Has your GEO agency told you about TurboQuant?

When an AI model processes a sequence of tokens, it stores a Key vector and a Value vector for every token already processed. This structure is called the KV cache, the model’s working memory. The KV cache is also the biggest cost driver in long-context inference. Two structural problems made it increasingly expensive: attention cost is quadratic in sequence length (O(N²)), so doubling context roughly quadruples cost; and even with large context windows, models attend unevenly, favouring the start and end and skimming the middle. Brute-force context expansion was a dead end. Compression was the only viable path.

TurboQuant solves this in two stages. PolarQuant (stage 1) re-encodes the cache vectors in polar coordinates: radius and angle instead of full Cartesian directions. For attention computation, direction matters more than magnitude, and angles cluster predictably across tokens, so they compress cheaply onto a small fixed codebook. QJL (stage 2, Quantized Johnson-Lindenstrauss) adds one extra bit per dimension that corrects stage 1’s small directional bias, making the inner-product estimates the model uses for attention scores provably unbiased: mathematically guaranteed not to drift. This is lossless compression of the model’s working memory: on average, the compressed answer matches the uncompressed answer exactly.

Prior compression algorithms required trade-offs that made them unsafe for production retrieval pipelines. TurboQuant is the first algorithm to combine all four properties required for reliable, high-performance deployment at web scale.

Three independent findings from WordLift (Andrea Volpini) establish the empirical case for structured entity pages over prose content. These are WordLift’s findings. Attribute them as such.

• +29.6% AI retrieval accuracy from dedicated entity pages with RDF structured data, over plain HTML and over HTML-plus-schema-without-RDF. (arXiv:2603.10700)

• 71% win rate for graph traversal (RLM-on-KG) over Microsoft GraphRAG on complex, scattered-evidence reasoning, which is the real-world case for most brands. (arXiv:2604.17056)

• 7% of AI citations go to pages outside the classical top-10 SERP, reached via entity links not rankings. Empirical proof that AI citation has already diverged from classical ranking.

Schema presence (has_schema) is also an independent variable in WLDM’s own 11M+ AI citations dataset. WLDM measures schema as a citation signal, not just asserts it.

Four categories of work make a brand computable at computation grade.

Schema skyscrapers

Comprehensive structured-data pages built so the model computes the brand’s entities without parsing prose. Maximum computable density: every entity the brand should be known for, represented in structured markup with stable identifiers, so the model can answer a broad range of queries from a single page traversal.

Article IDs and @id stability

Every schema entity, and every reference to it across the site, resolves to one canonical @id. Character-for-character consistency. Computation-grade requirement, not a best practice.

Wikidata entity engineering

The brand’s Wikidata QID anchors its entity identity across AI systems. Most enterprise QIDs exist but are thin: missing category-level and relationship properties. The work is property-completeness and on-page sameAs anchoring to the correct QID.

Off-page entity stacks

Consistent entity facts across third-party sources, with deliberate wording variation to validate the cluster. Coordinated with the AI Citations service.

Boundary with GEO

GEO delivers the broad on-page foundation: entity graph, @id consistency, crawler compliance, content structure. TurboQuant Optimisation is the specialised tier that pushes entity coverage to computation grade across the full question surface and enforces stability as a hard requirement.

Entity breadth is the computation-grade extension of GEO’s entity inventory. The principle: for every question a buyer asks AI in the category, the model should be able to compute a brand-relevant answer from structured facts without falling back to parsing prose.

Most enterprise brands have strong coverage for head-term queries (brand name, primary product, core use case) and near-zero coverage for the long tail: comparisons, methodology questions, use-case variations, integration queries. That long tail is where much of the computation-era citation concentrates, because those are the questions buyers ask AI before they have a shortlist.

The audit maps the full question surface for the category, scores current entity coverage against it, and identifies the gaps where a competitor or a generic source is being computed instead of the brand. The rebuild prioritises by the queries with the highest buyer-intent signal and the lowest current brand coverage.

Schema skyscrapers are the architecture for breadth at scale: a single structured-data page designed to answer a cluster of related buyer questions from computed entities, so the model never needs to fall back to a 6,000-word article for questions in that cluster.

For a model to compute with confidence, every reference to the same entity must resolve to the same canonical identifier, character for character. “Acme Corp”, “ACME Corp”, and “Acme Corporation” are three separate entities to a model performing entity computation. The correct canonical @id, applied consistently across every instance of the entity in schema markup, internal links, and structured data, is what consolidates three weak signals into one strong one. This is the @id consistency rule from GEO raised from a best practice to a non-negotiable computation requirement.

“Wikipedia is the human version; Wikidata is the data version for LLMs.”

A brand’s Wikidata QID is the anchor point AI systems use to resolve brand references across the web. Most enterprise brands have a QID that resolves the brand name but is missing the category-level properties, relationship properties, and industry context that make it computable for buyer questions. The work is property-completeness: ensuring the QID carries the properties AI systems actively query, and on-page sameAs anchoring so every relevant entity page links back to the correct QID.

The same facts, names, and properties need to appear in third-party sources across the web, with deliberate wording variation to prevent the cluster from collapsing to a single source.

The variation validates the cluster; the consistency validates the entity. As Brie puts it: “We create entity stacks across the internet — consistently we want to change the wording a little bit — ‘Kinsta is an enterprise-level WordPress hosting platform that focuses on XYZ’, then variate it. Once you start clustering those entities together with a little variation, you’re validating.” This work is coordinated with the AI Citations service.

TurboQuant is deployed now and its effect on retrieval behaviour is measurable now. The variable you control is not the compression algorithm, which runs inside the model. The variable you control is the quality, breadth, and consistency of your entity graph.

Full adoption across all AI systems sits on approximately an 18-month horizon. The entity coverage built during that window is what the model reaches for as adoption completes.