The GBMF state is not a score. It is a diagnosis. The reason the framework reports three separable measures and combines them into a state taxonomy is that each off-target state implies a different remediation path. A brand should not begin any remediation programme without identifying which state it is in, because the wrong intervention against a misdiagnosed state wastes effort and sometimes makes the picture worse.
The companion articles in this series cover the framework (three measures explainer) and the measurement walkthrough (how to run a first GBMF scan). This article walks the off-target states one by one and names the remediation for each. Where the tactical detail already lives in an existing article, this article points to it rather than repeating it.
The diagnostic principle
Two examples make the diagnostic argument concrete.
A brand diagnosed as Misrepresented AI Pariah (visible, inaccurately described, negative sentiment) that responds with a visibility push amplifies the misrepresentation. AI surfaces the brand more often with the same wrong description, and the negative reception often tracks the misrepresentation itself. The remediation is alignment work first.
A brand diagnosed as Undiscovered AI Contender (accurately described, neutrally received, just rarely surfaced) that responds with a brand-perception campaign solves the wrong problem. The reception is fine; the brand isn’t reaching the answer space. The remediation is reach.
Most teams over-invest in visibility regardless of state. The state diagnosis is the discipline that interrupts the default reflex.
AI Absent: exposure first
A brand classed as AI Absent is not on the shortlist. The buyer’s AI session never names it. The remediation is exposure. The work is building the third-party signal density that AI engines actually weight when assembling answers.
The evidence on what matters is reasonably consistent. Entities.org’s synthesis of Ahrefs (75,000 brand study) and Yext (17.2 million citations) found that brand mentions correlate with AI visibility at r=0.66, while backlinks correlate at only r=0.10. 5W Public Relations’ AI Platform Citation Source Index 2026, analysing 680 million citations across major platforms, reported that the top 15 domains capture 68% of consolidated AI citation share. Reddit is the top citation source at approximately 40%. Wikipedia accounts for 26 to 48% of ChatGPT’s top-10 cited domains. AI engines appear to require corroboration across two to three independent authoritative sources before recommending a brand with consistency.
The remediation steps depend on the engine. The existing articles on getting cited in ChatGPT, getting cited in Perplexity, and getting cited in Google AI Overviews cover engine-specific exposure tactics, and content structure for AI citations covers the structural signals that improve citation rates across platforms.
One failure mode is writing more content on the brand’s own blog. Owned-property writing is necessary for context but does not move the corroboration count. The signal AI engines weight is independent third-party coverage, not internal publishing. A B2B SaaS brand that publishes ten new blog posts and zero new third-party appearances has not made itself less AI Absent.
Misrepresented quadrants: fix the facts first
A brand in Visible & Misaligned (high MV, low MA) or Unseen & Misaligned (low MV, low MA) has an alignment problem. The AI describes the brand inaccurately, and the remediation is fixing the underlying source signals AI is drawing from.
What to do when AI gets your brand wrong is the operational guide. It covers the source-category mapping (which fix to run first based on the error type), the platform timeline differences (Perplexity hours, ChatGPT base model 18 to 24 months), and the structural reality that some sources cannot be directly corrected. The companion article on why AI gets brands wrong covers detection and root causes.
The state diagnosis adds a sequencing argument. Visible & Misaligned is the case where the misrepresentation is already reaching the answer space at volume, and the brand is being remembered for the wrong things. Unseen & Misaligned needs alignment work before visibility investment. Facts first, then volume. Building exposure on an inaccurate baseline amplifies the misrepresentation rather than correcting it.
Teams regularly correct the brand’s own website when the misrepresentation lives in G2, Capterra, Wikidata, or a high-authority comparison article. The owned-property update is the fastest step but the smallest signal contribution. As Ishtiaque Ahmed at ZipTie.dev puts it, what AI knows is what’s frequently mentioned, not what’s officially correct. Updating the sources AI actually weighs changes what appears most often in its context; updating the brand’s own website does not.
Undiscovered quadrants: exposure on a clean foundation
A brand in Aligned but Unseen or with the Undiscovered prefix has the accuracy foundation already in place. AI describes the brand correctly on the few occasions it surfaces it; the gap is reach.
The remediation looks similar to AI Absent in tactic (third-party signal density), but the risk profile is different. Adding visibility on a Visible & Misaligned baseline amplifies misrepresentation. Adding visibility on an Aligned but Unseen baseline just adds correct mentions. The engine-specific exposure articles (ChatGPT, Perplexity, Google AI Overviews) apply directly, and the content structure article covers what to fix in the brand’s own properties so that AI engines have something usable when they do retrieve.
Undiscovered states often get treated as a content-quality problem when the gap is reach. The content may be fine; the issue is that AI doesn’t surface the brand often enough. The remediation is reach, not rewriting.
AI Pariah: sentiment investigation before counter-narrative
A brand classed as AI Pariah is described unfavourably across the answer space. The remediation requires understanding what the AI is saying negatively and why.
How AI platforms form brand sentiment covers the formation mechanism, including the structural risk that a small number of high-authority critical sources can dominate the sentiment signal for a brand. The remediation work for an AI Pariah is more about understanding the sentiment drivers than about adding generic positive coverage.
The sequence that tends to work:
- Run a targeted sentiment investigation. Which prompts produce negative descriptions? Which sources are cited in those descriptions? What is the AI actually criticising?
- Address the underlying issue if there is one. If the AI is criticising old pricing or a feature that no longer exists, the remediation is alignment work, and the fix AI brand wrong article walks through the source-category mapping.
- Build correct sentiment-driver coverage in sources AI actually reaches. Specific responses to specific critiques carry more weight than generic positive PR.
Volume of positive coverage will not drown out high-authority criticism. AI weights frequency, but it also weights source authority. Five new positive blog posts on low-authority domains do not outweigh one critical TechCrunch article. The remediation is targeted source work, not volume.
AI Wildcard sometimes needs no remediation
The Wildcard state, unlike Pariah and Absent, can be a viable shortlist position. A brand with mixed sentiment (both positive and negative shares ≥ 25%) is on the shortlist as the bold-but-divisive option. Buyers consider it knowing the tradeoffs.
This is the editorial framing this article adds to the diagnostic. The framework reports the state, not what to do about it. Not every Wildcard needs to become a Champion. A brand that is divisive for reasons aligned with its commercial positioning may be in exactly the state it wants to be in. A challenger brand criticised by incumbents for being unproven, or a premium product criticised on price, can be a Wildcard by design.
Where a Wildcard does need remediation:
If the negative side of the sentiment is misaligned (the AI is citing wrong reasons for the criticism), treat the case as a Misrepresented brand. The fix is alignment work on the misrepresentation, not on the sentiment itself; the sentiment often follows.
If the negative side is aligned but commercially harmful (the AI is citing real product weaknesses the brand could address), treat the product issue and watch the sentiment shift. Sentiment is a downstream signal of what people actually say; changing what they say requires changing what is being criticised.
Treating every Wildcard as a brand-perception emergency wastes budget. Some are. Some are the cost of category positioning, and the cost of trying to neutralise them is higher than the cost of holding them.
What this playbook will not achieve
Timelines are not service-level commitments. The companion article on fixing AI brand errors documents practitioner-observed timelines: Perplexity 24 to 48 hours, ChatGPT Search and Google AI Overviews 1 to 6 weeks, ChatGPT base model 18 to 24 months for entrenched errors. These are observed ranges from documented cases, not commitments any platform makes. A specific brand correcting a specific error may move faster or slower depending on the source mix.
Reversal is a real risk. Lily Ray’s April 2026 documentation of the AI slop loop showed that corrections can be undone by AI-generated content scraping the previous error and reintroducing it. A correction that propagates can disappear weeks later if the source balance shifts back. Treating remediation as a one-time project rather than a monitoring function leaves the problem unattended.
Some sources cannot be directly corrected. Comparison articles that are no longer maintained, cached pages with low crawl frequency, and coverage establishing a previous positioning will not respond to outreach. For brands with years of coverage behind a previous positioning, the accumulated old content actively competes with the update, and for a long time it will win.
State changes lag remediation. Even when the source signals shift, the engines’ own retrieval and training cycles mean a Pariah does not become a Contender in a week. The state taxonomy reports a brand’s current standing, not the trajectory.
The framework diagnoses standing, not commercial outcomes. A brand can become an AI Champion across the engine set and still lose deals for reasons GBMF does not measure. The state is a diagnostic input, not a sales forecast.
Per-engine variation persists. A brand can be a Champion on ChatGPT and a Pariah on Perplexity simultaneously. State labels are reported per engine and then aggregated; the aggregate hides the disparity. The engine-level numbers carry the diagnostic information.
Aiviara is building infrastructure for monitoring AI brand citations and factual accuracy across LLM platforms. Early access information is available at aiviara.com.