A serious operator does not buy on a pitch, and a serious investor does not invest on one. Neither would we.
An independent data trust only earns the name if it holds up to scrutiny. So here are the questions a careful operator or investor should ask, answered plainly, including the ones we cannot fully answer yet. Where we are early, we say so.
Common Questions
We are building the first independent operator cohort in restoration, and we are early about it. That is exactly why a benchmark never appears until its cohort clears a minimum size: if the sample is too thin to be honest, or to protect anyone's anonymity, the number is withheld, not shown. You will never see a peer line built on two operators dressed up as a hundred. As the cohort grows, more benchmarks unlock, each gated on real coverage. And quality matters more than raw size here: every contributor is a working restoration operator sharing their own real operational data, never a scraped database, a bought list, or a survey panel, and never an unrelated trade. Segment-specific work within restoration, like exterior and roofing claims, gets its own scoped cohort rather than being folded into a generic average. A benchmark from a few hundred verified restorers is worth more than one averaged out of a big, mixed-trade dataset.
Benchmarks are cohort-filtered, so you are compared to operators like you, by size, service mix, and region, not the whole industry averaged into mush. Where a cohort is too small to be representative or to stay anonymous, it is not published. Representativeness is enforced by a floor, not assumed. Because contributors are restoration operators only, the population is the right one to begin with, not a general-contractor or multi-trade average stretched to look bigger than it is.
Every decision Verinode surfaces carries the dollar impact and the data behind it. You can trace a recommendation back to the specific inputs it read, the jobs, the invoices, the peer line, because every extraction ships with a receipt of what was read and what was corrected. Nothing is a black box you are asked to trust.
Little. There is nothing to migrate and no system to replace. Verinode reads from the documents, exports, and inbox you already have. You point it at your data once, and it does the structuring.
It stays yours. You can export it, and you are not locked in, because Verinode reads what your existing tools already produce rather than trapping your history in a new silo. Leaving is always an option, which is the only thing that makes a data promise worth anything.
An Operator Advisory Council, active operators only, with no carrier or TPA in the room, reviews every material change to the data-use policy before it ships and publishes its dissent if it disagrees. The commitment that your data is never given to carriers, competitors, or any other third party in identifiable form is written into the Terms, not just stated in marketing, and the policy and methodology are published in full.
No, and we will not imply otherwise. Neither certification is held today, and an external penetration test has not been completed either. A SOC 2 Type II audit and the penetration test are planned together. An attestation is something a third party grants, not a badge a company awards itself, so until an auditor has signed one we say so on our own security page rather than borrowing the language. What does exist is the whole control program, written out control by control and mapped to the SOC 2 Trust Services Criteria and ISO 27001 Annex A, alongside a plainly labeled list of what we have not done yet. So the accurate reading is both halves at once: our controls are self-reported rather than externally attested, and we document them and their gaps in more detail than a lot of certified vendors bother to. Read it at verinode.ai/security and hold us to it.
Every third party that processes anything on our behalf is named at verinode.ai/subprocessors, with what it is for, the specific data it sees, where it runs, and its contractual safeguards. Operator information is processed in the United States, with regional deployment including the EU available on enterprise agreements. Encrypted backups sit in Canada under a 90-day window and automatic deletion, and they carry ciphertext only, so the backup provider holds no key and cannot read them. The AI providers are named rather than hidden behind a general assurance: Anthropic and OpenAI work under paid API terms that bar training on our data, though prompts do sit in their standard 30-day abuse-monitoring window before deletion, so we do not call our posture zero-retention. Direct identifiers are replaced with placeholders on the text paths, meaning forwarded email bodies, pasted text, and voice transcripts; PDFs are deliberately exempt and go to the model as rendered pages, because masking a scan before it is read destroys extraction on invoices and estimates. We would rather name that exemption than let the masking sound universal. Where a Data Processing Agreement is still being executed, that page says so rather than leaving the vendor off the list, because a contract you can see is one you can ask about. A DPA for your own agreement is available from [email protected].
Yes. Benchmarking is a normal, lawful practice. What antitrust law scrutinizes is a narrower set of patterns: competitors exchanging current, identifiable, sensitive figures with each other, or using a shared facility to coordinate what they charge or pay. Verinode is built to avoid both. It is a neutral third-party administrator, so operators never see each other's raw data, and only anonymized aggregates are published, never below a minimum number of distinct operators. Competitively sensitive figures like prices, costs, margins, and wages are published only after a 90-day historical lag, not in real time, while operational metrics like cycle time stay current. Verinode never recommends, defaults, or scripts a price or a wage, and it never pools operators into a common position toward any carrier, TPA, or vendor. The Terms of Service also bar operators from using the benchmarks to coordinate terms with competitors. The full principles and controls are published at verinode.ai/fair-benchmarking and in the Transparency Center.
On the inputs that matter, a second model checks the first, so one model's mistake does not quietly become your number. Extractions are normalized against a known schema and carry a receipt, and the pipeline is built to fail loud rather than guess. The AI drafts decisions for you to approve; it does not act unseen.
Only eligible, provenanced contributions count toward a benchmark. Data passes an anonymization boundary and a provenance gate before it can influence a peer line, and every cohort enforces a minimum size, so no single operator can move or read a benchmark by stuffing it. Integrity is a gate, not an honor system.
We will not pretend to have years of proof we do not have. Verinode is early, and validating decision quality on real, messy operator data is the work of this first cohort. What we can show today is honest: every recommendation comes with the recoverable dollars and the math behind it, so you judge the return on your own numbers, not on a case study we cherry-picked. We also don't stop checking once you act on a recommendation: outcomes are tracked after submission, so a decision that looked right on paper but didn't hold up in practice feeds back into how confidently we surface it next time. The claim is the method, transparent and auditable, not a figure we ask you to take on faith.
Yes, in a conversation, and no, not as a published case study. We are working with our first design partners, so anything we posted publicly today would be a sample of one dressed up as evidence, and you would be right not to credit it. What we will do is put you in touch with operators close to your size and service mix so you can ask them directly, without us in the room. That is worth more than a testimonial page anyway. Ask, and we will arrange it.
Ask for all of it, and start scoped. The Data Processing Agreement and the membership agreement, from [email protected]. A reference conversation with operators like you. And a walkthrough of a real decision on your own data, where you make us show every input behind the dollar figure. That last one is the test that matters, because it is the one a polished pitch cannot fake. Then keep the first pass narrow. Verinode is read-only against systems you already own, writes nothing back, needs no migration, and locks you into nothing, so a scoped start costs you almost nothing to walk away from. Contributor membership is free and every account opens with a 14-day Executive trial with no card, which means you can judge the output on your own numbers before any money or any commitment is involved.
What We Won’t Pretend
We are early. We will not quote an operator count we have not earned or an ROI we have not proven. What we will do is show our work: every benchmark gated on real coverage, every recommendation traceable to its inputs, every promise written into the Terms and held to operator scrutiny through the Advisory Council. The integrity is the product. If we ever have to choose between an impressive claim and an honest one, this page is our answer to which wins.
Is It For You?
The lowest-risk way to find out is to see it on your own numbers. There is nothing to migrate and no lock-in. Verinode reads from the documents and inbox you already have, so you can judge it on your data before you commit to anything.
The Bar
That only works if the answers hold up. Read the policy, check the methodology, ask the Council. Then decide.