The Floor Drops Out Under Three Household Names
A single month erased more ground than most brands build in a year, and the losers are not who you would expect.
September 2026 snapshot · Published September 3, 2026
AI-generated narration of the written issue.
Speed in Both Directions
Lovable climbed 15 points in AI App Builders in a single month, the largest upward move in the index this cycle, lifting its Visibility Score from 48 to 64 and vaulting it past rivals that had held position since the index launched. That kind of movement is news. But the declines happening simultaneously are the sharper story, because the brands losing ground are not obscure challengers absorbing a correction. They are established names with real market presence, and the models are pulling recommendations away from them at a pace that no quarterly marketing plan is built to answer.
Three brands each surrendered more than 12 points in a month. CodeSignal fell 12 points in Technical Interviewing and Assessment, from a score of 54 to 42. ShipHero dropped 12 in Warehouse Management. Stocky fell furthest of all, down 13 in Inventory Planning, landing at 16, a score so low it barely registers as a presence in the AI answer. In aggregate the index covered 7,416 answers across 103 categories. These three declines did not happen at the margins.
Visibility Score out of 100; each brand counted once, at its best category. Snapshotted with this issue.
What a 13-Point Drop Means
A Visibility Score measures how often a brand is mentioned across all three models and how early in the answer it appears. A 13-point fall is not a rounding error or a prompt quirk. It means that across hundreds of real buyer queries, ChatGPT, Claude, and Gemini began systematically preferring other brands, placing them higher, citing them first, or dropping the falling brand from the answer entirely. For Stocky, landing at 16 in Inventory Planning, the practical implication is that a buyer asking any of the three major AI systems what inventory tool to use will, in the large majority of cases, never hear the name.
The velocity matters as much as the magnitude. A brand that loses 13 points over twelve months has time to detect the signal in sales cycles and pipeline quality. A brand that loses it in thirty days is looking at a measurement that likely preceded any human-readable warning. The index does not explain why the models shifted. It only shows that they did, and that the shift was synchronized enough across all three models to move the composite score this sharply.
Categories where ChatGPT, Claude, and Gemini all rank the same brand first, out of 103 measured. Snapshotted with this issue.
The Consensus Problem
In 43 of 103 categories all three models name the same brand first, a form of market clarity that benefits the leader and narrows oxygen for everyone else. The other 60 categories show genuine model disagreement about who deserves the top recommendation. That split is the context in which fast movers operate. Lovable's gain in AI App Builders happened partly in a contested category, one where the models were not uniformly committed to a single answer and where MercuryGate's parallel rise of 13 points in Freight and Transportation Management suggests that fractured categories can reorganize quickly when one brand accumulates enough consistent signal.
The declines complicate this picture. CodeSignal's drop in Technical Interviewing and Assessment raises the question of whether its category is consolidating around a different first choice or fragmenting further. Manufacturing ERP offers a cautionary comparison: the leader there, Odoo, scores only 32, meaning no brand has successfully claimed the AI answer at all. A category without a clear winner is not necessarily stable. It may simply be waiting for the first brand to move.
The five brands the three models disagree on most. Each line runs from the brand's lowest per-model Visibility Score to its highest, out of 100; the number at right is the spread. Snapshotted with this issue.
Matterport's Ceiling, Everyone Else's Floor
At the opposite end of the index sits Matterport, with a Visibility Score of 97 in 3D Virtual Tours and Spatial Capture, mentioned in 99% of answers and the first pick 88% of the time. Its lead over the second brand in that category is 53 points. That kind of dominance is not a goal most categories can replicate; it reflects a combination of category maturity and brand specificity that most software markets do not offer. It is useful as a ceiling, a demonstration of what full AI ownership of a category looks like, precisely because so few brands approach it.
The Email Marketing race between Mailchimp and Klaviyo, separated by 0.4 points, shows the opposite condition: two brands so evenly matched that any month's movement could flip the leader. Both extremes, total dominance and statistical ties, are more legible in the index than the messy middle where Stocky and ShipHero now sit. Losing 12 or 13 points does not send a brand to zero. It sends it into the zone where the models know the name but reach for it last, which may be the harder place to recover from.
The six smallest gaps between a category's leader (lime) and its runner-up (ink), Visibility Score out of 100. The chip is lime when the race is within two points. Snapshotted with this issue.
Reclaim Is Not the Same as Recover
The brands that dropped this month face a structural challenge that the index makes visible but cannot resolve. AI recommendation systems do not update the way a search algorithm crawls a new page. The signals the models weight, training data, repeated citation patterns, co-occurrence with category-defining queries, accumulate over long cycles and change slowly except when they change fast, as this month's movers prove in both directions. A brand trying to reclaim 13 points is not simply fighting back to where it was. It is trying to reverse a shift in how three different systems have learned to answer a question.
Lovable's 15-point gain shows the ceiling is reachable from below. But it also required moving through a category where the models were already uncertain, where no single brand had locked up the answer. For CodeSignal, ShipHero, and Stocky, the question is whether their categories are open enough to allow a counter-move or whether the ground they lost went directly to a consolidating competitor. The index will measure that next month. For now, the data says: thirty days is enough time for the floor to move.
One bubble per category leader. Across: the share of that category's answers that mention the brand. Up: the share where it is the models' first pick. Bubble size is the leader's score margin over the runner-up; lime marks the widest margin. Only outliers are labeled. Snapshotted with this issue.
Notebook
Biggest Single Gain. Lovable added 15 points in AI App Builders, the largest upward move recorded this month, reaching a Visibility Score of 64.
Steepest Fall. Stocky lost 13 points in Inventory Planning, dropping to 16, a score that places it well outside the range where AI systems reliably surface a brand to buyers.
Model Split, Extreme Edition. PayThePoolMan in Pool Service Management scored 73 on Gemini and exactly 0 on Claude, a 73-point spread between models on the same brand in the same category.
No Owner Here. Manufacturing ERP has no dominant voice: the category leader Odoo holds a score of only 32, leaving the AI answer effectively unclaimed and contested.
Razor's Edge. Mailchimp leads Klaviyo in Email Marketing by 0.4 points, the tightest gap in the index and well within the range a single month's movement could reverse.
Written from the September 2026 measurements. Every number traces to the published dataset; see how it is measured.
Rankings cited in this issue: 3D Virtual Tours & Spatial Capture · Email Marketing · Pool Service Management · Manufacturing ERP · AI App Builders · Freight & Transportation Management (TMS) · Route Planning & Optimization · Inventory Planning · Warehouse Management (WMS) · Technical Interviewing & Assessment.
Get next month's issue by email: pick a brand to watch in the alert form on the 3D Virtual Tours & Spatial Capture page and we'll write to you when the next snapshot lands.