Methodology
How to read the Behavioral Responsivity Framework
One page that defines what we claim, how we classify our evidence, and the vocabulary our articles reuse. If a claim in any article points here, this is what it points to.
How we classify evidence
Every claim we publish is tagged by the kind of evidence behind it. The point is to make the boundary visible — so you can always see where established research ends and our own synthesis begins. Very few frameworks in this field draw that line; it is the one we care about most.
Peer-reviewed behavioral science and information-retrieval findings — Joachims, Nisbett & Wilson, Scharkow, Pirolli & Card, Chapelle & Zhang, and equivalents.
Engineering intent documented in granted patents. Evidence of how a system was designed to work — not proof of current deployment.
Real-world implementation: court testimony, engineering disclosures, documented API behavior (e.g. US v. Google, 2023; the Content Warehouse API leak, 2024).
Conclusions derived from the evidence above but not directly documented by any external source — the framework's own predictions, labeled so you can tell them apart.
The Behavioral Signal Hierarchy
Implicit signals sort into four tiers, ordered by how hard each is to fake. This ordering reflects inferential reliability and manipulation resistance — not confirmed ranking weight, which is unknown and varies by query. Articles reference these tiers rather than re-defining them.
Tier 1 — Click
visible · manipulation cost: Low
Which result the user selects. The most visible signal, and the easiest to fake.
Tier 2 — Engagement
attention · manipulation cost: Medium
What happens after the click — dwell, scroll, session-level attention. Harder to fake at scale.
Tier 3 — Reformulation
verdict · manipulation cost: High
Whether the result resolved the need. Requires a full, failed session to manufacture.
Tier 4 — Longitudinal
durability · manipulation cost: Very high
Whether utility persisted after the task — return visits, branded demand. Must produce real recurring value.
↓ Top to bottom, manipulation cost rises — and with it, how far the signal can be trusted.
The full development of the hierarchy — with evidence for each tier — lives in The Behavioral Signal Hierarchy.
Reading your own signals
The catch is structural: the signals a search engine weighs most are the ones you can see least. Your tools expose far less than the engine's. So each tier below pairs a visible proxy with the thing it cannot tell you — and the question worth asking because of that gap. Articles reference this table rather than reprinting it.
🟣 These are diagnostics — questions the hierarchy implies, not a tested playbook with guaranteed outcomes.
Explicit — survey / feedback
- You can see
- Ratings, survey scores, NPS
- It cannot tell you
- Whether stated satisfaction matches behavior — self-report is the distorted channel
- So ask
- Do the people who rate it highly also act as if it helped them?
Tier 1 — Click
- You can see
- Search Console: impressions, clicks, CTR, average position
- It cannot tell you
- Raw CTR is confounded by position, and the engine's bias correction is invisible to you
- So ask
- Is CTR low for this position — a title or snippet problem — or just normal for the rank?
Tier 2 — Engagement
- You can see
- GA4: average engagement time, scroll depth
- It cannot tell you
- High dwell can mean interest or friction; the meaning depends on the page's intent
- So ask
- Is the dwell on this page appropriate to its intent, or a sign of confusion?
Tier 3 — Reformulation
- You can see
- Indirect only: a page that earns clicks but logs very short sessions
- It cannot tell you
- You cannot see the reformulation itself — that record lives with the search engine
- So ask
- When the click is earned but the session is short, was the need actually resolved?
Tier 4 — Longitudinal
- You can see
- GA4: returning-user rate, direct traffic. Search Console: branded-query growth
- It cannot tell you
- These are publisher-side shadows, not the engine's internal longitudinal signal
- So ask
- Does this page — or the brand — earn return visits and branded demand, or only first clicks?
↓ Top to bottom, manipulation cost rises — and with it, how far the signal can be trusted.
What follows from it
- —Prioritize pages and topics by their higher-tier proxies — returning-visitor and branded-search growth — rather than by raw click volume.
- —Before counting a page as a success, confirm the gain shows up in a proxy above Tier 1. A click-only win is the easiest kind to manufacture.
- —When two proxies disagree, trust the one that is harder to fake — the Manipulation-Cost logic, turned on your own dashboard.
- —Treat no single proxy as decision-grade. Act when two or three point the same way.
The Retrieval Loop
The system model the framework's six ontology layers map onto. Every retrieval system that learns from behavior runs some version of this cycle; each layer of the framework owns a stage of it. Articles reference the loop rather than re-defining it.
🟣 The loop is Framework Synthesis — a framework-level synthesis, not a diagram of any vendor's pipeline. It formalizes the functional cycle that behaviorally adaptive retrieval has to negotiate: representation, retrieval, interaction, evaluation, learning. What the framework specifies is the boundaries between those stages and the feedback relations among them. Individual systems may implement their internal models, signals, and update mechanisms differently.
1 — Representation
Ontology 5 + adjacent territory
What is this content about, and in what order does a person meet it? Entity-centric semantic SEO models the machine-readable half of this stage deeply — entities, attributes, semantic structure — and the framework takes those techniques as given. What the framework claims is the layer beneath them: which distinctions the representation has to preserve, and in what sequence a reader should encounter them, derived from the behavioral requirements of the task rather than from the markup vocabulary. That is Behavioral Information Organization.
2 — Retrieval
the pipeline
Candidates are selected and ranked for a query, using representation plus everything the system has learned so far.
3 — Interaction
Ontologies 0 + 1
Humans behave. Why they behave the way they do is the Behavioral Foundations layer; what engines can observe of it — the signal families and tiers — is User Behavior Signal Architecture.
4 — Evaluation
Ontologies 2 + 4
Behavior is read against the intent behind the query: did the response resolve the need? Intent-Response Alignment defines the test; Utility Divergence defines its limits — satisfaction itself stays latent.
5 — Learning
Ontology 3
Validated behavioral evidence — corrected for presentation bias, normalized for context, aggregated over time — updates the models. Adaptive Feedback Systems is this stage in depth. The loop then returns to an improved representation.
↺ Learning returns the loop to Representation — visibility is never granted once; it is continuously re-evaluated.
Where the evidence attaches: 🟠 US 8,938,463 describes learning from bias-corrected clicks; 🟠 US 9,002,867 describes unlearning — discounting accumulated behavior when a document substantially changes; 🔵 US v. Google testimony (2023) describes calibrating the learning to how much behavioral evidence a query actually has.
The Translation Chain
Behavioral research does not tell you where to put a specification table. Between a finding about human behavior and a decision about a page there is a translation step, and that step is where this framework does its own work. The chain below is the procedure — the same one applied in Field Report 001.
🟣 The chain is Framework Synthesis — a method for deriving design decisions, not a claim that any retrieval system reasons this way. Each link is auditable: you can ask which finding produced a given decision, and we should be able to answer.
1 — Behavioral grounding
Established findings about how people seek, evaluate, and use information, together with the task or intent that motivates the information-seeking episode.
2 — Information requirement
What that finding implies the reader must be able to obtain. Not yet a page: a statement of what has to be knowable, distinguishable, or comparable before the reader can act.
3 — Organization decision
How the information is grouped, sequenced, and connected to satisfy the requirement — which concepts belong together, which relationships matter, what has to be resolved before what.
4 — Representation
The same decision expressed in the forms a system can read: page structure, entity relationships, schema, internal links, headings, navigation. Several representations of one model, which is why they have to agree.
5 — Prediction
What should follow if the model is right — stated before measuring, in terms specific enough to be wrong.
6 — Measurement
The prediction tested under a fixed protocol, with the misses recorded alongside the hits, and the alternative explanations named.
↓ The direction is behavior → design → measurement. Never search engine → design.
The chain is what makes the framework accountable rather than decorative. A decision that cannot be traced back to a behavioral requirement is a preference, and we try to label it as one.
The objections we welcome
Sophisticated readers push back — and they should. We answer in a consistent register: concede the literal point, then sharpen to the actual claim.
“Patents don’t prove implementation.”
Correct — and we don’t use them as proof of deployment. We use patents as evidence of engineering intent and architectural possibility. The behavioral science doesn’t depend on any patent; the patents illustrate it.
“Correlation doesn’t imply causation.”
Agreed. We don’t claim behavioral signals cause rankings. We claim retrieval systems face a measurement problem — satisfaction is latent — and behavioral signals are the least-bad available proxies. Whether a given system uses a given signal causally is a separate question from whether that signal carries information about utility.
Google is the evidence, not the subject
This framework is not an attempt to reverse-engineer Google. It is an account of the behavioral constraints any retrieval system faces when it tries to infer utility from how people behave. Google, Microsoft, and other search engines are valuable case studies only because they operate at enough scale to expose those constraints publicly — through patents, testimony, and engineering disclosures.
The behavioral principles come first. The implementations follow from them. A patent can be retracted and a system renamed without touching the underlying claim: that satisfaction is latent, and behavior is the least-bad proxy a ranking system has for it.
That is the whole apparatus: a clear claim, four classes of evidence, three models the articles reuse — the hierarchy, the loop, and the translation chain — and a vocabulary that compounds across the cluster. The articles apply it.
Read the articles →