How To Vet A Cold Email Agency In 2026

A cold email agency in 2026 sells deliverability engineering with copywriting attached, and the ones that still sell it the other way round are the ones you can safely pass on. The tell is not the pitch deck. It is which rules they quote at you, and whether those rules govern the inboxes you are actually trying to reach.

Two sets of mailbox provider requirements get quoted in almost every outbound sales call. Google's bulk sender requirements, live since February 1, 2024. Microsoft's high volume sender requirements, enforced since May 5, 2025. Both are real, both are published, and both are worth reading. Neither one governs the inbox your buyer opens on a Tuesday morning.

That is not a technicality. It is the central fact about B2B outbound, and an agency that has not internalised it will spend your retainer optimising for a rulebook that does not apply to your list.

What Does A Cold Email Agency Actually Do?

A cold email agency owns four things: sending infrastructure, list accuracy, sequence content, and the measurement that tells you which of the three is failing. Agencies that own only the third one are copywriters with a sending tool, and they price like infrastructure specialists.

Infrastructure is domains, mailboxes, authentication records, warmup schedules, and the rotation policy that decides which mailbox sends to whom. List accuracy is verification, suppression, and the discipline to delete a contact rather than send to it on a guess. Content is the sequence. Measurement is the part almost nobody does honestly, because the numbers that matter are the ones that make the agency look replaceable.

Which Published Rules Govern B2B Cold Email?

Two published rulebooks govern bulk sending to consumer mailboxes, and both set their floor at more than 5,000 messages per day. Google's requirements are the more detailed of the two, and they are worth quoting precisely because the paraphrases in circulation are consistently wrong in the same direction: stricter than the text.

Per Google's sender guidelines, senders above the threshold need SPF and DKIM for the sending domain, a DMARC record whose enforcement policy may be set to none, and alignment between the From header domain and either the SPF domain or the DKIM domain. DKIM keys must be at least 1024 bits, with 2048 recommended. Marketing and subscribed messages must support one click unsubscribe through the List-Unsubscribe-Post and List-Unsubscribe headers, and carry a visible unsubscribe link in the body. PTR records, TLS and RFC 5322 formatting round out the list.

Microsoft's requirements for high volume senders cover domains sending more than 5,000 messages a day to outlook.com, hotmail.com and live.com. SPF and DKIM must pass, DMARC must be present at p=none or stricter with alignment to one of them. Non compliant mail is rejected outright with 550 5.7.15 Access denied, after an initial transition period in which it was routed to junk instead.

Read the bottom two rows. They are where your prospects live, they have no volume floor, and they return no code you can instrument against. There is nothing to comply with, which is exactly why the top two rows get quoted so much: they are the only part of the problem with a checklist.

Why The Consumer Rules Do Not Cover Your Prospects

Both published rulebooks are scoped to consumer mailbox domains, so a B2B program sending to corporate tenants can satisfy every requirement in both documents and still land in quarantine on every send. Compliance with the published rules is necessary and nowhere close to sufficient.

The mechanics differ in a way that matters. A consumer mailbox provider publishes thresholds, operates a feedback loop, and tells you through Postmaster Tools roughly how you are doing. A corporate tenant does none of that. Its filtering posture is set by an IT administrator who tuned it for phishing, not for your sequence, and whose quarantine digest is the only place your message appears. You get no bounce, no complaint rate, and no signal beyond a reply rate that quietly goes flat.

This is the same measurement gap we walked through when the industry's cold email benchmarks turned out to be citing each other rather than any underlying dataset. When a channel produces no trustworthy denominator, published averages become a way of not looking at your own numbers.

How Many Complaints Does It Take To Break A Domain?

Google's 0.30 percent complaint ceiling sounds generous until it is converted into absolute complaints at realistic B2B volumes, at which point it becomes a handful of annoyed recipients per day. The arithmetic below is illustrative, applied to Google's published threshold, and it is the single most clarifying calculation in outbound.

At the volumes a mid market B2B program actually runs, the ceiling is two or three people clicking the spam button. Google's recommended target of below 0.10 percent cuts that to one. No sequence survives that margin by being clever; it survives by being sent to people who plausibly want it, which is a list problem wearing a copy problem's clothing.

“Every deliverability crisis I have been called into was diagnosed as a copy problem and turned out to be a list problem with a copy symptom.”

The related failure is bounce handling, where the distinction between a permanent and a temporary rejection decides whether a retry is diligence or self harm. We covered that split in detail in what hard and soft bounces actually tell you, and any agency that retries hard bounces is telling you something about its tooling.

What Should You Ask A Cold Email Agency Before Signing?

Four questions separate a cold email agency that operates infrastructure from one that resells it, and all four have answers that are either specific or evasive, with very little room in between. Ask them in this order, because each one makes the next harder to deflect.

The third row is the one that ends most sales calls. Open rates on cold email are measured with tracking pixels that corporate security gateways pre fetch, strip, or block, so an open rate on a B2B list is a number about mail gateways rather than about humans. An agency presenting open rate as evidence of deliverability either does not know that or is counting on you not knowing it.

The fourth row matters because domain degradation is routine, not exceptional. A rotation policy that exists in writing before the first send is cheap; one invented during an incident costs you the quarter. We laid out the mechanics of that in how warmup, blocklists and domain rotation interact.

How Do You Tell An Operator From A Reseller?

An operator can show you the configuration, name the failure it last recovered from, and tell you which part of your list it refuses to send to. A reseller talks about volume, pipeline and copy frameworks, and gets quieter as the questions get more technical.

Neither model is dishonest by definition. Reselling a well run mailbox estate is a legitimate business, and a specialist copy team is genuinely valuable once infrastructure is solved. The problem is pricing one as the other, and the only defence is asking which you are buying before the contract rather than during the first bad month.

What Should The First 90 Days Produce?

A competent engagement produces documented infrastructure, a defensible list, and one honest number by day 90, in that order. An engagement that produces meetings in week two and no documentation by month three has borrowed against your domain reputation to buy early optimism.

Expect the first month to look slow. Authentication gets fixed, domains get warmed, the list gets cut, and send volume stays deliberately low while the reputation record is established. Agencies that skip this to show early pipeline are spending an asset you cannot easily rebuild, which is the same trade that makes the corporate promotions folder a one way door for a domain that lands there repeatedly.

Expect the second and third months to produce segmentation. Reply rate by recipient domain family, suppression counts, and a stated position on which parts of the list are not worth sending to. Our own cold email lead generation engagements open with that cut rather than with copy, and the volume floor usually falls before it rises. For B2B programs specifically, the cut is where the deliverability result actually comes from, because a list you are proud of is the only complaint rate control that scales.

See where you are cited today

A free snapshot audit of your rankings and AI citations before we ever talk.