Picture one of your subscribers on a Tuesday. They bought something at the weekend, so they are in the post-purchase flow. They browsed two other products on Monday, so they are in browse abandonment. It has been ninety days since their previous order, so replenishment has fired. The monthly newsletter goes out that morning. And a category campaign is scheduled for the afternoon because somebody in merchandising needs to move stock.

Five emails. Every one of them was built correctly, approved by somebody sensible, and defensible on its own terms. Together they are a business shouting, and the person receiving them does not experience five well-targeted messages, they experience one company that will not leave them alone.

Most programmes have done some work on exclusions and almost none on hierarchy, which are two different questions. Exclusions ask whether somebody should be in a send at all. Hierarchy asks which send wins when several of them are all, legitimately, allowed to go.

The second question is the one nobody answers, because every campaign and every flow gets built in isolation by whoever owns it, and nothing in the platform ever forces anyone to consider what else that person is receiving.

The terms

Three different mechanisms that get confused constantly

• Suppression is permanent and universal. Unsubscribed, complained, hard bounced, or asked you to stop. Applies everywhere, forever, across every system you own, and is never overridden by anybody for any campaign.

• Exclusion is contextual. Somebody should not receive this particular send, for a reason that applies right now. They just bought the thing you are promoting, they are in a live support conversation, they joined yesterday and are still in orientation.

• Precedence is a tie-break. Two or more sends are both valid and both allowed. Something has to decide which one goes, and if you have not written a rule, the answer is whichever one happened to be scheduled first, which is not a decision, it is an accident.

Get those three straight and most of the confusion clears. Suppression protects you from legal and reputational damage. Exclusions protect the customer from irrelevance. Precedence protects them from volume.

Exclusions

Exclusions improve your results by subtraction, which is why they get ignored. There is no launch, nothing to present, no before and after screenshot for the monthly report, and the benefit shows up as an absence of problems rather than as a number going up.

They are also the cheapest improvement available to most programmes, needing no budget, no redesign and no new tooling.

I also believe exclusions are the NEW segmentation

The example exclusion list

• Anyone who just bought the thing you are promoting. The single most common failure, and the one that most obviously signals that you are not paying attention.

• Anyone in a live sales conversation. A promotional email arriving mid-negotiation undercuts the person handling it, and occasionally undercuts the price.

• Anyone with an open service or support issue. Marketing to somebody whose order went missing is how a fixable complaint becomes a public one.

• New subscribers who are still in orientation. Let the first few sends do their job of establishing what your name means before you drop them into the general rotation.

• Consequential opt-ins, held back from hard commercial asks. Somebody who handed over an address to get a discount has not agreed to a relationship yet, and treating them as though they have is how complaint rates climb.

• Anyone who has already done the thing this campaign is asking for. Booked, registered, downloaded, referred. Continuing to ask is irritating and it corrupts your reporting.

• Anyone who has already received their limit this week. Which is where exclusions start turning into hierarchy.

• Your dormant group, held back from broad sends. The people showing nothing at all, inside or outside the email, are the ones costing you placement for everybody else.

Build them at system level, not campaign level

An exclusion applied campaign by campaign will eventually be forgotten on the campaign where it mattered most, usually during a busy week when somebody is covering for somebody else. Build the rules once, apply them automatically, and make overriding one a deliberate act rather than an oversight.

And test them by sending, not by looking at the interface. Run the send to a small audience, then go and check who was excluded and confirm the number matches what you expected. Trusting the count on screen is how people discover in November that an exclusion has silently not been applying since March.

The hierarchy

Which email wins

The organising principle is straightforward once you see it. The closer a message sits to something the person did, the higher it ranks. Broadcast sits at the bottom of the list because it is the least about them, not because it does not matter.

Working from the top down:

1. Legal, security and safety. Password resets, security alerts, regulatory notices, product recalls. Never blocked, never capped, never delayed by anything in marketing.

2. Transactional. Order confirmations, delivery updates, invoices, account changes. Exempt from frequency caps because they are not marketing, and they must never share a sending reputation with your campaigns.

3. Time-critical service. The appointment tomorrow, the delivery slot this morning, the thing that is useless if it arrives late. Beats everything below it because lateness destroys the value entirely.

4. Triggered by the person’s own action. Abandoned basket, back in stock, browse follow-up, form submitted. They did something within the last few days, which makes this the most relevant marketing you can send.

5. Lifecycle automation. Orientation, replenishment, milestone, win-back. Driven by where they are in a relationship rather than by what they did on Monday.

6. Targeted campaign. A send built for a defined segment with a reason behind the targeting.

7. Broadcast. Everybody, or nearly everybody. Perfectly legitimate and the lowest priority, because it is the message least shaped by anything this person has done.

8. Optional extras. Surveys, referral asks, review requests, cross-sell prompts. Valuable, and the first things that should give way when the day is busy.
   
   

Operating rules that make the ranking work

• One primary ask per person per day. Not one email, one ask. Two messages that each want something different is where the sense of being harassed comes from.

• Cap frequency across campaigns and flows together. Almost every programme caps one or the other. Flows do not pause because your campaign calendar got busy, and the recipient does not know which system a message came from.

• Never let marketing block anything above it. A frequency cap that suppresses a delivery update has done more damage than the extra email would have.

• Decide whether a losing email holds or drops. Time-bound content should drop, because a sale email arriving two days after the sale is worse than nothing. Evergreen content should hold and go tomorrow.

• Name who arbitrates. When the rules conflict, one person decides, and it is written down in advance rather than settled in a meeting at four on a Friday.

The common one

Should the flow go, or the newsletter?

The question I get asked more than any other in this area, and the default answer is the flow, for the reason above. A flow email is responding to something the person did, and a newsletter is going out because it is the second Tuesday.

The better answer, though, is to stop deciding this send by send.

Suppress for the duration, rather than arbitrating each time

When somebody enters a flow that matters, take them out of broadcast for as long as they are in it. Onboarding runs for two weeks, so they get no newsletter and no campaigns for two weeks. Post-purchase runs for ten days, same thing. They come back into the general rotation when they exit.

Which sounds like it costs you reach and does not, because the flow is almost always the higher-performing message for that person at that moment, and you have removed an entire category of decision from your weekly workload at the same time.

The exceptions worth writing down

• Time-bound content that will not repeat. A closing date, an event, a change to terms. Let it through, and count it against the cap.

• Flows with low-value filler in them. If a flow email is weak enough that you would rather send the newsletter, the right fix is to improve or remove the flow email rather than to override the rule.

• Very long flows. Suppressing somebody from everything for eight weeks is not protection, it is neglect. Long flows should overlap with broadcast at a reduced rate rather than blocking it entirely.

The friction

What a collision feels like from the other side

Worth spending a moment on the experience rather than the mechanics, because the damage is not proportional to the number of emails.

• Contradiction. A win-back saying you have not heard from them, arriving the day after they placed an order. Nothing damages credibility faster than visibly not knowing who somebody is.

• Repetition of the ask. Three emails in a day that each want a purchase reads as desperation rather than as three offers.

• Loss of signal. When everything arrives at once, the important message is buried inside the unimportant ones, and the one that mattered gets deleted with the rest.

• The complaint shortcut. Somebody who feels overwhelmed reaches for whichever exit is nearest, and on a phone the junk button is nearer than your unsubscribe link.

Which is the part that makes this a deliverability subject rather than a customer experience one. Collisions produce complaints, complaints damage your sending reputation, and the reputation damage lands on every email you send afterwards, including the ones that were perfectly judged.

How to build it

The conclusion

Most programmes are not suffering from a shortage of good emails. They are suffering from an absence of anybody deciding which good email goes when several of them could, and that decision currently gets made by scheduling order, which is to say by chance.

Exclusions and hierarchy are unglamorous, they will never win an award, and they will do more for your results this year than any redesign or subject line test you have planned. Improvement by subtraction is still improvement, and it compounds, because every collision you prevent is a complaint that never happened.