Point of view

01

Roughly 96% of cold email ignores the one thing that doubles replies.

Only about 3.8% of cold emails open on a genuine buying signal, something that actually happened at the company in the last few days. Those emails reply at roughly twice the rate of everything else.

The gap has stayed open because signal-led sending is operationally annoying. It means monitoring sources daily, acting inside 48 hours, and accepting a smaller list. Volume tooling makes the opposite easy.

This gap is the entire reason Miles exists in the shape it does. The daily radar is not a feature, it is the cost of being in the 3.8%.

Source: Unify, analysis of 25M+ outbound emails, 350+ customers

02

Personalizing on job title is worse than not personalizing at all.

Merging in a role performs about 26% worse than sending the same email with no personalization. It is the clearest negative result in the data, and almost every template on the market does it.

The mechanism is simple. "As a VP of Engineering, you probably..." tells the reader you know nothing except what their title says, and it announces the mail merge in the first line. Personalization is only worth the tokens when the variable is something the reader knows is hard to find.

One real variable beats two synthetic ones. Tier A prospects get a researched line about something that actually happened. Everyone else gets a clean segment-level email with no fake intimacy.

Source: Unify, analysis of 25M+ outbound emails

03

Deep personalization is a losing trade below a 5% reply rate.

Research-heavy sending lifts replies substantially, but it costs real time per prospect. Below roughly a 5% expected reply rate, the arithmetic inverts: you spend more on research per booked meeting than the meeting is worth.

This is why "personalize everything" is bad advice sold by people who do not carry the delivery cost. The correct answer is to spend research effort where the prior is already high and to stop everywhere else.

We tier every list before writing anything. Deep research goes to accounts where two signals already agree. It is also why we drop Tier C rather than send to it.

Source: Unify personalization-depth analysis, cross-checked against our own send economics

04

Optimizing reply rate suppresses qualified meetings.

Reply rate is easy to move in the wrong direction. Vague curiosity-bait openers, "are you the right person?", and soft asks all lift replies while producing conversations that never become pipeline. An agency reporting reply rate is reporting the metric it can most easily flatter.

The honest metric is qualified meetings held, which is slower, noisier, and much harder to fake.

Our guarantee is denominated in held meetings with a decision-maker present, not replies, not "leads", not booked calls that no-show. That choice costs us the easiest number to win on.

Mechanism, not a benchmark. Test it by asking any agency for meetings held rather than replies.

05

The best time to reach a funded company is week six, not week one.

A funding announcement triggers a stampede. Every vendor in the category emails the founder within days, and the inbox becomes unusable for a fortnight.

The useful window opens later. By week six the noise has cleared, budget has moved from announced to allocated, and the executives hired with the round have started. A VP in their first month is the least defended and most motivated buyer in the company, because they are looking for something to change.

We treat a funding round as the start of a watch window, not a send trigger. The round tells us who to watch; the hire that follows tells us when to write.

Mechanism. Verifiable against your own reply data by segmenting on days-since-announcement.

06

Google deleted the domain reputation score. Dashboards still report it.

Google removed the Domain Reputation and IP Reputation panels from Postmaster Tools when the v1 dashboard was retired on 30 September 2025, and shut the v1 API down entirely by the end of that year. Google's stated reason: the low, medium, high and bad labels were routinely misread and lagged real sending behaviour by weeks.

The replacement reports spam complaint rate, authentication status, delivery errors and compliance. Those are the numbers that now decide whether mail lands.

If a deliverability report still shows a reputation grade, ask where the number comes from. We report complaint rate, delivery rate and authentication status, because those are the ones Gmail still acts on.

Source: Google Postmaster Tools v1 deprecation notice, September 2025; v2 API general availability, February 2026