Start with the research scope
You provide a public website, confirm a company profile and select competitors and buyer questions. Those choices define the brand panel and the decisions the research should inform. A snapshot is a collected sample for that scope, not a census of a market.
Results depend on the selected brands, questions, keywords, market and collection settings. Compare snapshots with matching scope. Changing the competitor set or question panel can change a share even when the brand’s underlying activity has not changed.
Sources and collection
Website evidence comes from readable public pages and search results. Earned-media research samples public news indexes over your selected research timeframe, excludes identified brand-owned domains and reviews brand matches. Social research samples public keyword matches on supported networks and excludes identified owned posts from its share calculation. Ownership identification and platform access can be incomplete.
Organic search uses Google ranking and search-volume records supplied through DataForSEO. AI answer collection targets ChatGPT, Gemini, Copilot, Perplexity, Google AI Mode and Google AI Overview through a search-answer collection provider. The snapshot identifies the collected assistants, questions and available citations. If a separately configured model API is used instead, its model identity is retained; an API response should not be treated as a consumer-interface result.
Paid-ad research uses bounded public Google, Meta and LinkedIn ad-library samples. An observed ad establishes the message in that sample. It does not establish spend, reach, conversions or effectiveness.
Dates, refreshes and missing results
A source’s publication date and the date Digittle retrieved it are different. Collection timestamps describe when an observation was captured; publication dates are retained where available. An AI answer is an observation at collection time, not a permanent answer from that assistant.
Weekly research starts Monday at 5 AM Eastern. On-demand research uses a separate per-brand monthly allowance. Essentials and Agency allow a 2-year window; Intelligence allows 5 years. Provider archives and sampling limits determine the evidence available within it. Organic rankings may be reused within a week, and monthly search-volume weights may be reused within the same month and scope. A refresh is not a claim that every upstream record was created that day.
Collection warnings identify missing or partial inputs. A failed request is not zero visibility. Incomplete channels cannot establish a complete composite score. Available source dates and sample counts should be read alongside the percentages.
Measurement definitions
Share of voice shows your share of the visibility measured within a defined comparison set. It does not represent your share of the entire market. The calculation and comparison set differ by channel.
| Measure | Calculation and counting rule |
|---|---|
| Earned media | Articles mentioning your brand ÷ the sum of article counts for all tracked brands × 100 Each tracked brand counts once per retained article. Repeated mentions add no credit. Identified brand-owned publications are excluded. |
| Social conversation | Posts mentioning your brand ÷ the sum of post counts for all tracked brands × 100 Each tracked brand counts once per retained post. Engagement does not affect share. Identified brand-owned posts are excluded by default; the include-owned control changes that comparison. Post counts are pooled across the platforms included in the view, rather than averaging platform percentages. The saved DVS uses the snapshot’s eligible posts and excludes identified brand-owned posts; view filters do not change it. |
| AI share of voice | Collected AI answers mentioning your brand ÷ the sum of answer counts for all tracked brands × 100 Each configured brand name is matched without regard to letter case and counts once per answer when named in the answer text. Source URLs, image labels and citation-only Sources or References sections are excluded. Repeated mentions add no credit. Returned source links are counted separately; a citation-only result does not add a brand mention. Answer counts are pooled across the assistants included in the view, rather than averaging assistant percentages. DVS uses this share from the saved set of answers, not answer presence or citation counts; view filters do not change it. |
| Estimated search visibility | Your domain’s estimated visibility ÷ the sum of estimated visibility for all observed ranking domains × 100 For each tracked keyword, monthly search volume is multiplied by a benchmark click-through rate for the domain’s observed organic position. These estimates are summed across usable keywords. Each domain contributes its highest-ranking organic result within positions 1–20, not every result. Tracked-brand subdomains are grouped with the tracked domain. The denominator includes other observed ranking sites. This measures estimated visibility, not actual traffic. Paid placements and AI Overviews are excluded. |
| Paid advertising | No advertising share-of-voice percentage is calculated. Public ad-library samples show observed ads and messaging. The samples used here do not establish a consistent denominator for calculating share of advertising impressions or spend. |
Counting example: An article mentioning three tracked brands contributes one appearance to each brand and three appearances to the denominator.
AI answer presence: Answers naming your brand ÷ all collected answers × 100. This uses a different denominator from AI share of voice.
Reporting scope and limitations
Reporting period: articles and posts use available publication dates within the selected research window. AI answers and organic rankings describe their collection dates, not a reconstruction of past answers or positions. The monthly earned-media matrix groups the selected report’s dated articles by publication month; it does not establish complete monthly archive coverage.
Comparison set: earned media, social and AI share compare your brand with the competitors frozen in the saved research snapshot. Estimated search visibility also includes other observed ranking domains. Changing the brands, keywords, questions, assistants, market, counting version or filters changes the comparison. AI trend comparisons exclude snapshots with a different mention-counting version.
Duplicate handling: identified duplicate news URLs and matching normalized titles from the same publisher are merged, retaining the combined brand matches. Posts are deduplicated by retained item identity and normalized URLs. Each brand counts at most once within a retained item. Syndicated stories on different publisher domains can remain separate; duplicate detection is not perfect.
Source coverage: these are bounded public-source samples. Private posts, inaccessible pages, missing dates, provider limits and collection failures can leave gaps. Partial collection is labeled. Not available means the denominator is zero or collection is unavailable or failed; 0% is used only when a usable comparison exists and the brand has no counted visibility in it.
From an observation to a recommendation
Recommendations distinguish an observed source signal from its interpretation and a proposed action. Supporting excerpts and source links let you inspect the basis for a finding. Before implementing a content recommendation, review your existing material and verify the relevant product claims.
Lift and payoff assessments help prioritize work; they are estimates, not measured financial returns. A recommendation can be saved, annotated and marked implemented. Those actions record your team’s decisions. They do not validate the underlying hypothesis or prove an outcome.
Digittle Visibility Score™
Share of voice tells you who is loudest. Share of visibility shows who buyers are likely to see, across media, social, AI answers and search.
DVS™ combines organic search, AI visibility, earned media and social conversation into one visibility score from 0 to 100. It compares visibility within the saved research scope, not your share of the entire market. The four inputs are modeled organic search share, AI share of voice, earned-media share and social-conversation share. AI share of voice counts brand appearances across collected answers, rather than the percentage of answers naming your brand. Paid advertising is researched separately and is excluded.
Social platforms feed the social conversation measure. AI assistants feed the AI visibility measure. Within each measure, counts are pooled across the collected posts or answers; platform and assistant percentages are not averaged. A source with more qualifying observations contributes more to that measure, but does not receive a separate DVS weight.
DVS uses the eligible evidence saved with the research snapshot. Identified brand-owned articles and posts are excluded. Filtering a channel view, including its include-owned control, does not recalculate the saved DVS. Its comparison set, collection scope and policy remain attached to that snapshot.
DVS uses a fixed, approved policy version and server-side arithmetic. AI does not select the weights or assign the score. The same saved evidence, comparison set, settings and policy produce the same numeric scores and ranks. A new collection can produce different results because its evidence has changed.
Digittle records the method version, comparison scope, collection status and calculation time. New calculations also retain a private audit record identifying their inputs, policy and results. Scores are rounded once to one decimal place; equal displayed scores share a rank. A changed measurement scope or scoring policy must be treated separately when comparing results.
All four visibility measures are required. Missing or failed measures are not assigned zero, and weights are never redistributed to force a score. A genuine observed absence can contribute zero when its measure has a usable denominator. Partial coverage is disclosed beside a provisional score.
The current approved policy weights each of the four measures at 25%. This weighting has not been empirically validated against external business outcomes. DVS™ measures visibility in the stated research scope; it is not total-market share, audience reach, customer preference or a validated predictor of traffic, leads or revenue.
Limits of the measures
Samples may miss relevant sources, misidentify brands or reflect platform and language bias. AI answers can vary between requests and can contain incorrect claims. A citation in an AI answer is not independent verification of the cited claim. Absence from a collected sample does not prove absence from the market.
Digittle Visibility Score™ (DVS™) uses a fixed, versioned calculation under an approved scoring policy. It requires measured results from all four visibility measures. It can use partial samples, with incomplete coverage and available sample counts disclosed beside the score. If any measure is missing or failed, the score is withheld; weights are never redistributed to force a result. The score describes measured visibility within the selected comparison scope. Read it alongside the four individual measures. DVS™ is not a validated predictor of sales, revenue or recommendation success.
The interactive workspace walkthrough uses fictional brands, invented counts and illustrative recommendations. It demonstrates the format and interactions, not customer performance. Customer reports reflect their stated scope and collection limits; edited narrative is not independently verified.