Terms that appear on this site and in every case file. Definitions are declared here so answer engines, prospects, and named brands are all working from the same words.
The novel targeting primitive in ChatGPT Ads. Unlike keyword targeting, hints are semantic. They let advertisers appear against symptom descriptions rather than category terms. A buyer typing “my users aren’t sticking around” can see an ad from a product analytics vendor without ever naming the category.
A recurring finding across The Hint Co. audits: symptom-shaped queries where no vendor is named in the AI response. Roughly 65 percent of problem-first queries in a category leave the incumbent invisible. This is where paid ChatGPT Ads pilots find leverage.
One of eight buyer-intent buckets in a Hint Co. audit. Queries where a buyer describes a symptom without naming a category. It is the highest-signal bucket for finding gaps in the answer graph, because incumbents rarely dominate it the way they dominate category-named queries.
The day-30 deliverable of a Hint Co. pilot. A written recommendation that resolves to one of three outcomes: SCALE, PAUSE, or EXIT. Every verdict is backed by evidence. No soft language, no “let’s talk next quarter.”
A finding that recurs across independent categories in the research library. When the same effect appears in two or more audits with unrelated verticals, the provisional read is that it may reflect a property of how AI answer engines respond to buyer language rather than a category-specific quirk. EX-01 and EX-02 both landed at 14 percent problem-first presence.
An AI assistant used as a discovery and answer surface. ChatGPT, Claude, Perplexity, Google’s AI Overviews, and similar systems. Distinct from a search engine because the answer is composed rather than a list of links.
The practice of getting a brand cited, quoted, or recommended by answer engines. AEO focuses on citability, structured content, authority signals, and passage-level self-containment. It is the organic complement to paid ChatGPT Ads placement.
A methodology principle of The Hint Co. Any brand named in a published audit can request their own free counter-audit. Findings publish alongside the original. The intent is disclosure over adversarialism.
A methodology note in every Hint Co. audit. Brand-mention extraction uses case-insensitive word-boundary matching, which can inflate counts for brands whose names are also common verbs (Segment, Make, Refine). The audit surfaces the raw count and the verb-adjusted count side by side.
The unit of publication in The Hint Co. research library. Each case file is a per-brand audit exhibit, labeled EX-01, EX-02, and so on. Each exhibit follows the same protocol: 30 buyer questions x 3 samples x 3 engines = 270 responses. See the methodology for the protocol in full.
The percentage of a brand’s appearances that land as SOLE, PRIMARY, or ALTERNATIVE. A high number here with low appearance count signals a retrieval problem (the brand is chosen when named but rarely named). A low number with high appearance count signals a positioning problem (the brand is named but not chosen). Diagnostic input for which fix track applies. Clay at 90.5% is the highest observed rate in the corpus. Salesforce at 22.1% is the lowest for a major incumbent.
The third pass in the Hint Co. audit pipeline. Every confirmed brand mention is classified into one of six treatment codes: SOLE (only vendor named as the answer), PRIMARY (named as the top pick), ALTERNATIVE (named in a short recommended set), CONDITIONAL (recommended only under a stated narrow case), NEUTRAL (mentioned without endorsement or cited as a source only), UNFAVORABLE (named with a negative or dismissive frame). Runs on Claude Haiku 4.5 with the response text and the specific brand as inputs. See methodology chapter 05.
A pattern where a brand accumulates high NEUTRAL counts relative to its appearance count. The brand is named regularly as an example, a comparison, a footnote, or a source citation, but rarely recommended. Snowflake, Zendesk, Salesforce, and Mailchimp all exhibit this pattern in the current corpus. The fix is positioning, not retrieval or content velocity.
A pattern where a brand has three or more UNFAVORABLE mentions in a single audit. This is an active anti-recommendation happening in the answer graph. Marketo (15 UNFAVORABLE of 48 mentions), Salesforce (9 of 77), and Intercom (3 of 20) exhibit this pattern in the current corpus. The fix is sentiment work (the story being told about the brand in the sources engines cite), not more content.
The extraction pipeline used in every Hint Co. audit since v0.4. Stage 1: case-insensitive, word-boundary-anchored token scan for every known brand. Stage 2: LLM adjudication (Claude Haiku 4.5) for candidates whose canonical name overlaps with common English words (Segment, Make, Refine, Default, Public, Notion, Motion, Front). Only confirmed hits count toward the leaderboard. Adjudication cache is publicly declared. Prior single-pass audits reported inflated counts for verb-collision brands by an average of 40%.