The Bowstone Index (BX) is a weekly measure of sports brand visibility across two signals that increasingly shape how audiences discover and encounter teams, athletes, and leagues:
Every tracked entity receives two BX scores:
Bowstone also tracks Market Signal, including franchise valuation, as a separate contextual measure rather than a component of BX — see Correlation View for how it's used to spot divergence between AI visibility and market value.
The Index is designed to answer a question traditional sports analytics does not:
Which sports brands are winning the AI conversation — and how does that visibility compare with social momentum and real-world market value?
Scores are recalculated after each weekly collection run and updated every Monday.
For most of the internet era, digital visibility was largely link-based. A fan searched Google. Google ranked webpages. Brands measured rankings, clicks, impressions, traffic, and social engagement.
That model is changing.
Traditional search remains enormously important, but discovery is increasingly AI-mediated. Consumers now encounter generated answers through standalone AI platforms such as ChatGPT, Claude, Gemini, and Perplexity, while generative answers are also becoming part of traditional search itself through products such as Google AI Overviews and AI Mode.
That creates a different kind of visibility. In a traditional search result, a brand competes for a position among links. In an AI-generated response, the system decides which teams, athletes, leagues, and sources become part of the answer at all. A user may never click a website. They may simply ask:
The entities included in the response receive exposure. Those omitted do not.
Bowstone does not assume AI will replace Google, social media, or traditional sports coverage. It measures a growing discovery layer that those existing metrics do not capture.
Search used to be primarily about whether you ranked. AI visibility adds another question: are you part of the answer?
AI Visibility measures how often and how consistently an entity appears when leading AI systems are asked natural questions about its sport, league, or peer group.
What we measure: Bowstone runs a structured battery of 50+ prompts across four AI platforms every week — Claude (Anthropic), GPT-4o (OpenAI), Gemini (Google), and Perplexity (Perplexity AI). Prompts are designed to resemble questions a sports fan, journalist, analyst, marketer, sponsor, or executive might naturally ask, organized into five categories:
Bowstone does not prompt models to mention particular entities. The goal is to observe which brands surface naturally.
How it's scored: For each prompt run, Bowstone records the entities mentioned by each model. An entity's AI Visibility score reflects its mention share — the proportion of tracked mentions attributable to that entity — combined with a cross-model consistency bonus for entities appearing across several AI systems rather than concentrated on one.
Scores are then normalized within the appropriate comparison group (see BX League vs. BX Overall below). This prevents structurally larger sports — particularly leagues generating substantially more overall AI discussion — from overwhelming entities in smaller sports simply because their raw mention universe is larger.
Why it matters: AI systems increasingly act as intermediaries between a question and the information a user receives. That makes inclusion itself measurable. A sports brand that appears repeatedly across models and across different types of questions has a different AI footprint from one that appears only occasionally or only on a single platform. Bowstone measures that footprint over time.
Weight: 65% — AI Visibility receives the largest BX weighting because it is the most distinctive signal Bowstone measures and remains substantially less developed as an analytics category than conventional social or market metrics.
Social Presence measures current audience momentum across Instagram and Twitter/X. The objective is not to reward the entity with the largest existing following — it's to identify who is gaining attention now.
What we measure: Bowstone collects weekly follower counts for tracked entities on Instagram and Twitter/X. The signal uses week-over-week growth rather than absolute follower count. An entity with 2 million followers gaining 50,000 in a week may demonstrate greater current momentum than an entity with 20 million followers gaining 10,000.
How it's scored: Follower growth is calculated as week-over-week percentage change and normalized within the relevant peer group. Growth rates above 25% week-over-week are winsorized at 25% when determining the normalization ceiling, preventing viral spikes among newly-signed or small-base athletes from distorting scores for established accounts.
Data collection: Social data is collected weekly by a dedicated research contractor and imported into the Bowstone database each Monday morning. Baseline collection began in July 2026. Because social growth is inherently volatile, weekly changes reach full reliability from week four of data collection onward, once short-term anomalies (viral moments, roster announcements, playoff runs) become distinguishable from sustained momentum.
Platforms not currently included: Facebook and TikTok are not part of the current Social Presence calculation.
Weight: 35% — Social Presence provides a useful corroborating signal. An entity gaining both AI visibility and social momentum presents a different pattern from one experiencing an isolated spike on only one surface.
Market Signal incorporates franchise valuation data and is displayed alongside BX rather than included within the composite score.
What we measure: Bowstone uses the most recently published franchise valuations from leading sports-business sources, currently including Forbes and Sportico. Valuations are updated as new reports become available.
How it's scored: A team's valuation is evaluated relative to the median valuation within its league. This allows market context to be interpreted within the economics of the relevant sport rather than treating nominal franchise values across leagues as directly equivalent — a $1.5B NHL franchise may be among the most valuable in hockey while sitting far below the NFL average.
Players and other individual entities: Market Signal is not assigned to individual athletes, coaches, or other entities for which franchise valuation has no direct equivalent.
Why it's separate: Franchise values don't change weekly the way AI Visibility and Social Presence do, and they have no equivalent for individual players. Bowstone treats valuation as a comparison signal, not a BX ingredient — which is useful because divergence itself can be meaningful. A highly valuable franchise may have surprisingly weak AI visibility. A less valuable franchise may dramatically outperform its market position in AI conversation. Those gaps are exactly what the Correlation view is built to surface.
When Social Presence data is unavailable for an entity, its weight is reassigned to AI Visibility rather than leaving part of the score empty.
| Situation | AI Visibility | Social Presence |
|---|---|---|
| Social data available | 65% | 35% |
| Social data unavailable | 100% | — |
AI-generated answers are probabilistic. The same question can produce somewhat different responses across repeated runs even when the wording and model remain unchanged. Bowstone treats that variance as something to measure, not ignore.
Periodically, Bowstone repeats each prompt ten times in a single session across all four AI models with nothing else changing, and compares which entities appear, how frequently, and how consistently those results repeat. That produces a prompt stability score from 0 (effectively random) to 1 (identical across repeated runs).
Current calibration results:
| Prompt Type | Example | Stability | Confidence |
|---|---|---|---|
| Value | "Who are the five most valuable NFL franchises?" | 0.88 | High |
| Top Teams | "Which NBA teams are the best right now?" | 0.77 | Medium |
| Top Players | "Name the top NFL players right now." | 0.74 | Medium |
| Brand | "Which NFL teams have the strongest brand?" | 0.63 | Medium |
These figures reflect Bowstone's July 2026 calibration study across Claude, GPT-4o, and Perplexity. Gemini was excluded from this round due to a model deprecation issue on our end, since resolved — future calibration studies will include all four platforms. Calibration is repeated quarterly and after any major AI model release.
An entity's confidence score is derived from the stability of the prompts contributing to its visibility score, weighted by the entity's mentions across those prompt types.
Confidence thresholds:
Each entity in the dashboard displays a confidence indicator — a colored dot next to the entity name. Teal indicates high confidence, yellow indicates medium, and red indicates low.
BX is intended to identify direction and patterns, not to turn a single weekly score into an absolute statement about brand value. A few principles matter when interpreting movement.
Movement should be read alongside confidence. A meaningful BX movement associated with a highly stable prompt set is more informative than the same movement generated from lower-stability prompts. Raw mention counts should also be considered — when BX movement and raw mentions move materially in the same direction, the signal is stronger.
Different AI platforms can react at different speeds. AI systems do not all access, retrieve, weight, and synthesize information the same way. Perplexity draws on live web sources and can reflect breaking news within days. Claude and GPT-4o draw primarily on training data and may take weeks or months to update following a major event. The LeBron James signing in July 2026 is a clear example — Perplexity's mention counts shifted within the first weekly run after the signing, while Claude and GPT-4o initially reflected earlier team associations. Bowstone does not expect identical movement across every model, and treats that divergence as a finding, not an error.
New entries are baselines, not movers. Entities entering the index for the first time — flagged with a NEW badge in the Top Movers table — have no prior run to compare against. Bowstone recommends observing at least two to three subsequent runs before drawing conclusions about trend direction.
Model divergence is signal, not noise. If one AI platform repeatedly surfaces an entity while another rarely does, that may indicate the entity's AI visibility is platform-dependent — which itself matters to brands evaluating where and how their narratives are reaching audiences.
Bowstone calculates two BX variants from the same underlying AI Visibility and Social Presence signals and the same 65/35 weighting — the difference is the comparison pool.
BX Overall compares each entity with entities of the same type across Bowstone's entire sports universe — players vs. players, teams vs. teams, leagues vs. leagues, brands vs. brands — regardless of league. This avoids comparisons in which structurally different entity types distort one another: league-level entities (e.g. "NBA" as its own tracked entity) draw far more raw mentions than any individual player or team simply by being the broader subject of more prompts, so pooling every type together would let leagues set the ceiling for players and teams. BX Overall is the headline score displayed on entity pages.
BX League evaluates an entity relative to all entities — across every type — within its own league. It answers: how strong is this entity's AI and social position within its immediate sports ecosystem?
Bowstone currently covers: NFL, NBA, MLB, NHL, MLS, WNBA, and College Football.
Interpreting the number: A BX score is a normalized 0–100 composite score. A BX League score of 78 means the entity earned 78 points on Bowstone's normalized weighted-signal scale relative to the relevant league comparison pool. It does not mean 78 mentions, 78% social share, or necessarily the 78th percentile. BX Overall uses the same 0–100 framework, but compares the entity against its type peers across all covered sports.
Rankings: Wherever Bowstone displays rankings, entities are ranked within comparable entity-type pools — players against players, teams against teams — never mixed together, regardless of which BX variant is being ranked.
| Platform | Developer |
|---|---|
| Claude | Anthropic |
| GPT-4o | OpenAI |
| Gemini | |
| Perplexity | Perplexity AI |
Bowstone's platform universe may change over time. Models can be added, replaced, or retired as user behavior and the AI-discovery landscape evolve. The objective is not to permanently privilege a particular model — it's to measure the AI surfaces through which meaningful sports discovery is taking place.
Bowstone runs 50+ active prompts per weekly cycle across five categories — top teams, top players, brand, value, and news (see AI Visibility above for detail on each). The battery is intentionally broader than a single ranking question: an entity that appears across different prompt types and multiple AI platforms demonstrates more durable AI visibility than one appearing only in response to a narrow query.
| Signal | Frequency | Source |
|---|---|---|
| AI Visibility | Weekly (Sunday night) | Automated query pipeline |
| Social Presence | Weekly (Monday morning) | Manual contractor import |
| Market Signal | As published | Forbes / Sportico |
AI query runs are executed via two automated cron jobs every Sunday night, split across the first and second halves of the prompt battery to stay within platform rate limits. Results are processed and available in the dashboard by Monday morning.
No index captures every dimension of brand strength. Bowstone currently does not include the following within BX:
Athletic performance. Win-loss record, standings, and championships are not direct BX components. Those events may indirectly affect AI or social visibility, but BX measures the resulting brand signal rather than awarding points for athletic success itself. A struggling team with strong historical brand equity may score well; a championship team in a small market may score lower than expected.
Sponsorship revenue or deal activity. Commercial sponsorship activity is not currently part of BX — this distinction is intentional and can itself produce useful findings. A major sponsorship announcement, for example, may generate substantial commercial activity without producing any immediate change in AI visibility. This is a planned future signal.
AI sentiment. The current AI Visibility signal measures whether and how consistently an entity is mentioned, not whether those mentions are positive, negative, or neutral. Sentiment classification is a planned future enhancement.
Traditional search volume. Google search volume is not currently part of BX. Bowstone is not designed to replace traditional search analytics — it measures a different layer of discovery: which entities AI systems choose to surface in generated answers.
Traditional search, AI visibility, social momentum, sponsorship activity, and market value are separate signals. Understanding where they agree — and where they diverge — is one of the central purposes of Bowstone.
Bowstone tracks 250+ sports entities across the covered leagues, including teams, athletes, leagues, and selected sports brands. The tracked universe is updated periodically to reflect roster changes, emerging athletes, new coverage areas, and entities warranting inclusion.
Bowstone began its initial data collection in July 2026. AI Visibility can be measured beginning with the first collection run, while trend analysis becomes more informative as additional weekly observations accumulate. Social growth rate data reaches full reliability from week four onward, as the baseline period required for meaningful week-over-week comparison is established.
Every BX observation is dated so that users can distinguish current measurements from historical results. For that reason, Bowstone emphasizes direction, persistence, and cross-signal comparison over isolated weekly scores.
Bowstone is built around a simple premise: brand visibility is no longer occurring on one surface.
A sports organization can have enormous social reach but weak AI visibility; high franchise value but limited AI presence; a major commercial announcement that produces little change in generative search; or rapidly increasing AI visibility before traditional market indicators react. None of those signals invalidates the others — they measure different things.
As search and discovery become increasingly AI-mediated, brands need a way to measure not only whether audiences can find their websites, but whether AI systems recognize them as relevant enough to become part of the answer.
Bowstone measures that emerging layer of sports brand visibility.
Bowstone's methodology is informed by the changing search and discovery environment. Recent industry and platform data show:
These developments do not imply the end of traditional search. They support Bowstone's narrower premise: AI-generated answers have become large enough, distinct enough, and measurable enough to warrant their own visibility metric.