When nurture sequences start annoying people
The email that damages you is almost never badly written. It is the fifth introductory message sent to someone who bought last month, or the case study that arrives the day after a prospect told your salesperson no. Nurture goes wrong through missing exit conditions rather than weak copy, and that failure is invisible in the reporting, because a sequence sending to the wrong person still reports a healthy delivery rate.
What actually makes a sequence irritating?
Sending something the recipient has already moved past. A person who has spoken to your sales team, bought, cancelled, or explicitly said not now, has a state that the sequence does not know about, and continuing to address them as a fresh prospect signals that nobody is paying attention. That is the specific irritation, and it is about relevance to their situation rather than frequency.
The second cause is arriving alongside other messages from the same company. Three separate flows each sending weekly are, from the recipient's side, one company mailing three times a week. Nobody designed that; it emerged because each flow was designed on its own.
The third is asking repeatedly for something already given. A form asking for a name and company from someone whose record already holds both, or a sequence that continues to push a download that was completed, both tell the reader that the system does not recognise them.
Why do exit criteria get forgotten?
Because building a sequence is a creative task and exits are an administrative one, and the platform will happily launch a flow with entry criteria and no exits at all. The entry rule is the interesting part and gets the attention; the question of who should stop receiving this, and on which event, gets left as an assumption.
Exits also require knowing about events that happen in another system. A deal moving to closed-won lives in the CRM, and unless that state syncs back promptly, the sequence has no way to know. The gap between the deal closing and the platform learning about it is exactly the window in which the wrong email goes out.
So the discipline is to write the exits before the emails. For every sequence, list the events that should remove someone immediately, and confirm that each of those events is visible to the platform in something close to real time rather than in a nightly sync.
| Recipient state | What should happen | Common failure | How to detect it |
|---|---|---|---|
| Replied to a sequence email | Pause the sequence, hand to a human | Sequence continues around the conversation | Search the reply inbox for threads that received later automated mail |
| Has an open opportunity with sales | Suppress prospecting flows | Prospecting continues in parallel with negotiation | Cross-reference active sequences against open deals |
| Bought | Exit immediately, move to onboarding | Acquisition sequence keeps running for weeks | List customers currently enrolled in any acquisition flow |
| Said no, or asked for later contact | Exit, with a dated re-entry if agreed | No mechanism exists, so nothing changes | Check whether a not-now field exists at all |
| In three or more active sequences | Frequency cap applies, lowest priority pauses | Each flow sends independently | Count contacts enrolled in multiple live flows |
| No engagement for a long period | Move to a low-frequency track, then suppress | Sends forever at full frequency | Segment by last engagement date and check what is still sending |
Do unsubscribes tell you when you have gone too far?
Not really, and relying on them is the reason teams believe their programme is fine. Unsubscribing takes effort and creates a small social cost, so most annoyed recipients simply stop opening, or move the sender to a folder, or mark as spam without ever unsubscribing. The unsubscribe rate measures the most conscientious fraction of the annoyed.
The quieter signal is engagement decay within a cohort. Take everyone who joined the list in a given month and watch their open and click behaviour across the following months. A steep decline means the programme is exhausting its audience faster than it is earning attention, and that will show up long before unsubscribes move.
Complaint rate is the signal that matters operationally, because mailbox providers act on it. It is also the one that responds fastest to sending people things they did not expect, which links the annoyance question directly to whether your mail continues to arrive at all.
How often is too often?
There is no universal number, and any answer given as one is guesswork. The useful frame is that frequency should be set per audience by watching engagement decay, and capped globally so that no combination of flows can exceed it regardless of how many sequences a person qualifies for.
A global frequency cap is the single most valuable setting in most platforms and the least used. Without one, a person who happens to match four segments receives four times the intended volume, and no individual flow owner sees anything wrong.
Milestone-triggered messages beat calendar-driven ones on the same budget of attention. A message that arrives because the recipient did something is expected. The same message on a fixed weekly schedule is an interruption, even when the content is identical.
What does a good nurture sequence look like structurally?
Short, with a defined end, and a stated purpose that can be tested. Four to six messages moving someone from a specific starting state to a specific next action, after which they leave the flow rather than looping. An indefinite sequence is a subscription the recipient never asked for.
Each message should be able to justify itself alone: if a reader received only this one, would it be worth their attention. Sequences designed as a narrative arc fail this test, because messages two and three exist only to set up message four and the reader has no reason to care.
Then instrument replies. A nurture flow that generates replies is working, whatever the click rate says, and replies are the only feedback in the system that arrives in the recipient's own words. They land in an inbox rather than a dashboard, which is why most teams never read them.
How do you audit the sequences you already have?
Produce one list: every live sequence, its entry criteria, its exit criteria, its owner, and the date it was last edited. Most teams cannot produce this without an afternoon of clicking, and the difficulty is itself the finding. Anything with no named owner or no edit in over a year should be paused rather than reviewed, because pausing is reversible and reviewing never happens.
Then run the cross-checks from the table above against live data. The counts that matter are customers enrolled in acquisition flows, contacts in three or more live sequences, and contacts receiving mail after replying to a human. Each of these is a specific bug with a specific fix.
Do this quarterly. Sequence estates grow by accretion, because adding a flow is a decision one person can make and removing one feels like deleting someone else's work.
Common questions
- Why do nurture sequences annoy people?
- Usually because they address a state the recipient has already moved past. Someone who has spoken to sales, bought, cancelled or said not now keeps receiving messages written for a fresh prospect, which signals that nobody is paying attention. The second cause is several flows sending independently, so the recipient experiences one company contacting them three times a week that nobody designed.
- What exit criteria should every nurture sequence have?
- At minimum: replied to a message, opened an opportunity with sales, bought, asked not to be contacted for now, or exceeded a global frequency cap. Each exit depends on an event that often lives in the CRM rather than the sending platform, so the sync delay between the two is exactly the window in which the wrong email goes out. Write the exits before the emails.
- Is the unsubscribe rate a good measure of whether you are sending too much?
- No, it measures only the most conscientious fraction of annoyed recipients. Most people stop opening, filter the sender, or mark as spam without unsubscribing. Better signals are engagement decay within a joining cohort tracked over months, and the spam complaint rate, which mailbox providers act on directly and which responds quickly to sending people things they did not expect.
- How many emails should a nurture sequence contain?
- Few enough to have a defined end, commonly four to six, moving someone from one specific state to one specific next action and then exiting. Indefinite sequences are a subscription the recipient never agreed to. Each message should justify itself in isolation, because sequences built as a narrative arc rely on the reader caring about messages that exist only to set up a later one.
- What is a global frequency cap and why does it matter?
- A limit on how many messages any one person can receive from all flows combined in a given period, applied above the individual sequences. Without it, someone matching four segments receives four times the intended volume and no individual flow owner sees anything unusual. It is commonly the most valuable setting in a marketing platform and among the least used.
