/ranking logic, per its own changelog; in September its developer changelog added a "Fast Search" tier now used by all Agent API presets except xhigh (about 800 ms faster, $1.00 vs $2.50 per 1,000 invocations), new Image Search and Finance Search tools, and a fetch_url full-page retrieval tool as the legacy Sonar API cut over to the Agent API on Sep 27. Observed: every answer carries numbered, clickable citations; Perplexity has published execution-reliability metrics for its search pipeline (81.9% to 92.6%). Inferred: recency and clear factual structure likely favor citation; the shift to a faster, cheaper default search tier and full-page fetch suggests crawlable, fast-loading pages with complete on-page content matter more, though Perplexity has not published a full ranking algorithm.
What's New with Perplexity
- Launched Hybrid Compute on Mac for Pro, Max, and Enterprise, pairing cloud reasoning and web search with a local model for private files and on-device actions. (Sep 1)
- Released Portable Computer for Windows, running model, harness, orchestrator, and scheduler entirely on-device on NVIDIA RTX GPUs. (Sep 14)
- HP began pre-loading the Perplexity Windows app on its PCs, starting with the ZBook Ultra G3a. (Sep 15)
- Added Effort Mode to Computer on web (Light/Standard/High/Ultra). (Sep 17)
- Added GPT-6 Astra as a Computer model, a public Skills Marketplace, and Side Chat. (Sep 21)
Overview
Anthropic's AI assistant and model platform, used for reasoning, writing, and agentic coding/work tasks, with web search and tool-use as one capability rather than its core design. AI assistant / model platform, not primarily a search engine.
- Strong reasoning and writing quality with a 1M-token default context across the new Fable 5.1, Opus 5.5, and Sonnet 5.5 models
- Distributed as an embedded model through Amazon Bedrock, Google Cloud, and Microsoft Foundry as well as its own apps
- Always-on citations and source links when web search runs
- Less search-centric than Google, Bing, or Perplexity by design
- Web search runs only when enabled and Anthropic publishes no source selection or ranking method
- September 2026 releases focused on models, agents, and enterprise tooling rather than search or discovery
What Drives Placement
Documented: Claude's web search retrieves and cites sources only when enabled; Anthropic's help center states Claude processes multiple sources to find relevant content and that every search-backed response includes citations, with searches triggered automatically for current-information questions or on explicit user request. Inferred: as of September 2026, Anthropic still has not published placement or ranking criteria for what Claude cites or mentions; September release notes (Fable 5.1, Opus 5.5, Sonnet 5.5, Cowork in every conversation) contain no changes to web search or citation behavior.
What's New with Claude
- Launched Claude Fable 5.1 and Claude Mythos 5.1 with a 1M-token context window, 128k max output, always-on adaptive thinking, cheaper cache reads, and text watermarking plus C2PA Content Credentials on generated files. (Sep 1)
- Published "Detecting and countering misuse of AI: September 2026"; launched Smart reports (beta) for Claude Enterprise. (Sep 10)
- Added on-demand conversation compaction to the Messages API in beta. (Sep 14)
- Brought Claude Cowork to every conversation and enabled creating designs, decks, and docs in any chat across all plans. (Sep 16)
- Introduced the Life Sciences Verification Program; Sep 18 announced an Accenture partnership on embedded evaluation. (Sep 17)
Overview
xAI's AI assistant embedded in X, with growing standalone apps and third-party distribution. Hybrid: AI assistant embedded in a social platform (X) with web search.
- Only one of the six with native, real-time access to a major social platform's live conversation data
- Separate web and X search tools with an explicit citations field and domain allow/exclude controls
- Rapid model release cadence (Grok 4.7 on Sep 21 with a 500k context window)
- Fast-expanding third-party distribution (Cursor, Grok Build, Microsoft Foundry, Gemini Enterprise Agent Platform)
- X-centric data weights the source mix toward social conversation rather than authoritative publications, a property of Grok's source mix, not a documented ranking weakness
- Domain filters cap at 5 domains, limiting fine-grained source control
- September releases added no transparency on how web vs. X results are weighted or how sources are chosen
What Drives Placement
Documented: Grok has direct, real-time access to X/Twitter posts as a distinct data source from its general web search mode, per xAI's own product documentation; the API exposes separate Web Search and X Search tools, returns a citations field listing the sources behind each answer, and lets developers restrict or exclude up to 5 domains per search via allowed_domains and excluded_domains (docs last updated May 27, 2026, no September change). Inferred: what drives citation or mention likely differs between the X-data path and the general web-search path, though xAI has not documented this; September's releases (Grok 4.7, Team Bots, Grok Bot for Enterprise, Voice Transcribe 2.0) were model, agent, and speech updates with no stated search or citation changes.
What's New with Grok
- Published "Biosecurity at the frontier," reporting third-party evaluation showing Grok outperformed competitors at detecting dangerous biological queries. (Sep 1)
- Launched Grok Bot for Enterprise, autonomous agents with governance controls running in isolated instances. (Sep 3)
- Added persistent Memory to Grok Build so agents retain context across sessions. (Sep 16)
- Released Grok Voice Transcribe 2.0, ranked first among 32 streaming speech-to-text models on the Artificial Analysis leaderboard. (Sep 18)
- Announced Grok 4.7 with a 500k context window, $2/$6 per million tokens, a 2x-speed Fast variant, and immediate availability in Cursor, Grok Build, and the API. (Sep 21)