How AI Search Is Recommending Eating Disorder Treatment Centers: Monthly Trends

Executive Summary

The Emily Program leads the category in October 2026 at 18.5% valid recommendation coverage, ahead of Monte Nido at 13.9% and Alsana at 12.3%. The gap from first to third is 6.2 points, and the gap from first to second is 4.6 points. The Emily Program held the lead at the start of the tracked series in July 2026 (7.7%) and holds it again in October, but the field moved through a different configuration in between: ERC Pathlight took the lead in September 2026 at 23.5%, a 7.3-point advantage over The Emily Program's second-place reading of 16.2% that month, before falling back sharply in October.

The strongest riser against the baseline is Monte Nido, up 11.7 points from 2.2% in July 2026 to 13.9% in October 2026, a move the benchmark classifies as significant. Monte Nido also recorded 4.6% top-three placement and 0.0% rank-one placement in October 2026, so its gains came from second- and third-position recommendations rather than first place. Alsana is up 7.9 points over the same span on a three-month upward streak, and it also recorded a rank-one placement this month (1.5% rank-one rate) even though its cumulative gain falls just under the benchmark's significance threshold.

Against the prior month, the sharpest single move belongs to ERC Pathlight, down 17.3 points from 23.5% in September 2026 to 6.2% in October 2026, a decline the benchmark classifies as significant. No brand is classified as a significant decliner against the July baseline, but ERC Pathlight's month-over-month drop is the largest single-brand move in the current reporting period and reverses its September leadership position, leaving it tied for sixth place in October. The category leader for October 2026 is The Emily Program at 18.5%, up 10.8 points from 7.7% in July 2026, a cumulative gain that sits just inside the benchmark's normal variation band.

Each monthly run begins with 320 prompt-surface observations (246 unique questions) in October 2026, down from 213 prompt-surface observations (209 unique questions) in July 2026. Of those, 320 mentioned a tracked brand or competitor in October and 213 in July; 220 were relevant and 100 were irrelevant in October, versus 138 relevant and 75 irrelevant in July. The public metrics use the 65 qualified observations in October 2026 and 91 in July 2026 that survive both qualification stages. August 2026 contributed 71 qualified observations and September 2026 contributed 68.

Key Findings

AI Response Inconsistency Alerts

Questions This Section Answers

• Where did AI platforms describe the same treatment center in incompatible ways?
• Which organizations and platforms were involved in the critical or high severity conflicts?

The benchmark's conflict review detected 5 critical or high severity factual inconsistencies across 4 AI platforms: ChatGPT, Copilot, Gemini, and Perplexity. Two companies were involved.

Center for Discovery

A critical severity factual inconsistency concerns what the New York Center for Discovery is. When asked "What is the Center for Discovery in New York?", ChatGPT stated that the Center for Discovery in New York is a specialized eating disorder and mental health treatment provider offering residential, PHP, IOP, and outpatient care, while Gemini stated that the Center for Discovery is a nonprofit providing education, healthcare, residential, and research services for children and adults with complex disabilities, medical frailties, and autism. Gemini's answer cited the Sullivan County Partnership investor directory, a PR Newswire release on the organization's economic impact in New York State, and a family network resource guide. The conflict review flagged the PR Newswire release and the organization's own site as sources supporting the disability services identity. Both answers claim to describe the primary Center for Discovery in New York but assign it incompatible core identities.

A high severity factual inconsistency concerns the location of the New York Center for Discovery. When asked "What is the Center for Discovery in New York?", ChatGPT described Center for Discovery as a treatment provider with programs in New York without giving a specific campus location, while Perplexity stated that the Center for Discovery is located at 641 Old Route 17, Monticello, NY 12701, with a mailing address at P.O. Box 840, Harris, NY 12742. Perplexity cited ZoomInfo, the organization's About Us page, and LinkedIn. The review flagged the organization's contact page as supporting the Monticello address and two pages describing residential and complex disability services as supporting the eating disorder treatment framing, reflecting that the two answers are describing different organizations under the same name.

Monte Nido

A high severity factual inconsistency concerns current ownership. When asked "Who owns Monte Nido?", ChatGPT stated that Monte Nido is owned by Revelstoke Capital Partners alone, while Copilot stated that Monte Nido is owned by Revelstoke Capital Partners and Constitution Capital Partners. ChatGPT cited a Revelstoke acquisition announcement and a Centre Partners investment notice, while Copilot cited the Revelstoke announcement, a news aggregator, and a mergers and acquisitions database. The review flagged the Revelstoke announcement as supporting the sole owner claim at lower confidence, and flagged two database entries stating that Monte Nido is owned by BPEA Private Equity, Constitution Capital Partners, and Revelstoke Capital Partners as supporting the co-ownership claim.

A high severity factual inconsistency concerns the population of Monte Nido, California. When asked "What is the population of Monte Nido CA?", ChatGPT stated roughly 9,500 residents as a broader neighborhood estimate, while Perplexity stated roughly 26,200 to 26,300 residents. ChatGPT cited a census reference site, a real estate demographics page, and Wikipedia. Perplexity cited Apple Maps, a real estate neighborhood page, and a places directory. The review flagged the real estate neighborhood page as stating that 26,205 people live in Monte Nido.

A high severity factual inconsistency concerns the same population figure. When asked "What is the population of Monte Nido CA?", Gemini stated approximately 9,515 residents, while Perplexity stated roughly 26,200 to 26,300 residents. Gemini cited a real estate demographics page, and Perplexity cited the same three sources as above. The review flagged the same real estate neighborhood page as the source of the higher figure.

Benchmark Context

Questions This Section Answers

• How many prompts qualified for the public benchmark versus the full collection, and why does the gap matter?
• Which AI surfaces contributed qualified observations in October 2026 compared with earlier months?

The report separates the raw collection universe from the qualified analysis set. Brand-level recommendation percentages are calculated within the qualified benchmark set.

July 2026 captured observations across two AI surface families, AI Mode and AI Overviews. August 2026 added Gemini, and both September 2026 and October 2026 carried three qualified surfaces. October 2026 replaced Gemini with Copilot as the third qualified surface alongside AI Mode and AI Overviews. The qualified observation count fell from 91 in July 2026 to 65 in October 2026 even as the raw prompt collection grew, so a larger share of collected prompts did not meet the qualification criteria. September 2026 sat at 68 qualified observations, the closest reading to the current month.

Benchmark-Level Metrics

The valid recommendation shortlist share reached 29.2% in October 2026, its highest reading in the tracked series and up from 19.1% in September 2026, meaning AI systems produced actionable recommendations in a larger share of qualified observations even as the qualified set itself shrank.

AI Recommendation Trend

Questions This Section Answers

• Which eating disorder treatment centers lead the category in October 2026, and by how much?
• Which brands moved more than normal month-to-month variation?
• Which tracked brand recorded no recommendations and no presence across the entire series?

Three Brands Now Separate the Leader from the Rest of the Field

Three brands exceeded normal month-to-month variation from July 2026 to October 2026: Center for Discovery, Monte Nido, and Rogers Behavioral Health. The Emily Program leads the current reading at 18.5%, with Monte Nido at 13.9% and Alsana at 12.3% trailing by as little as 4.6 points and as much as 6.2 points. The remaining movement came from the combination of several smaller baseline-to-current shifts, with seven of nine brands recording higher coverage in October 2026 than in July 2026 and one brand, Veritas Collaborative, flat at zero.

What Changed This Month

The Emily Program: Holding the Lead as the Field Compresses

The Emily Program's valid recommendation coverage rose from 7.7% in July 2026 to 18.5% in October 2026, a 10.8-point gain that sits just inside the benchmark's normal variation band. Against the prior month, coverage moved up 2.3 points from 16.2% in September 2026, a smaller step than the September gain of 9.2 points, which was not enough to overtake ERC Pathlight's 23.5% reading that month.

Its placement profile strengthened materially. Rank-one rate moved from 2.2% in July 2026 to 7.7% in October 2026, and top-three rate moved from 5.5% to 10.8% over the same span. Its valid recommendation count reached 12 in the 65 qualified October observations, with 29 total brand mentions and 13 positive ones.

The distinction to notice is that The Emily Program holds the lead on coverage while Monte Nido sits 4.6 points behind with substantially higher presence at 58.5%. The Emily Program leads on recommendations while Monte Nido leads on visibility, and the two measures are no longer tracking together for either brand.

Highest-priority diagnostic: Which prompt types and surfaces carry The Emily Program's rank-one placements, and where does Monte Nido's higher presence fail to convert into recommendations?

Monte Nido: Significant Cumulative Gain with Presence Still Outrunning Recommendations

Monte Nido's valid recommendation coverage rose from 2.2% in July 2026 to 13.9% in October 2026, an 11.7-point gain the benchmark classifies as significant. Against the prior month, coverage moved up 3.6 points from 10.3% in September 2026, extending an upward streak to two months.

Its placement profile improved unevenly. Top-three rate moved from 1.1% in July 2026 to 4.6% in October 2026, while rank-one rate remained 0.0% in both months. Its valid recommendation count reached 9, with an average recommended rank of 4.75, meaning most of its recommendations land below the top three. Presence moved from 72.5% of July observations down to 58.5% in October 2026, a 14.0-point decline.

Monte Nido is the clearest case of a brand whose visibility and recommendation credit remain out of step. It was surfaced in more than half of qualified October observations but converted that presence into a valid recommendation in 9 of 65. Average recommended rank of 4.75 shows the gap is positional as well as numeric.

Highest-priority diagnostic: Which prompts surface Monte Nido without recommending it, and what separates the contexts where it lands a recommendation from those where it only appears?

ERC Pathlight: Largest Single-Month Retreat in the Series

ERC Pathlight's valid recommendation coverage fell from 23.5% in September 2026 to 6.2% in October 2026, a 17.3-point drop the benchmark classifies as significant against the prior month. Against the July 2026 baseline, coverage moved down 0.4 points from 6.6%, a flat cumulative reading that masks the swing between September and October.

Its presence collapsed alongside coverage. Raw mention presence rate fell from 79.4% in September 2026 to 24.6% in October 2026, which sits 41.3 points below the July baseline of 65.9%. Top-three rate moved from 4.4% in July 2026 to 6.2% in October 2026, while rank-one rate moved from 4.4% to 3.1% over the same span.

This is the sharpest repositioning in the series. ERC Pathlight held the category lead in September 2026 at 23.5% and now sits tied for sixth at 6.2%. Its valid recommendation count fell from 16 in September to 4 in October, a 12-recommendation drop, while its rank-one count fell from 4 in July to 2 in October, a comparatively smaller decline suggesting that much of the September strength sat in second- and third-position placements rather than first place.

Highest-priority diagnostic: Which prompts and surfaces carried ERC Pathlight's September recommendations that no longer appear in October, and whether the change reflects the shift from Gemini to Copilot as the third qualified surface.

Alsana: Three-Month Upward Streak Inside Normal Variation

Alsana's valid recommendation coverage rose from 4.4% in July 2026 to 12.3% in October 2026, a 7.9-point gain the benchmark does not classify as significant, though it marks a third consecutive monthly rise. Against the prior month, coverage moved up 2.0 points from 10.3% in September 2026.

Its top-three rate is the standout supporting movement. Top-three placement rose from 1.1% in July 2026 to 12.3% in October 2026, an 11.2-point gain, and its rank-one rate moved from 0.0% to 1.5%. Its valid recommendation count reached 8 of 65 qualified observations, with 8 positive mentions out of 12 total mentions.

Alsana stands out for a top-three placement gain that is larger in percentage-point terms than its overall coverage gain, even though its cumulative coverage movement stays within the benchmark's normal variation band. Its recommendations are clustering in the upper positions even as its total coverage stays within normal variation.

Highest-priority diagnostic: Which prompts produce Alsana's top-three placements, and whether that positional strength can be extended into broader coverage.

Center for Discovery, Rogers Behavioral Health, and The Renfrew Center: A Second Tier Moving Up Together

Center for Discovery's valid recommendation coverage rose from 0.0% in July 2026 to 6.2% in October 2026, a 6.2-point gain the benchmark classifies as significant and a second consecutive month with recommendations after two zero months. Its valid recommendation count reached 4 of 65, with 12 total mentions and 4 positive ones. Presence moved from 20.9% in July to 18.5% in October 2026, a 2.4-point decline, so the brand began receiving recommendations while being surfaced slightly less often.

Rogers Behavioral Health's coverage rose from 2.2% in July 2026 to 10.8% in October 2026, an 8.6-point gain the benchmark classifies as significant and a third consecutive monthly rise. Its rank-one rate moved from 0.0% to 1.5%, and its net sentiment score reached 0.9, the highest among tracked brands. Its valid recommendation count reached 7 of 65.

The Renfrew Center's coverage rose from 1.1% in July 2026 to 7.7% in October 2026, a 6.6-point gain just outside the significance threshold and a third consecutive monthly rise. Its presence fell from 35.2% to 16.9% over the same span, an 18.3-point decline. Its recommendations rose even as its visibility fell, the same pattern Center for Discovery shows.

Highest-priority diagnostic: Which prompts convert presence into recommendation for these brands, and whether the presence declines reflect prompt-mix changes or surface changes.

Walden Behavioral Care and Veritas Collaborative: The Wide Gap Persists

Walden Behavioral Care's valid recommendation coverage rose from 0.0% in July 2026 to 4.6% in October 2026, a 4.6-point gain just below the significance threshold and a second consecutive month with recommendations. Its valid recommendation count reached 3 of 65, with an average recommended rank of 7.33 and no top-three placements. Presence fell from 11.0% in July to 6.2% in October 2026.

Veritas Collaborative recorded no presence and no valid recommendations in any tracked month from July 2026 through October 2026. Its coverage is 0.0% across the full series.

The notable gaps in the series are category-level observations about repositioning rather than any single brand's problem. The gap between Monte Nido and Veritas Collaborative widened by 11.7 points from July 2026 to October 2026. The gap between ERC Pathlight and Monte Nido moved from a 4.4-point ERC Pathlight advantage in July 2026 to a 7.7-point Monte Nido advantage in October 2026, a 12.1-point swing across the series.

Highest-priority diagnostic: Whether Veritas Collaborative appears in the benchmark's prompt universe at all, and which remaining presence contexts surface Walden Behavioral Care without recommending it.

Buyer-Intent Interpretation

Questions This Section Answers

• Which buyer-intent clusters did the qualified observations actually cover, and which remained empty?
• What can the benchmark not yet answer about how AI systems frame cost, insurance, or head-to-head treatment center comparisons?

In October 2026 all 65 qualified observations fell into the Brand Recommendation cluster, the same concentration seen across the 91 qualified observations in July 2026, 71 in August 2026, and 68 in September 2026. No qualified observations fell into the Pricing & Value or Multi-Brand Comparison clusters in any tracked month. October 2026 did record one pricing analysis response and one comparison analysis response in the broader collection, but neither produced a qualified observation in the public benchmark set. For an eating disorder treatment category where financial accessibility and program fit are often decisive, the public benchmark cannot yet answer how AI systems frame cost, insurance, or direct treatment center comparisons.

Brand Opportunity Summary

Questions This Section Answers

• Which treatment centers show the strongest current signals, and what diagnostic should each one investigate next?
• What does the table identify for brands losing ground or failing to convert visibility into recommendations?

The benchmark identifies where attention is warranted; a company-level analysis is needed to explain why.

Evidence Behind the Benchmark

Questions This Section Answers

• What does each prompt-level observation capture, and what should readers avoid assuming about causation from these metrics?

The aggregate metrics are built from prompt-level observations capturing the query, surface, recommendation outcome, rank, sentiment, and citations where exposed. Company-level analysis can go deeper into prompt, competitor, surface, and evidence patterns. Source presence is not automatically treated as proof of causation.

About This Benchmark

This report is part of the CiteWorks Studio AI Visibility Industry Market research program.

• AI Visibility Industry Market Methodology
• AI Visibility Industry Market Metrics
• AI Visibility Industry Market Standards

Report-Specific Interpretation Notes

• Small observation counts limit what can be concluded: the October 2026 qualified set is 65 observations, down from 91 in July 2026, 71 in August 2026, and 68 in September 2026. Brand-level percentages rest on small absolute counts (for example, 12 valid recommendations for The Emily Program and 3 for Walden Behavioral Care), so single-prompt shifts can move percentages by several points.
• Movement between months identifies changes worth investigating; it does not establish cause. ERC Pathlight's 17.3-point retreat flags a question, not an explanation.
• The qualified denominator differs from the raw collection: 65 of 320 October prompts, 91 of 213 July prompts, 71 of 270 August prompts, and 68 of 327 September prompts survived qualification stages, so public percentages reflect the qualified set only.
• The data describes the output distribution, not its cause. Attribution requires company-level audit.
• The benchmark's third qualified surface changed from Gemini in August 2026 and September 2026 to Copilot in October 2026. Surface-mix changes can shift which brands receive recommendations without any change in brand behavior.

Next Step

The Public Benchmark Shows Where a Brand Is Winning or Losing. A Company-Level Audit Shows Why.

Beneath the aggregate percentages sit the questions that determine real-world impact: which high-intent prompts does a brand win or lose, which competitor takes the recommendation when a brand loses, what attributes AI systems associate with each treatment center, and which external sources shape those answers. The conflict review found 5 critical or high severity factual inconsistencies about Center for Discovery and Monte Nido across ChatGPT, Copilot, Gemini, and Perplexity, showing that AI systems are not always describing a single organization consistently.

A company-specific AI visibility audit maps those prompt, surface, competitor, ranking, sentiment, and evidence-source patterns into a prioritized visibility strategy. It can help explain why ERC Pathlight's recommendations fell 17.3 points from September, why Monte Nido's 58.5% presence converts into only 9 recommendations at an average rank of 4.75, and why The Emily Program holds the lead on coverage while ranking second on visibility, then turn those answers into concrete action.

Request an AI visibility audit