Berkey AI Market Strategy Report - Water Filter Systems

Answer Capsule

Berkey is visible in AI-generated water filter recommendations but is not being chosen. In September 2026, the brand appeared in 19.4% of qualified observations yet received a valid recommendation in only 8.4%, the widest presence-to-recommendation gap among the ten tracked brands. Berkey's clearest weakness is recommendation conversion: presence held flat from July 2026 while valid recommendations fell from 99 to 60. The clearest opportunity is the brand's still-positive framing and a small but real rank-one pocket that a focused answer-layer and citation buildout can defend and extend.

Who This Report Is For

This report is for Berkey's marketing, ecommerce, and brand leadership teams, and for category planners in home water treatment who need to understand how AI systems are framing Berkey at the moment buyers ask which water filter system to choose.

Report Card

Executive Summary

Berkey is present but under-recommended in AI-generated water filter recommendations. The September 2026 LLM Authority Index benchmark recorded the brand in 19.4% of qualified observations, unchanged from July 2026, while valid recommendation coverage fell to 8.4% from 14.2%, a decline of 5.8 points that the benchmark classified as significant. That is the defining pattern for the brand this month: the same number of appearances, fewer of them carrying a recommendation.

The gap between presence and recommendation is the widest in the tracked set. Berkey appears in roughly one in five qualified observations but is recommended in fewer than one in twelve. By comparison, Aquasana converts 66.7% presence into 53.4% recommendation coverage, and iSpring converts 56.1% presence into 48.8%. Berkey's conversion rate is the weakest of the ten brands, and it is the clearest diagnostic signal in the dataset.

The decline is a recommendation quality problem, not a visibility problem. Berkey's presence rate held flat at 19.4%, while its valid recommendation count fell from 99 observations in July 2026 to 60 in September 2026. The brand is still being mentioned in the same prompts; it is simply being mentioned in a different way, less often as an answer and more often as context, comparison, or cautionary reference.

Placement metrics confirm the pattern. Berkey's top-three rate fell to 3.5% and its rank-one rate sits at 0.3%, or two observations in the current month. The brand's average recommended rank of 3.43 is the weakest among brands with rank-eligible recommendations, meaning that even when Berkey is recommended, it is recommended late in the list.

The strongest platform signal for Berkey is Copilot, where the brand recorded an 11.9% valid recommendation coverage and a 0.0119 rank-one rate, both above its overall averages. The weakest platform signal is Gemini, where Berkey recorded a 2.15% valid recommendation coverage and no rank-one placements. AI Mode carried the largest share of Berkey's neutral mentions, with 54 of the brand's 71 neutral mentions occurring on that surface, which suggests the brand is frequently referenced as context rather than as a recommendation.

Sentiment remains positive overall at 0.4173, but it is the second-lowest net sentiment score in the tracked set, behind only Pentair at 0.3729. Berkey's framing is not negative; it is thin. The brand is being described in neutral, factual terms more often than it is being endorsed.

The clearest opportunity is to convert Berkey's existing presence into recommendation credit. The brand already appears in the prompts that matter. The work is to make the brand's public evidence layer easier for AI systems to retrieve, cite, and synthesize as a recommendation rather than as a reference.

What Berkey Is Winning

Questions This Section Answers

• Where is Berkey still converting AI mentions into actual recommendations?
• Which platform is Berkey's strongest recommendation signal?
• What does Berkey's positive sentiment and rank-one pocket tell us about its current position?

Berkey's wins this month are narrow but real. The brand retains a positive net sentiment score of 0.4173, which means that among the mentions that were classified, positive framing still outweighs negative framing. That is a foundation to build on, not a problem to fix.

Berkey also retains a small rank-one pocket. Two observations in September 2026 placed the brand first, and the brand's average recommended rank of 3.43, while the weakest in the set, still reflects a real placement when the brand is recommended. The brand is not absent from the recommendation layer; it is thinly represented in it.

Copilot is Berkey's strongest platform signal. The brand recorded an 11.9% valid recommendation coverage on Copilot, above its 8.4% overall coverage, and a 0.0119 rank-one rate, above its 0.0028 overall rate. Copilot also produced the brand's highest positive visibility rate at 11.9%. This is the platform where Berkey's evidence layer is converting most effectively.

Berkey's presence rate of 19.4% is also a win in the sense that it has not deteriorated. The brand is still being surfaced in the same share of qualified observations as it was in July 2026. The decline is concentrated in how those appearances are framed, not in whether they occur.

Where Berkey Has the Clearest AI Visibility Gaps

Questions This Section Answers

• Where is Berkey losing the most recommendation share relative to its AI presence?
• Why is Berkey appearing in AI answers without being recommended?
• Which brands are taking the recommendation positions Berkey is not converting?

Berkey's clearest gap is recommendation conversion. The brand appears in 19.4% of qualified observations but receives a valid recommendation in only 8.4%. That 11.0-point gap is the widest in the tracked set and the single most important number in this report. Competitors with similar or lower presence rates convert far more effectively: APEC Water Systems appears in 38.5% of observations and converts to 33.6% coverage, and Clearly Filtered appears in 49.2% and converts to 38.6%.

The gap is also widening. In July 2026, Berkey converted 19.4% presence into 14.2% coverage, a gap of 5.2 points. By September 2026, the same presence converted into 8.4% coverage, a gap of 11.0 points. The brand's appearances are becoming less recommendation-shaped over time.

Placement is the second gap. Berkey's top-three rate of 3.5% and rank-one rate of 0.3% place the brand at the bottom of the tracked set on both measures. When Berkey is recommended, it is recommended late. The brand's average recommended rank of 3.43 is the weakest among brands with rank-eligible recommendations, behind Pentair at 3.30 and Brita at 3.24.

Platform coverage is the third gap. Berkey's strongest platform, Copilot, produced only 10 valid recommendations in September 2026. Gemini produced 2. AI Mode produced 16, but 54 of the brand's 71 neutral mentions occurred there, which suggests the surface is treating Berkey as context rather than as a recommendation. The brand has no platform where it holds a strong recommendation position.

The competitive displacement pattern is clear. iSpring, Aquasana, and APEC Water Systems hold the recommendation positions that Berkey is not converting into. iSpring recorded a 30.96% top-three rate and a 10.46% rank-one rate; Aquasana recorded 36.68% and 16.46%; APEC recorded 24.83% and 13.25%. These are the brands being named when buyers ask which water filter system to choose, and Berkey is being mentioned alongside them rather than instead of them.

Biggest Opportunity

Questions This Section Answers

• What should Berkey do to turn AI presence into actual recommendations?
• Which high-intent prompts offer the clearest conversion opportunity for Berkey?

Berkey's biggest opportunity is to convert its existing presence into recommendation credit by strengthening the public evidence layer that AI systems retrieve when they form a shortlist. The brand already appears in the prompts that matter. The work is to make the brand's owned and earned content easier for AI systems to cite as a recommendation rather than as a reference.

The specific path is to build answer-shaped content around the high-intent prompts where Berkey already appears but is not being recommended. The benchmark's prompt examples include "best water filtration system for home," "What is the best water filter for drinking," and "Which type of water filter is best." These are the prompts where Berkey's presence is not converting. A focused answer-layer buildout that pairs each prompt with a clear, citable, brand-owned answer is the most direct route from reference to recommendation.

Competitive Landscape

Questions This Section Answers

• Who leads the water filter category in AI recommendation rankings?
• How far behind the leaders is Berkey on top-three and rank-one recommendation rates?

Aquasana and iSpring hold the strongest recommendation-stage positions in the Water Filter Systems category, with Aquasana leading on top-three and rank-one rates and iSpring leading on overall captured recommendation value. Berkey sits ninth of ten on valid recommendation coverage and last on rank-one rate, with a presence-to-recommendation gap that is the widest in the tracked set.

Average recommended rank covers rank-eligible recommendations only.

Berkey's position in the table reflects a brand that is present in the category conversation but not in the recommendation layer. Its top-three rate is less than one-seventh of Aquasana's, and its rank-one rate is the lowest of any brand with rank-eligible recommendations. The sentiment column shows that Berkey's framing is not the problem; the brand's mentions are more positive than Brita's or PUR's, both of which convert presence into recommendations more effectively.

Prompt Evidence

AI Mode / Brand Recommendation Prompt: "What is the best whole house water filtration system?" Result: Berkey was mentioned as a comparison reference but was not placed in the top three recommendations, contributing to the brand's 54 neutral mentions on AI Mode.

Copilot / Brand Recommendation Prompt: "What is the best water filter for drinking?" Result: Berkey received a valid recommendation and appeared in the top three, one of the brand's 10 valid recommendations on Copilot and its strongest platform signal.

Gemini / Brand Recommendation Prompt: "Which type of water filter is best?" Result: Berkey was mentioned but not recommended, consistent with the brand's 2.15% valid recommendation coverage on Gemini and its zero rank-one placements on that surface.

ChatGPT / Brand Recommendation Prompt: "best water filtration system for home" Result: Berkey was mentioned but received no valid recommendation, one of the 3 observations where the brand appeared on ChatGPT without converting to a recommendation.

What CiteWorks Studio Would Do Next

Phase 1: AI Market Discovery Audit Map every prompt where Berkey appears but is not recommended, and identify the specific surfaces, clusters, and competitor displacements driving the 11.0-point presence-to-recommendation gap.

Phase 2: Recommendation Readiness Plan Prioritize the prompts where Berkey's presence is highest and its conversion is lowest, and define the answer shape, evidence type, and citation target required to convert each one.

Phase 3: Owned Answer Layer Buildout Build brand-owned, citable answers for the high-intent prompts where Berkey is currently referenced as context, starting with the "best water filter for drinking" and "best whole house water filtration system" clusters.

Phase 4: Citation / Authority Layer Development Strengthen the public evidence layer that AI systems retrieve when forming water filter shortlists, including comparison pages, specification content, and third-party sources that AI systems already cite in this category.

Phase 5: Monthly AI Visibility and Recommendation Tracking Track Berkey's presence, valid recommendation coverage, top-three rate, rank-one rate, and sentiment month over month, with the presence-to-recommendation gap as the primary KPI.

Why This Matters

AI presence alone is not enough. Berkey is mentioned in roughly one in five qualified observations, but it is recommended in fewer than one in twelve. Buyers who ask an AI system which water filter system to choose are not seeing Berkey as an answer; they are seeing Berkey as a reference point in someone else's answer. That is a materially different position, and it is the position that determines whether the brand enters the buyer shortlist.

The next move is targeted correction of the prompt, page, and citation layers. The benchmark shows where Berkey is present and where it is not converting. The work is to close that gap prompt by prompt, page by page, and source by source, so that the brand's existing visibility converts into recommendation credit at the moment buyers are forming their shortlist.

Core Metrics

Sentiment Score

Questions This Section Answers

• Why can a high number of AI mentions still mask a recommendation problem for a brand like Berkey?
• How is Berkey's sentiment score calculated in this benchmark?

Sentiment Score = (positive mentions x 1 + neutral mentions x 0 + negative mentions x -1) / total mentions

For Berkey in September 2026, that is (63 x 1 + 71 x 0 + 5 x -1) / 139 = 0.4173.

This matters because unclassified mention counts are misleading. A brand can appear frequently in AI responses and still be losing the recommendation stage if most of those appearances are neutral references, comparison anchors, or cautionary mentions. Berkey's 139 mentions look substantial, but 71 of them are neutral and only 63 are positive. The brand's share of voice in the category is not the same as its share of recommendations.

Share of voice is a diagnostic metric, not a business KPI. A positive recommendation, a neutral reference, a cautionary mention, and a competitor-displaced mention are not equal. Counting all mentions as wins is bad measurement. Berkey's 139 mentions include 71 that carry no recommendation weight at all, and the brand's 5 negative mentions are the only ones that actively work against it. Classified sentiment is required before interpreting AI visibility, because the difference between "Berkey was mentioned" and "Berkey was recommended" is the difference between presence and pipeline.

Sentiment by Platform

Questions This Section Answers

• On which AI platform is Berkey's sentiment strongest, and where is it being treated as context rather than a recommendation?
• Which platforms show Berkey's mentions are positive enough to convert, but still not recommendation-led?

Methodology

1. This report is a benchmark-based analysis of Berkey's position in the Water Filter Systems category, produced from the September 2026 LLM Authority Index AI Market Discovery benchmark and the associated metrics aggregation dataset.
2. The reporting window is September 2026, with July 2026 as the baseline comparison month and August 2026 as an intermediate measurement point.
3. Six AI and search surfaces were tracked: ChatGPT, Copilot, Gemini, Perplexity, AI Overviews, and AI Mode. All six produced at least one qualified observation in the reporting window.
4. The benchmark collected 800 prompt-surface observations in September 2026, of which 594 were unique questions. After relevance filtering and qualification, 717 observations formed the public denominator.
5. Ten brands were tracked: Aquasana, APEC Water Systems, Berkey, Brita, Clearly Filtered, Culligan, iSpring, Pentair, PUR, and SpringWell Water.
6. All 717 qualified observations in September 2026 fell into the Brand Recommendation cluster. The benchmark captured no qualified observations in the Pricing & Value or Multi-Brand Comparison clusters in any month of the series.
7. Stage 0 extraction retained the query, AI or search surface, answer, brand outcome, recommendation placement, sentiment, and, where exposed, citations or attributable evidence sources.
8. A mention is any qualified observation in which Berkey appears in the AI response, regardless of whether the brand is recommended.
9. A valid recommendation is a qualified observation in which Berkey receives a recommendation that fits the query. Neutral, cautionary, comparison-anchor, and listed-only mentions are not counted as valid recommendations.
10. Top-three rate is the share of qualified observations where Berkey appears in the first three recommended positions. Rank-one rate is the share where Berkey is the first recommendation. Average recommended rank covers rank-eligible recommendations only.
11. Brand-level percentages use the qualified observation set of 717 as the denominator, not the 800 raw prompt-surface runs. The qualified set excludes prompts that did not mention a tracked brand or were deemed irrelevant.
12. The benchmark treats month-over-month movement as a signal for investigation, not as proof of cause. Berkey's 5.8-point decline in valid recommendation coverage was classified as significant against the benchmark's movement thresholds. Explaining why that movement occurred requires a deeper, company-level analysis of prompts, surfaces, and evidence sources.

See Where AI Is Recommending Your Brand

The public benchmark shows where Berkey is present and where it is not converting. A company-level AI visibility audit maps the specific prompts, surfaces, competitors, and evidence sources behind the brand's 11.0-point presence-to-recommendation gap, and turns the benchmark's directional signals into a prioritized plan for closing it.