Selling Without Inflating: The Evidence Discipline Behind Our Cold Outreach

Most sales teams say they follow one rule: use only real numbers. Every responsible company follows another: never expose confidential client data. In cold outreach, these two rules collide — and the collision is where credibility quietly dies.

Here is the collision, stated plainly. Your most impressive real numbers are usually your clients’ numbers. The more real and the more impressive a figure is, the more likely it is confidential. So “use real numbers” and “protect client data” pull in opposite directions the moment you try to prove, to a stranger, that you are good at something.

We hit this building our own outbound. What follows is the machinery we put in place so the tension resolves the same way every time — instead of being re-litigated, and leaked, one email at a time.

Part 1 — Three tiers of evidence

We sort every piece of proof into one of three tiers before it can go near an outgoing message.

• Tier A — publishable. Our own operating data, publicly reproducible results, the methodology itself, and synthetic or self-data demos. This is the default, and for cold outreach it is the only tier we send.
• Tier B — not publishable by default. A client’s real data, even anonymized or aggregated. This requires that client’s written consent before it can leave the building. “We anonymized it” is not consent; it is a decision to hope nobody reconstructs it.
• Tier C — never. Anything that identifies a client: names, part numbers, prices, inventory, deal terms.

Two things make the tiers work. First, the default is Tier A only. If a draft reaches for Tier B, that is an escalation to a human owner, not a judgment call for whoever happens to be writing the email under a deadline. Second, “anonymized” is not a promotion. Anonymized client data is its own do-not-publish category; the word does not launder a Tier B fact into a Tier A one. We wrote that down explicitly, because the failure mode is always the same — someone reasons “it’s anonymized, so it’s fine,” and the boundary dissolves in a single well-meaning sentence.

The uncomfortable consequence: our best-sounding material is often exactly what we cannot send. We send the tier we own instead.

Part 2 — Eight rules that turn “don’t exaggerate” into something checkable

“Don’t exaggerate” is unfollowable as written. It is a feeling, and feelings drift under a quota. So we turned it into eight yes/no checks. Every number in an outgoing message has to pass all eight, or it gets cut:

1. Traceable — it comes from a specific event we can point to.
2. Has a denominator — a count is stated against what it is a count of.
3. Has a time period — the “when” is stated or bounded.
4. No max-as-representative — we never present a best case as a typical one.
5. Declared selection bias — if we are only counting the wins, we say so.
6. No causation from correlation — a thing that preceded a good outcome is not claimed to have caused it.
7. No favorable rounding — approximations round honestly, not in our direction.
8. Estimates marked as estimates — anything not directly measured says so.

When a figure fails one of these, we do not soften it. We remove the number and describe the mechanism in words. A described mechanism the reader can reproduce beats a number they have to take on faith.

Part 3 — Why this is a reply-rate problem, not a virtue

The instinct is to treat honesty as a tax on persuasion — the right thing, paid for in conversion. In cold outreach it is the opposite.

We have been on the receiving end of a pitch built around an impressive-sounding rate with no denominator and no time window: a chart, and nothing behind it. It did not make us want to buy. It made us trust every other number in that deck less. One unbacked figure is not a free boost; it is a discount applied to everything near it.

Under-claiming is what makes the claims you keep believable. A sender who marks their own limits is a sender whose remaining numbers you can act on. So the discipline is not ethics-first. It is the deliverability of belief — and belief is the only thing a cold email is actually trying to earn.

What we can’t tell you yet

This piece is about not inflating, so it would be dishonest to end it with a reply-rate lift we do not have.

We designed this discipline. We have not yet run enough outreach to report an outcome number, and we are not going to invent one. When we have the data — with a denominator and a time window — we will publish it, held to the same eight rules above. Until then, the claim is deliberately narrow, and true: our outreach is built so that its credibility survives contact with a skeptical reader.

That is a smaller promise than most sales content makes. It is also one we can keep.