How to Set a Contact Frequency That Does Not Annoy

Why this matters

Every shop that has ever been told to "stay in front of customers" ends up sending more than it thinks and less usefully than it hopes. The damage does not arrive as complaints. It arrives as a slow decay in which your best customers stop opening anything from you, so the one message that mattered, the due notice or the coverage expiry, lands in the same ignored pile as the seasonal promotion. By the time you notice, the fix is expensive: you cannot un-train a customer who has learned that your messages are not worth reading. This card is about volume control specifically. What to contact them about is a separate subject and is covered in the cadence reference at the end.

Step 1: Count what you actually send, because you are undercounting

Ask an owner how often they contact a customer and the answer is usually three or four times a year. The real number is several times that, because messages leave the shop from separate places and nobody has ever added them up: dispatch confirmations, day-before reminders, on-the-way texts, invoices, payment reminders, review requests, post-job follow-ups, service due notices, seasonal offers, a newsletter, and whatever a technician sends from their own phone.

The audit is simple and it must be done per customer, not per system. Pick three real customers with different profiles. For each, reconstruct every single message they received in the last twelve months, from every source including the technician's phone. Put them in date order.

Why per customer: each system looks reasonable on its own. A newsletter at monthly is defensible. Review requests after each job are defensible. Due notices with one resend are defensible. The collision only becomes visible on a single timeline, and the collision is what the customer experiences.

Step 2: Split the count into operational and non-operational

Operational messages are attached to a job in progress: booking confirmation, reminder, arrival notice, invoice, payment reminder. These are expected, they are wanted, and they do not count against your budget. A customer with a job on the calendar is not annoyed by hearing about that job.

Non-operational messages are everything else: due notices, follow-ups, review requests, seasonal notes, promotions, newsletters. These are the entire budget. Every one of them spends attention.

Getting the split wrong in either direction causes real problems. Counting operational messages against the budget makes shops suppress arrival notices, which produces actual complaints. Counting a promotion as operational because it went out with an invoice is the oldest trick in the book and customers see through it immediately.

Step 3: Set a hard annual budget per segment

Commit to numbers. "As needed" resolves to nothing in the busy season and a panic blast in the slow one, which is the worst possible pattern.

Those are starting points to tune, not laws. But pick one for each segment and write it down, because a budget you have not written is not a budget.

Step 4: Write the suppression rules

The budget controls total volume. Suppression rules control clustering, which is what actually generates irritation. Six rules cover almost everything.

1. No non-operational message within 7 days of another non-operational message, to anyone, for any reason.
2. No marketing message while a job is open or an invoice is unpaid. A promotion arriving while a customer is waiting on a repair or disputing a bill reads as tone-deafness, and it undermines the invoice conversation.
3. No review request within 30 days of a complaint, callback, or warranty visit on that record. This one prevents the single most self-inflicted bad review a shop can generate.
4. One channel per message. The same content by text and email is not thoroughness, it is two messages. Duplicating across channels is the most common way a shop doubles its real frequency while believing its numbers are unchanged.
5. Three consecutive non-responses moves the customer to the low-engagement budget until they interact with something.
6. A ten-day blackout on non-operational sends after any price or terms change notice. Let that message stand alone.

Safety notices and recall notices are exempt from all six. They go out regardless of budget, timing, suppression, or engagement status, they go alone, and nothing is bundled with them.

Step 5: Instrument for irritation, not for engagement

Open rates and click rates measure the people who are still paying attention. They are structurally incapable of showing you the damage, because the damage is people who quietly stopped. Track four things instead.

• Opt-out rate per send, against your own trailing average. Do not chase an industry benchmark. Investigate any individual send whose opt-out rate is more than double your own trailing average across the previous several sends. That comparison is reliable regardless of what your baseline happens to be.
• Explicit stop requests by any route, including a customer telling a technician on site. These are worth ten opt-out clicks each as a signal, because a customer who bothers to say it out loud represents a much larger group who did not.
• Reply-rate decay across a sequence. If message one gets replies and message four gets none, the sequence is too long, not the audience too small.
• Booking rate per send. A message with strong opens and no bookings is consuming budget and producing nothing. Cut it before you cut a low-open message that reliably books.

The asymmetry that makes opt-out rate misleading: the customers most likely to click opt out are the ones with the least invested in you. Your best customers rarely opt out. They simply stop reading, keep receiving, and remain in your counts looking perfectly healthy right up until the day they use somebody else. So a low opt-out rate is not evidence that the frequency is fine. Booking rate on anchored messages is the number that would notice.

Step 6: A real audit, worked through

All figures illustrative, and the profile is common. One residential customer, annual-interval trade, twelve months, one repair job plus one maintenance visit.

Reconstructing every message from every source:

• Booking confirmations, 2
• Day-before reminders, 2
• On-the-way texts, 2
• Invoices and one payment reminder, 3
• Review requests, 2
• Post-job follow-up, 1
• Service due notice plus one resend, 2
• Seasonal offers, 4
• Newsletter, 12

Total, 30 messages in twelve months. The owner's estimate before the audit was four or five.

Split it. Operational, meaning job-attached: 2 plus 2 plus 2 plus 3, which is 9. Non-operational, meaning everything spending the budget: 2 plus 1 plus 2 plus 4 plus 12, which is 21.

So this customer received 21 non-operational messages in a year against a target budget of 6 to 8 for an active annual-interval customer. And 12 of the 30 total messages, which is 40% of everything they got from the shop that year, were newsletters carrying no fact about their property.

Rebuild it inside budget. Keep the 2 due-notice touches, the 1 post-job follow-up, and 1 review request rather than 2, since the second job was a maintenance visit on a customer who had already reviewed. Keep 2 seasonal notes, but only where the item genuinely applies to equipment they have. Add 1 coverage or age anchor from the file. That is 7 non-operational messages, comfortably in range, and every one of the 7 is anchored to something about them.

Read the change carefully. Non-operational volume goes from 21 to 7, a cut of 14, which is about a two-thirds reduction. Total contact including operational goes from 30 to 16. The newsletter, which was 12 of the 21 non-operational messages, or a bit over half of the entire budget spend, is gone, and it was the only category that could not pass the test of containing a fact about the customer.

What the shop gets back for the cut is not fewer complaints, because there were not many complaints. It is the due notice landing in an inbox where the last thing from this shop was useful, rather than in one where the last eleven things were not.

Step 7: Tune one variable per quarter

Change one thing per quarter and watch the four irritation measures for a full quarter before changing another. Frequency, channel, timing, and content all interact, and moving two at once means you learn nothing.

Order of operations that works: cut the lowest-booking category first, because that is a pure gain. Then fix duplication across channels. Then adjust timing. Only then consider adding volume back, and only to the segment that is booking.

What changes the numbers

Plan and agreement members tolerate and expect more, because they are paying for a relationship rather than a transaction. Their budget is legitimately at the top of the table, and under-contacting a member is its own failure that shows up at renewal as "I never heard from you all year."

Commercial contacts want fewer messages and different ones. The individual reading them is doing a job, not maintaining a relationship with you, and they are measuring you on whether the property gets serviced and the paperwork is clean. Cut the relationship-building volume hard and route anything that matters through whatever documentation channel they actually use.

A trade with an emergency component can support slightly higher frequency, because being top of mind has real value at the moment of a burst, an outage, or a failure. That is an argument for one or two more anchored touches, not for a newsletter.

A stated preference always wins, and an opt-out must be honored across every channel you hold for that customer, not only the one it arrived on. For commercial email specifically, the federal CAN-SPAM Act requires an opt-out to be honored within 10 business days, and text messaging carries its own consent rules; the operational habit worth building is to process any stop request the same day it is received, on all channels, which keeps you clear of the deadlines rather than close to them.

How to verify you got this right

Re-run the per-customer audit from step 1 twelve months later, on three different customers, and compare non-operational counts against the budget you set. That is the direct check and it is the only one that catches a new message source that appeared quietly during the year, which is the usual way a budget breaks. A new review-request tool, a technician who started texting customers directly, or a supplier-provided seasonal mailer can each add several messages a year that no one counted.

The second check is a sample of ten recent non-operational messages read cold. For each, ask whether it contained a fact about that specific customer's property or history. If fewer than half do, your frequency is not the real problem and cutting it further will not help; the content is inert and would be ignored at any volume.

References

• Federal Trade Commission, CAN-SPAM Act compliance guide for business, including opt-out handling requirements for commercial email
• Federal Communications Commission, consent and revocation rules for calls and text messages to consumer numbers
• See related: The Communication Cadence Between Jobs; The Channels a Customer Actually Wants You to Use; Texting Customers the Professional Way; The Customer Text Campaign Done Right