Skip to content
Saleswoman

Cold email software that refuses to guess about your prospect

Volume is the last decision in cold email, not the first. Four things have to be right before the number is even worth discussing, and three of them can be checked automatically on the draft.

Run it on a prospect
  1. Every claim about them is true, and you can show why

    A cold email exists to say one specific true thing about somebody who did not ask to hear from you. Get that sentence wrong and nothing else in the message matters, because the reader now knows you were guessing. This is also the failure that scales worst: a person guesses about one account, a bot guesses about four thousand.

    How the software handles it

    Enforced by the claim ledger. Every factual sentence is cited to a source you connected or marked unverified, and an unresolved unverified claim holds the send.

  2. The message is lawful to send

    Not a formality, and not one rule. In the United States a commercial message needs honest headers, a subject that matches the content, a physical postal address and a working opt out, and penalties are assessed per email rather than per campaign. Elsewhere the bar is higher: Germany treats unsolicited commercial email as a nuisance under UWG section 7, and Canada requires consent or a documented relationship under CASL.

    How the software handles it

    Enforced by three pass or fail checks on every draft: unsubscribe present, address block present, subject line not deceptive. Each one holds the send on its own.

  3. It reads like something a person would send

    Not as a matter of taste. Trigger phrases, an image only body, three links, a wall of text at graduate reading level and a run of exclamation marks are the specific things filters weight, and they are also the specific things that make a reader reach for the spam button rather than the delete key.

    How the software handles it

    Enforced by the spam score, which names every trigger it found rather than returning a number you cannot argue with. Reading grade and word count are reported next to their target ranges.

  4. You are entitled to contact this person at all

    This is the one no software can do for you. Whether you have a lawful basis for this contact, in this jurisdiction, on this list, is a judgement a person makes and records. A tool can store the answer, enforce a suppression list and refuse to write to a suppressed address. It cannot make the decision, and any vendor telling you the AI handles compliance is describing a product that does not exist.

    How the software handles it

    Not enforced. Recorded. The lawful basis is a field on the contact, it travels with the contact, and it is in the export when somebody asks.

Only then, the volume

Once those four are settled, the number is a straightforward operational decision, and it should be made from your complaint rate rather than from your pipeline gap.

Google's sender guidance names the level to watch: keep the spam rate reported in Postmaster Tools below 0.30%, and aim to stay under 0.10%. Those requirements, along with SPF, DKIM, DMARC and one click unsubscribe, have applied to bulk senders since February 2024. Crossing the line is not a fine. Delivery degrades for everything from that domain.

email outreach platform, for how the cap is enforced.

A first touch that clears all three mechanical checks, queued rather than sent.

PASS

Spam risk 14/100 Grade 7.0 79 words 1 of 40 today

Cleared every check. Both factual sentences cite a source you connected.

Claim ledger

2 of 2 cited

  • You opened a second facility in Reno in March. CITED

    Source: their press page, read at draft time.

  • Returns are still reconciled across two systems. CITED

    Source: CRM note from the call on 2 March.

Spam triggers found: none. One link. Unsubscribe and address block present. Subject line describes the content.

The ten line checklist, and who does each line

Nine of the ten are mechanical and run on every draft before it can queue. The tenth is yours, permanently.

The from name and reply address are real and monitored

Software

The subject describes what is inside

Software

The commercial purpose is obvious in the first two lines

Software

A valid physical postal address is present

Software

There is an opt out that needs nothing but an email address

Software

Opt outs are suppressed everywhere within hours

Software

Every factual claim cites a source

Software

SPF, DKIM and DMARC pass on the sending domain

Software, when the mailbox is connected

Complaint rate stays under the published thresholds

Software watches it, you set the cap

You are entitled to contact this person

You, and it is recorded per contact

The full version of that list, with what each requirement actually means for a cold email, is in can spam compliance.

The whole product

What it is, why it holds a draft, and the demo it all runs in.

AI SDR

What it costs

Flat monthly, never per seat, and nothing is charged on this site.

AI SDR software pricing

The guide on this

Seven requirements, what each one means for a cold email specifically, and the two that outbound teams get wrong most often.

can spam compliance

The rest of the product, by job

The same product, entered from the job you came here to do. Each page states its own limits rather than repeating the pitch.

  • AI email writer

    The drafting side on its own page: what it writes, and what it refuses to assert without a source.

  • Spam score checker

    What the score is made of, line by line, and the two things it deliberately does not test.

  • Email outreach platform

    The cap, the queue and the suppression list, which is the part of a platform nobody demos.

  • CRM automation

    Activity logging, deal notes and next steps, written from the thread instead of from memory.

  • Prospecting tools

    Where a research tool ends and a list vendor begins, and which one this is.

  • Sales automation tools

    What to automate, what to review, and what a person still has to decide.

Run your current template through it.

Paste an email you already send. The interesting result is not the score, it is the list of sentences it cannot find a source for.

Run it on a prospect

Your data stays yours. Prospect data is never used to train a model. Nothing is charged today.