Email Outreach Is Four Decisions, and Only One of Them Is Writing

Email outreach is the practice of writing to people who did not ask to hear from you, in a way that earns a reply. For an agency selling to ecommerce stores it contains four decisions: who you are any good for, when to write to them, what the message has to earn, and what happens after the first send. Only the third is a writing problem.

The disclosure first. We build Keaz Signals, which sells store data and live buying signals for ecommerce outreach, so we have a commercial interest in the argument below. We are a competitor to whatever else you are evaluating, and two sections further down set out where this approach is the wrong one.

The term gets used for two different things, and the confusion is expensive. Sometimes it means the practice above. Sometimes it means the software that executes it. A team that buys the software while never settling the four decisions ends up with a faster version of a motion that was not working, which is the most common way outreach budgets disappear without anyone being able to say what went wrong.

The first decision is which kind of store you are genuinely good for, and it is the only one of the four that no software supplies. A segment builder asks you for conditions and returns a count. The conditions come from knowing why your best three clients bought, which is knowledge that lives in your own history rather than in any database.

This decision changes rarely. Twice a year is normal. It is also the decision teams revisit weekly when results disappoint, because re-cutting a segment feels like work and produces a new number to look at. The segment was usually not the problem; the timing layer underneath it was missing, so every cut of the market performed the same.

On our side the conditions are country, platform, follower count and similar attributes, with a live audience estimate as you narrow them, drawn from the ecommerce leads database. We went further into how a storefront audience differs from a conventional company list in Customer Segmentation When the Customer Is a Storefront.

The second decision is when the message goes out, and it is the one you cannot practise your way into. Copy improves by iteration: write fifty subject lines, keep the three that worked. Timing does not, because the variable is an event at the other company rather than anything inside your own process.

That makes it a data question rather than a skill question. Either you know that a store installed an email marketing tool this week, or you do not, and no amount of craft substitutes for the knowing. We argued the full version of this from the storefront side in Cold Outreach That Starts From the Storefront, so the short form here: a dated event answers why this store and why this week in one clause, and nothing else in the message has to carry that weight.

Freshness is the part that decays fastest. Our own catalogue stores every signal with the date it was last seen, and we surface events that are days old rather than weeks, because a three-week-old signal is noise dressed up as an opportunity. The buying signal catalogue lists the seven event types that currently fire and what each one is evidence of.

The third decision is what the message has to earn, and the answer is narrower than most outreach advice suggests. A first email does not have to sell the engagement. It has to earn one reply from a person who has never heard of you, which means it has to be obviously about them and obviously short.

This is where the first two decisions pay out. If you know the segment and you know what just happened at that store, the opening clause writes itself and it is specific in a way no template library can fake. If you do not, the message reaches for personalisation tokens instead, which is why so much outreach reads as though it was assembled rather than written. It was.

The signal type also decides the angle, not just the opener. A rising ad count and a fresh storefront rebuild are different problems arriving at different desks, and a message that treats them as interchangeable has already lost the specificity it spent the signal to get. We wrote one worked example per event type in Cold Email Templates, One Per Buying Signal, and took the opening line apart separately in Cold Email Subject Lines Driven by a Buying Signal.

The fourth decision is what happens after the first send, and most teams answer it with a calendar. Step two on Thursday, step three the following Tuesday. That rhythm is a property of your scheduler rather than of the situation, and it keeps running long after the reason for the first message has expired.

The alternative is to space follow-ups by how long the original event stays relevant. A product launch is a two-week window. A storefront rebuild runs for months. Same sequence length, different spacing, and the difference is visible to the recipient even when they cannot articulate why one email felt timely and another felt automated. We took this apart properly in Cold Email Follow-Up, Timed by the Signal Not the Calendar.

The other half of this decision is what a reply costs you. On our own pricing one credit is one person you contact, once, and sequence steps, enrichment and re-contacting are free, so a three-step campaign costs the same as a one-step one and coming back in six months costs nothing again. That is a deliberate choice about which behaviour to make cheap: following up properly, and not writing to more people.

Sending tools sit under the four decisions rather than over them. A sequencer schedules steps, handles replies, rotates mailboxes and protects your sending reputation. None of those functions has an opinion about whether the store receiving the message is worth writing to, which is the decision that most often goes unmade.

This is not a criticism of the category, and we are not neutral about it: our own campaigns are created in the customer's own Instantly workspace, on their own domains, and we take no cut of what goes out. The sending layer is genuinely difficult and worth paying for. It is simply answering a different question from the one the four decisions ask.

The confusion between the two is worth naming because it has a predictable failure mode. When results disappoint, the visible levers are all in the tooling: more mailboxes, a better sequencer, another vendor. Changing the list has no receipt and nothing to put in a status update, so it does not get changed. We wrote the long version of that in Cold Email Software Is Not a Lead Source and covered what the platforms themselves do in What a Sales Engagement Platform Does.

We are not going to tell you what email outreach will do to your reply rate, because we have not measured your market. We have one measurement and it is ours: across 660,000 emails sent for our own customer acquisition, as of August 2026, writing to stores in the week an event fired produced a reply rate of 8.5 percent against 3.9 percent without.

Read that as a description of what happened to us rather than as a forecast. Same team, same offer, our market, our sending setup, an unchanged message on both sides of the comparison. Internal consistency is its only real virtue. It is not a controlled study and we are not going to dress it up as one.

Two other numbers we will not give you. We will not quote an industry-average reply rate for cold email, because we have not run the study that would let us publish one and the figures in circulation rarely say what they measured. And we will not tell you how many hours a week this saves a three-person team, for the same reason. An hours-saved figure is a preference wearing arithmetic, and it travels much further than the caveat attached to it.

We make no deliverability or reply-rate guarantees of any kind. Your domains, your mailboxes and your sending reputation are yours, and no list rescues a misconfigured domain.

This approach is the wrong one in four cases, and each is cheap to check before you spend anything. They are limits rather than roadmap items, and we would rather you found them here than three weeks in.

• Your buyers do not run Shopify or WooCommerce stores. We watch no other platform, so the segment builder has nothing to show you.
• Your market sits outside Europe and North America. Same answer, for the same reason.
• Your motion depends on LinkedIn or the phone. There is no LinkedIn sending here and no dialler; this writes and sends email, and nothing else.
• Your calendar is already filled by referrals. Then you do not need any of this yet, and we say so on our own homepage.

Several event types people ask for are marked planned on our own catalogue rather than live, including hiring signals and registry triggers. Planned means they do not exist. Access is capped as well: we release it market by market, because leads lose their value when four agencies write to the same founder in the same week, and that is a limit on our own growth before it is anything else.

Running the loop means answering the four decisions once and then letting the last three repeat. The segment is set by a person and revisited rarely. The event decides the week and the angle, and the follow-up spacing comes from how long that event stays true.

On compliance we hold a deliberately narrow line: we are GDPR-conscious by design and will not claim more than that, and responsibility for what you send stays with you. That is a statement about how the system is built, not a claim about anyone's compliance.

As of September 2026 we watch 4.1 million Shopify and WooCommerce stores across Europe and North America, re-scraped weekly, and roughly 41,200 carried a fresh signal in the last seven days. Those figures are ours and they move, which is why they are dated. Stores arrive with contacts attached through the ecommerce leads database. Access is released market by market, so to check whether yours is still open, join the waitlist on the pricing page.

• Credit model, market caps and the 660,000-email reply-rate measurement — our own pricing page, figures as of August 2026, retrieved 16 September 2026
• Which events fire, their last-seen dating, and which are marked planned rather than live — our own buying signal catalogue, retrieved 16 September 2026
• What the agent writes, and the email-only boundary — our own AI sales agent page, retrieved 16 September 2026
• Store coverage and weekly fresh-signal counts — our own figures, as of September 2026