The research program began with scale, but scale was never the conclusion.
The inventory contains 1,779,805 unique discussion URLs from approximately ten years of FunkyMEDIA-owned publishing infrastructure. That gave us a large body of commercial discussion to work with, but the first useful step was not to count everything as a Brand Mention.
It was to ask what different kinds of brand visibility actually look like.
A link can point to a company without visibly naming it.
A brand can be named without any link.
A recommendation can repeat wording already used elsewhere.
A brand can enter a conversation before anybody recommends it.
One mention can contain only a company name, while another can connect the same brand with a product, problem, feature, use case and result.
Those differences became the structure of the research.
The pages in this section each isolate one of those questions.
Start with the flagship study
The 1.7 Million Conversations Brand Mentions Study
The flagship report explains the full research architecture and brings the main findings together in one place.
It separates the Full Research Inventory, the Representative Validation sample, Gold Dataset v1 and the Macro Subset, so the scale of the corpus is not confused with the smaller datasets used for deep analysis.
It is the best starting point for understanding why the research exists and how the different findings connect.
Six questions that change how Brand Mentions are read.
Each study isolates one measurement problem so that links, repetition, intent and semantic context are not collapsed into one headline number.
Research question 1: Are all Brand Mentions the same kind of evidence?
Not All Brand Mentions Are Equal
Gold Dataset v1 contains 416 linked explicit Brand Mentions, 328 unlinked Brand Mentions and 378 Linked Entity References.
Those records can look similar in a spreadsheet, but they describe different things.
The first two explicitly name the brand.
The third may identify the entity only through the linked destination.
This research page explains why a single raw mention count can hide several different kinds of visibility and why the type of mention has to be separated before volume can be interpreted.
Research question 2: Does a link prove that a brand was explicitly mentioned?
A Link Is Not Always a Brand Mention
The Gold Dataset shows that hyperlink presence and visible brand naming overlap, but they are not the same measurement.
Some pages link to an entity-owned destination without naming the brand in the visible wording.
At the same time, 328 explicit Brand Mentions in Gold Dataset v1 were unlinked.
That means a link can exist without an explicit Brand Mention, and an explicit Brand Mention can exist without a link.
The research isolates those cases and explains why link analysis should not be used as a shortcut for Brand Mention analysis.
Research question 3: Does a high mention count represent many different statements?
The Repetition Problem
Among 744 explicit Brand Mentions in Gold Dataset v1, 63 were classified as exact formulaic mentions and grouped into 21 exact clusters.
Another 681 remained unique-like under the current method.
This does not mean that unique-like wording is organic.
It means the current repetition method did not classify it as an exact or near-text duplicate.
The finding matters because a high number of mentions can represent either a broad range of language or repeated formulations.
Those are different information environments, even when the headline volume is identical.
Research question 4: When does a brand enter the decision process?
Brand Mentions Before the Decision
A recommendation is only one stage of a commercial conversation.
In the Gold Dataset calibration layer, 248 records contained question or consideration signals, compared with 179 recommendation signals.
That means brands often appear while users are still verifying, considering or trying to understand an option.
The research separates those earlier stages from direct recommendation so that brand visibility is not measured only at the final endorsement stage.
Research question 5: How much information can exist inside one Brand Mention?
The Anatomy of a High-Context Brand Mention
One mention can contain much more than a company name.
A single passage may connect a brand with a product, problem, feature, service, use case, location, experience or claimed result.
The semantic calibration extracted 536 relationship records and 276 unique Brand → Relation → Entity pairs.
This research page looks at what makes some Brand Mentions much richer in context than others and why mention frequency alone cannot capture that difference.
Research question 6: Is one mention the same as one conversation?
One Mention vs One Conversation
A single Brand Mention can be analysed as one occurrence, but a longer conversation can add information that the isolated sentence does not contain.
A thread may introduce alternatives, objections, product details, another brand, a later experience claim or a reported result.
This research question looks at how the surrounding conversation changes the amount of context available around the same brand.
The page remains useful only where the source material supports that distinction; it is not used to claim that longer discussions are automatically better or more trustworthy.
What connects these studies
The six research questions are different views of the same problem.
A Brand Mention is not simply a binary event that either happened or did not happen.
It has structure.
The brand may or may not be explicitly named.
A link may or may not be present.
The wording may be repeated or varied.
The mention may appear during consideration, recommendation or warning.
The surrounding text may contain almost no additional information or may generate several semantic relationships.
The research pages separate those dimensions so that one headline metric does not have to carry all of them at once.
What this section is not
This is not a list of universal ranking factors.
The studies do not claim that one mention type automatically improves AI visibility, search visibility or commercial performance.
They document structures observed in the research datasets.
That distinction is important because the first goal is to measure Brand Mentions consistently.
Questions about downstream effects can only be tested meaningfully once the underlying units are clear.
Where to go next
If you want the complete narrative, begin with the flagship study.
If you want the strongest single finding, start with Not All Brand Mentions Are Equal.
If your interest is link analysis, read A Link Is Not Always a Brand Mention.
If you are evaluating publication volume, read The Repetition Problem.
If you are interested in the decision journey, read Brand Mentions Before the Decision.
If you want to understand semantic richness, continue to The Anatomy of a High-Context Brand Mention and then to the Brand Context Graph.