Antitrust and Fair Benchmarking

Benchmarking that is fair by design.

A useful peer benchmark and a lawful one are the same thing when the benchmark is built correctly. Verinode turns the data already sitting in an operator’s inbox and tools into an honest picture of where they stand, without ever becoming a channel through which competitors trade sensitive figures or move together. This page explains why that is lawful and, more usefully, how Verinode is deliberately designed to keep it that way.

Last updated: July 2026

Design and governance, not legal advice

This is a description of how Verinode’s benchmarking is designed and governed. It is not a legal opinion, and it is not a guarantee. Several of the controls described here are being hardened in code as the network matures. Where that is the case, the text describes the committed design and the standing commitment rather than a finished feature.

Why an independent benchmark is lawful

Benchmarking is a normal, established, and lawful practice. Industries run peer surveys and performance benchmarks all the time, because knowing how your own operation compares to your peers is how any business improves. Competition law does not prohibit that. What it scrutinizes is a narrower set of patterns: competitors sharing current, identifiable, competitively sensitive figures with each other, or using a shared facility to coordinate what they charge, pay, or accept.

The safeguards that keep a benchmark on the right side of that line are well understood. A neutral administrator collects the data instead of competitors trading it directly. The output is aggregated and anonymized, so no participant’s own figures are exposed. Sensitive numbers are aged rather than shown live. And the facility informs independent decisions rather than steering everyone toward the same one.

Verinode is built around exactly those safeguards. The seven principles below are how they are put into practice, and how the product is designed so the risky patterns are not merely discouraged but structurally absent.

Principle 1

A neutral third party sits in the middle

Verinode is the administrator of the benchmark, not a party to it. Members contribute their own operational data to Verinode. Members never see each other's raw data, and no member can read another member's numbers. The only thing that ever comes back out is a synthesized aggregate: a peer median, a middle range, a distribution.

This is the structural difference between a lawful benchmark and an unlawful exchange of information. Competitors do not share figures with each other here. Each operator shares with an independent trust, and receives back a picture of the whole that no single contributor can reverse-engineer into anyone else's books.

In practice

  • No operator can query, read, or export another operator's raw or identifiable records. The public API serves an operator only their own data, never the shared intelligence layer.
  • Identifying fields, such as names, addresses, claim numbers, and contact details, are column-encrypted per operator and kept separate from the analytical dimensions used to build aggregates.

Principle 2

Aggregated, anonymized, and never below a participation floor

No statistic is published unless enough distinct operators have contributed to it. Below that floor, the number is simply not shown. This keeps any single operator from being read back out of an aggregate, and it keeps any single operator from dominating a figure. Sensitive financial categories, such as margins, wages, and labor cost, carry a higher floor than operational ones, because the risk of working a number back to one business is greatest there.

Operator identity enters the network only as a one-way cryptographic hash, with no path from a published benchmark back to a contributor. Each operator contributes one vote no matter how many records they submit, so volume cannot buy influence. Benchmarks report the median and the middle range rather than the average, so one extreme value cannot drag the number.

The exact participation floors and the aggregation rules are published in the Methodology and the Data Use Policy. This page describes why they exist; those pages state how they are set.

In practice

  • A benchmark is published only above a distinct-operator floor: a published minimum of 5 operators at national, state, or group scope, and 10 for sensitive categories such as financials, labor cost, and margins.
  • One operator, one vote, no matter how many records they submit, so volume cannot buy influence.
  • Benchmarks report the median and the middle (interquartile) range, not the average, so a single extreme value cannot move the number.
  • Operator identity enters the network only as a one-way, salted hash, and running aggregates never store or expose an individual row.
  • A new operator's data joins peer cohorts only after a short maturation window, and demo or test data is held in fully separate cohorts.
  • Only figures machine-extracted from real documents and connected systems, or bounded attested ratings, can move a benchmark; hand-keyed numbers are kept for the operator's own analysis and excluded from peer cohorts.

Principle 3

Sensitive figures are released on a historical lag, not in real time

Verinode's design draws a hard line between two kinds of number. Operational metrics, such as cycle time and days-to-pay, are shown current, because timeliness is what makes them useful and they carry little competitive sensitivity. Competitively sensitive figures, such as prices, costs, margins, and wages, are treated differently: they are never shown in real time, and are published only after a 90-day historical lag.

The reason is deliberate. A benchmark that broadcast current pricing or current wages across a market could look like a signal to move together. A benchmark built on figures that have already aged cannot serve that function. It tells an operator where they have stood; it cannot coordinate where a market is about to go.

In practice

  • Operational, low-sensitivity metrics such as cycle time and days-to-pay are shown current, because timeliness is what makes them useful.
  • Competitively sensitive figures, such as prices, costs, margins, and wages, carry the higher participation floor and are published only after a 90-day historical lag, never in real time.
  • The two tiers are handled separately end to end, so timeliness where it helps never comes at the cost of exposing live competitive figures.

Principle 4

Verinode never recommends, defaults, or scripts a price or a wage

Verinode shows an operator where they stand. It never tells them what to charge or what to pay. There is no recommended price, no suggested wage, no default rate, no script that nudges every operator toward the same number. A benchmark is a mirror, not an instruction.

This holds across the whole product, including the AI Co-COO. IQ is advisory by design: it surfaces analysis, context, and where an operator sits against peers, and the operator decides. It does not set competitive terms, and it is built so that it cannot quietly converge operators onto a common one.

In practice

  • IQ and every specialist agent are advisory: they recommend, you decide. Every recommendation requires your acceptance and can be edited or rejected before it has any effect.
  • There is no recommended price, suggested wage, default rate, or script that would nudge operators toward a common number.
  • IQ takes no outward-facing or hard-to-reverse action, such as sending a message, moving money, or changing another system, without your explicit, specific confirmation.
  • Numeric claims are traced back to their source, and the agents are tested against adversarial prompts designed to coax a fabricated benchmark or a dropped confidence caveat.

Principle 5

No common position toward any carrier, TPA, or vendor

Verinode never aggregates operators into a shared position toward any carrier, TPA, or vendor, and offers no feature that would coordinate what members charge, pay, or accept. Members act independently. Verinode helps each one understand their own situation against the wider picture; it does not organize them to act as one.

This is a line the product is built not to cross. Showing an operator how their own outcomes compare to a peer distribution is benchmarking. Coordinating operators to move together would be something else entirely, and Verinode is designed so that it is neither offered nor possible.

In practice

  • Verinode ships no feature that would aggregate operators into a shared position toward any carrier, TPA, or vendor, or coordinate what they charge, pay, or accept.
  • The Terms of Service bar operators from using benchmarks or platform output to coordinate prices, wages, or terms with competitors, or as a facility for any agreement or concerted practice that restricts competition.
  • Benchmarks are provided to inform each operator's own independent decisions; nothing in the product organizes operators to act together.

Principle 6

Different corporate structures are handled differently, with consent throughout

Not every comparison carries the same considerations. A single company comparing the performance of its own branches is not the same as independent competitors comparing notes. Verinode is built to treat these situations differently rather than flatten them into one, so that a within-company view and a cross-operator benchmark are governed by rules appropriate to each.

Consent runs through all of it. Contributing to peer benchmarks is disclosed and agreed at sign-up, described plainly in the Data Use Policy, and an operator can export or delete their data at any time. Nothing about how a benchmark is built is hidden behind a setting an operator has to go find.

In practice

  • A single company comparing its own branches and independent operators comparing across businesses are governed by rules appropriate to each, rather than flattened into one.
  • Contribution to peer benchmarks is disclosed and agreed at sign-up, described plainly in the Data Use Policy, and revocable: an operator can export or delete their data at any time.

Principle 7

Governed independently, and enforced in code

These are not just words in a policy. The rules that build every number are implemented in the system itself: the participation floor, the one-vote-per-operator rule, and the anonymization are enforced by the code, not left to intention. The separation of sensitive figures from real-time exposure is being brought under that same enforcement as the historical-lag controls roll out.

Governance sits on top of that. The Operator Advisory Council, active operators with no carrier or TPA in the room, reviews the Data Use Policy and the methodology, and publishes its concerns. The Council advises; it does not govern the company. Several of the controls named here are being hardened as the network grows, and where a control is still being rolled out the commitment stands and remains under this review, alongside outside counsel.

Why the design promotes competition

Competition law does not only ask whether a benchmark avoids the risky patterns. It also weighs whether an arrangement, on balance, helps a market work better. Verinode’s position is that its design is affirmatively pro-competitive, not merely inside the lines.

It widens access to information. Independent operators have historically lacked the aggregate market view that larger, data-rich participants already hold. Verinode gives an operator an anonymized read of how their own peer group performs, so a smaller operator can make the same well-informed independent decisions a larger one already can. Broadening access to market information tends to strengthen competition, because better-informed independent choices are the opposite of coordinated ones.

It is a benchmark, not a shared pricing engine. Enforcers have challenged arrangements where competing businesses fed non-public data into a single algorithm that then recommended or aligned their prices. Verinode is the opposite arrangement: it runs no common pricing algorithm across operators, it never sets, recommends, or optimizes any operator’s prices, and each operator decides independently. Using software changes nothing here; what matters is that no shared engine is harmonizing anyone’s behavior.

It looks backward, not forward. The patterns that draw the most scrutiny involve competitors sharing forward-looking information: future prices, upcoming marketing plans, or how they will divide customers or territories. Verinode shares none of that. It works from historical execution records, releases sensitive figures only after a 90-day lag, and never touches future plans, customer allocation, or territory division.

Projecting the data forward does not cross the line, because the line is on the output.A fair objection is that historical data can still be run forward by a model. The safeguard is therefore drawn at what the AI produces, not only at the age of what it reads. The AI Co-COO forecasts an operator’s own demand, cash flow, and margin trajectory, but it never forecasts, recommends, or optimizes a price or a wage, forward-looking or otherwise. Each operator’s analysis is private and scoped to their own business, so there is no common recommendation for operators to converge on; the forecasts are built from that operator’s own history and public signals such as weather and claim-frequency data, not from any competitor’s live figures; and a person makes every decision, with nothing in the product that could monitor or enforce what any operator actually charges. Coordination needs a shared signal and a way to police it. Verinode provides neither.

It is oriented to execution, not to consumer pricing.Verinode’s benchmarks focus on operational performance and how the network is paid on standard, third-party-defined line items, not on what an operator should charge a homeowner or a local customer. The comparison runs along an operator’s own execution and their dealings on the claim, not along the horizontal axis of what competitors charge the same buyers.

Contact

This describes committed design and governance, not a legal opinion or a guarantee. It is reviewed alongside the Data Use Policy and the methodology, and updated as the controls it describes are hardened.