The reciprocal-NDA reference model.

The standard objection

A prospect on the first call frequently asks for a public client list. Most B2B services have one. Logos on the home page is a standard trust signal. Labs does not have one. The absence prompts skepticism, and the skepticism is reasonable.

This article explains why the absence is deliberate, what we do instead, and why a prospect insisting on the public list is, in our view, asking the wrong question.

What a public client list signals

For most B2B services, a public client list is a strong trust signal because:

• It commits the vendor to a level of service the named clients implicitly endorse
• It allows prospects to compare their organisation to the vendor's typical clientele
• It demonstrates that the vendor has actual customers

For most B2B services, this works. The downside is small.

For an audit firm, the downside is not small.

Why audit firms publishing clients is structurally wrong

Three reasons the dynamic does not transfer to a cybersecurity audit context:

1. The client list discloses the audit perimeter. If Labs publishes "we audited Company X," an adversary now knows the public surface of Company X has been scrutinised by a firm with a known methodology. The adversary updates: spend less time on the obvious patterns Labs documents, more time on the patterns Labs is known to skip. The audit firm's customer becomes a more attractive target, not less, simply because the audit was disclosed.

2. The client list creates a comparison effect. If a prospect sees "Labs audited Company X" and Company X is similar to the prospect's organisation, the prospect infers "if Company X needed an audit, so do we." This is mostly correct, but the inverse — "if Company Y is not on the list, they were not audited" — is also drawn, and that inference is information leakage about which buyers chose to publicise their audit status. Buyers who chose not to publicise are now visible by negation.

3. The client list anchors the engagement to a brand narrative. When the audit firm uses its client list for marketing, the firm's incentive to find uncomfortable findings in those clients diminishes. The buyer is now also a brand asset. The relationship has a conflict of interest the buyer should be aware of.

For all three reasons, the audit-firm category — in its responsible form — does not publish a client list as a default.

What Labs does instead

The Labs trust model uses three substitutes for the public client list.

1. The reciprocal-NDA reference call. A prospect under reciprocal NDA can request a reference call with a prior client. The prior client must have explicitly consented to the call; we ask each time, not once. The call is direct: the prospect and the prior client speak without Labs on the line. The prior client speaks about their engagement experience in their own words. We do not script the conversation.

Two prior clients have consented to this to date. The contact path is shared during scoping with qualified buyers.

2. The published methodology. The methodology — AI-Alliance Challenging, the evidence ladder, the Proof-of-Threat doctrine — is published in detail on /how-we-work and in the research section. A prospect can evaluate the firm's working method without knowing the firm's clients. The method is the public artefact.

3. The redacted artefact set. Redacted Challenge Logs, redacted finding examples, and redacted deliverable structures are published on /proof. The prospect can read the shape of what the engagement produces, without any identifying detail of the engagement.

The three substitutes, together, produce a stronger trust signal than a logo wall would, for a prospect who reads them carefully. The prospect who does not read them carefully is, in our experience, not a Labs buyer regardless.

What the prospect should ask instead

A prospect evaluating Labs should ask, in approximately this order:

• "Can I read a redacted Challenge Log sample?" (Yes, on /proof.)
• "Can I speak with a prior client under reciprocal NDA?" (Yes, with two prior clients to date who have consented.)
• "What does your contract explicitly prohibit you from doing?" (Probing, testing credentials, accessing infrastructure beyond the public surface, naming clients publicly.)
• "What is the founder's professional standing, and what insurance covers the work?" (Disclosed under reciprocal NDA during scoping.)
• "What happens if I am dissatisfied with the deliverable?" (Defined in the contract: revision pass, refund clause for the Discovery Brief tier, escalation contact.)

These questions produce more useful information than a logo wall ever would. They are the questions a careful procurement function asks of an audit firm.

What the prospect should NOT ask

Three questions that occasionally come up in early conversations and that we politely decline:

• "Can you name one client you audited last year, even off the record?" — No. Off-the-record disclosure to a prospect is still disclosure. The prior client did not consent to it.
• "What CVE numbers did you discover during prior engagements?" — We do not publish CVE work; that is a different category (vulnerability research), and we are not a CVE issuer.
• "Can you share a screenshot of a real deliverable, redacted?" — The redaction work for a single screenshot is asymmetric with the information value to the prospect. The redacted Challenge Log on /proof is the substitute.

The trust-by-default problem

The most-asked underlying question — "why should I trust a small, new audit firm I have not heard of" — has no quick answer. The honest answer:

• The firm is new. You probably should not trust it on faith.
• The firm publishes its methodology. You can audit the methodology before signing.
• The firm has a small number of prior engagements. You can speak to those clients.
• The firm offers a small entry tier (Discovery Brief, 3 days, fixed price). You can test the firm before committing to a Board-Ready engagement.
• The firm signs the work. You have legal recourse if the signed attestation is materially wrong.

This is a slow trust-building model. It is slower than a logo wall. It is also the model that survives scrutiny in regulated contexts, which is the long-term game.

A note on logo walls

Some Labs prospects will, after this conversation, still want a logo wall. They are reasonable to want one; the rest of their procurement vendor-list is full of them. They are not Labs buyers, and we tell them so during the first call.

Labs is for the buyer who recognises that the absence of the logo wall is the signal. The buyer who needs the signal to point in the opposite direction has a different category of vendor to choose from. We are not the right firm for them, and saying so quickly saves both sides time.