What this template is for
Draw up the policy as soon as separate teams or tools can reach the same customer, and before any new sending tool, channel or team with its own sequences starts. Do it straight away if people complain to the service team, or in public, that the business contacts them too much: the troubleshooting page on customers who get too many marketing messages traces that complaint back to a missing rail, and this policy is where the rail gets written so it does not slip again.
Where another document already owns a piece of the problem, the policy points to it instead of copying it. The automation governance guide sets out the shared rails and maintains the automation inventoryThe single list of every active automated workflow, kept outside the platform, with its purpose, owner, trigger, exits, dependencies, measure and review dates., where each workflow's row shows which workflows outrank it and which rails apply to it; the rails themselves are decided here. The email marketing calendar template carries the email cap at the top and tests each week's plan against it, while this policy is where that cap is chosen next to the others. The email sign-up source register template keeps what each form promised about content and how often, and the caps have to honour those promises. The pre-send check for texts and app messages looks up its channel's cap and order before a send, and the SOP for marketing opt-out requests processes each request to stop.
Neighbouring problems have pages of their own. When sequences meant for new leads reach existing customers, start with the diagnosis for customers receiving prospect messages; when a single email is followed by a rush of unsubscribes, with the unsubscribe spike diagnosis; and when app notifications are being disabled, with the page on app users turning off notifications. When a fix from one of them alters a cap, the order or a choice offered to people, write the change into this policy.
The template
Copy the table into a shared document with one row for each field. Where a field varies between channels, give each channel its own column instead of adding rows, so that every limit one person faces can be read along a single line. Complete the senders and the service test before anything else: a cap means nothing until you know what it counts.
| Field | What to write |
|---|---|
| Senders in scope | Each tool and team able to put a marketing message in front of one person (the email, text and app tools, and any sales, membership or loyalty sequence), with its owner |
| Service or marketing | The test that decides whether a message is service or marketing, the service messages that sit outside every cap, and the rule for a service message that carries an offer |
| Cap for each channel | The promotional cap per person for each channel, the period it counts over, where the check runs before every send, and who owns it |
| Texts and WhatsApp | The cap across every messaging flow, kept within the sending frequency the opt-in wording promised, and the rule that moves unresponsive people to service messages only |
| App messages | One cap for push and in-app messages shared by every team and tool that sends them, with promotions at the back of the queue |
| Priority order and exclusions | Which flow, campaign or sales sequence goes first when several qualify for the same person, which may not run side by side, and how the order is enforced |
| Quiet hours | The hours in each recipient's local time when promotional texts and push wait, set as a system rule, with service messages let through and hard limits left to each market's legal review |
| Topic choices | The topics people can choose by, named in their words, which of them count towards a cap, the owner of each topic, and the matching choice in the email preference centre |
| Holds while an issue is open | The reasons that put a temporary hold on promotion, who sets each one, the service messages that keep running, and the rule and owner that release it |
| Suppression and requests to stop | The suppressions every send reads, such as unsubscribed, bounced, complained and customers kept out of prospect sequences, and how a request to stop is applied to the scope the person gave |
| Inactive contacts | What counts as inactive for each channel, the re-permission step if one is used, and what happens to people who do not respond |
| Survey and research invitations | How research invitations are spaced across every team that sends them, kept free of promotion, and kept apart from people's marketing choices |
| Changes, tests and review | Who owns the policy, how a change to a cap, the order or a rail is reviewed and logged, the seed contacts that test it, the look at the most-messaged customers, and the date of the last review |
Above the table, record the owner, the date of the latest change and, for each tool, where each rule is built: the shared condition, segment or setting that enforces it. A rule written only in the document is one that no send ever checks.
How to fill each section
Senders in scope
Name every tool and team that can reach one person with marketing: the email platform, the tool that sends texts and WhatsApp messages, the app engagement tool, and any sequence that sales, account management, membership, loyalty or store teams run separately, which may live in the CRM rather than in the marketing platform. The too-many-messages diagnosis begins by piecing together one complaining customer's weeks from every one of those send logs. Put customer lists uploaded to ad platforms on a line of their own, because the holds and requests to stop further down must still reach them. For each tool, write its owner and whether it can check a shared cap before sending. A cap counted per record also fails for someone stored twice, so name the diagnosis for duplicate contacts in the CRM as the page that traces where the second record comes from.
Service or marketing
Write the test first, then list the service messages that pass it. The comparison of promotional and transactional email decides by purpose: a message about something the person began, or one the relationship obliges the business to send, is service, while a message meant to promote is marketing whatever the campaign is called, which makes a reminder about an abandoned purchase marketing. Mixed messages are the difficult ones, such as a shipping confirmation carrying product suggestions, and the comparison suggests picturing the message with the promotion stripped out. Either count such messages against the caps or remove the promotion, and when a message could go either way, treat it as promotional, as the topics tutorial recommends. List the transactional and service messagesAn email triggered by a person's own action or account, such as a receipt, password reset or shipping notice, rather than a marketing send. the business actually sends, such as confirmations, appointment reminders, receipts, account notices and replies to complaints, because they sit outside every cap below, carry on during a hold and never give way to marketing.
Cap for each channel
Write each channel's promotional cap, the period it covers, where the check is built in each tool and its owner. The automation governance guide supplies the control itself: one limit per channel on the marketing messages a single person can get within a set period, tested before each send and not applied to service messages. Its reasoning: workflows that are each sensible can still add up to too much message FrequencyAverage times a person saw an ad. for the person, and only a channel cap looks at the total. Name the cap in each tool so it is plainly a protected, shared asset, as the guide asks for shared assets generally. For email, one cap covers broadcasts and automated flows together, and the calendar repeats it at the top. No guide on the site sets a figure: set it from what people were promised when they signed up and from your own record of complaints, unsubscribes and disables after heavy weeks, and note the reason next to it.
Texts and WhatsApp
Write one cap across all messaging flows, the opt-in wording it must respect, and what happens to people who never respond. The guide to SMS and WhatsApp automation treats texts more strictly than email because they interrupt the moment they arrive. Its example is one person who qualifies for a basket reminder, a renewal flow and a back-in-stock alert together, each reasonable alone; the answer is a single cap for the channel, checked as a condition ahead of each promotional send, with correctly classified transactional messages usually left outside it. The guide also reports that the CTIA's messaging principles expect the frequency promised at opt-in to be kept, so the cap can never exceed what the opt-in wording said. Store the wording and the cap together, so that changing one forces a look at the other. Finally, write the guide's stop rule: anyone with no reply, click or purchase across several promotional texts moves to service messages only.
App messages
Write one cap that push and in-app messages share, the teams and tools bound by it, and where promotions sit in the queue. The push and in-app messaging guide wants the limit set per person and per period and applied to every campaign from every team able to send, since campaigns judged one at a time each look fine while the person gets the lot. It adds that most engagement platforms offer global caps and priority settings, and it keeps transactional messages outside the cap with promotions ranked last. The guide to designing in-app messages brings in-app messages under the same cap, because to the person, push and in-app are a single app speaking. Note which topics count against the cap.
Priority order and exclusions
Write the order from highest to lowest, the pairs that must never run together, and what a losing message does: waits, or is dropped. The governance guide settles, for a single contact, which workflows are mutually exclusive and which takes precedence, for example a demo request over a nurture, a basket flow over a newsletter, and service messages over all else. On a platform without built-in priority, entry conditions do the work, together with a flag raised while a higher-ranked workflow is active. The lifecycle marketing guide explains where scheduled sends belong: a newsletter or a seasonal announcement sits on top of the triggered flows rather than replacing them. Bring in the sequences run by sales, membership and loyalty teams, since they reach the same people, and record each workflow's rank in the exits and priority column of its inventory row.
Quiet hours
Write the window for texts and for push, how each person's local time is worked out, and what becomes of a message that lands inside the window. The texts guide sends by the recipient's own clock, keeping a stored time zone or deriving one from the country code and the address, and keeps promotional sends back until daytime. It points out that quiet hoursThe hours when promotional texts, WhatsApp messages and push notifications are held for the recipient's daytime, set by law in places and courtesy elsewhere. are a legal requirement in some places and a courtesy elsewhere, that none of its sources fixes one window for everyone, and that each market's legal review decides the hard limits. The push guide treats quiet hours as a rule of the whole system, not something each campaign sets, holding non-urgent messages until morning while service messages pass. Take the texts guide's timing point too: a step that cannot go out promptly and inside the window is usually better skipped than sent late, so say which held messages wait and which are discarded.
Topic choices
Write the topics, whether each is service or promotional, whether it counts against a cap, who owns it and the matching email option. The tutorial on notification preferences by topic sorts every notification into a small number of topics described the way customers would describe them. It keeps service and promotional messages apart, so that someone who switches off offers still gets everything they rely on; each topic has a single owner who signs off any new message added to it, and every sending tool checks the same preferences before it sends. The tutorial also writes down which topics count against the cap, so the rule outlasts whoever builds the next campaign. The too-many-messages diagnosis connects this to email: adding a topic in the app means adding the same option to the email preference centreA page where a subscriber chooses which messages, topics and channels they receive and updates their own details, instead of an all-or-nothing unsubscribe. as part of one change, and the list building and consent guide asks the centre to offer only options the team will honour, kept short, with a reduced-frequency option respected as a genuine choice.
Holds while an issue is open
Write what starts a hold, who sets each kind, what stops and what carries on, and the release rule with its owner. The tutorial on pausing promotion while a customer's complaint or dispute is open builds the hold as a flag on the contact that every send and sequence reads, takes held customers out of the uploaded lists that target them while leaving them in the exclusion lists, and sets out who places a hold, what keeps running and how it ends; this policy records only the business's own choices on those points, not the build. A hold is never used in place of a request to stop, which is an objection. When a problem touches all customers together, the answer is a pause across the whole business, set out in the tutorial on marketing leadership during a business-wide crisis.
Suppression and requests to stop
Write the suppressions that every send checks and how a request to stop is carried out. The governance guide's global suppressionA list of addresses or contacts excluded from every send, such as unsubscribes, hard bounces and spam complaints. list is read on entry and again before each send, and it covers people who unsubscribed, bounced or complained, those who asked not to be contacted, anyone under a legal hold and any regional exclusions. Beside it sit the customer and pipeline suppressions, which keep current customers out of prospect sequences, contacts with an open opportunity away from automated sales nudges, and people judged not a fit from entering nurture again. The SOP for marketing opt-out requests handles a request to stop, however it is worded and wherever it arrives, as an opt-out and applies it to the scope the person gave in every tool on the consent map. The regulator's direct marketing guidance cited in the sources, as the too-many-messages diagnosis summarises it, expects marketing to end once a person objects, and the SOP keeps every opt-out on the suppression list so no import or sync can add the person back. Note the routes here and let the procedure do the work.
Inactive contacts
Write what inactive means in each channel, any re-permission step, and what follows for people who stay silent. The email list hygiene SOP starts from an inactivity definition agreed beforehand, flags inactive contacts during its routine pass, runs re-permission and removes those who do not respond. For texts, the stop rule in the texts field already applies. Keep the definition in this policy so the calendar, the flows and the hygiene routine all work from the same one, and leave decisions on archiving or deleting inactive records where the governance guide puts them: with legal and privacy colleagues, under the business's own retention rules.
Survey and research invitations
Write the minimum gap between research invitations to one person, who keeps the record of who has been asked, and the rule that invitations promote nothing. Recorded here, that gap becomes the business's contact ruleA business-wide rule that a customer surveyed or interviewed recently is not asked to take part again for a set period, whichever team wants to ask. for research, separate from the marketing caps elsewhere in this policy. The tutorial on sampling a customer list for a survey takes out anyone already surveyed or interviewed inside the period the contact rule sets, whichever team asked them, and then applies whatever the data protection owner decided about people who opted out of marketing. It also treats an invitation that pushes a product as marketing, whatever the subject line claims. Whether invitations count against the marketing caps is for the business to decide; record the decision and its reason, so that a survey wave landing in the same week as a sale is deliberate rather than accidental.
Changes, tests and review
Write the owner, the review each change needs, the tests and the regular checks. The governance guide requires a second person to review any change to the frequency cap, the global suppression list or consent conditions before it goes live, since each one reaches every workflow; every change gets a log entry, and seed contactsAn inbox or test contact the business controls, added to a send or run through a workflow so the team sees exactly what real recipients would receive. are sent down every branch afterwards. Its monthly audit uses seed contacts to spot-check suppression and consent conditions and looks for unauthorised changes to shared assets. Use a RACI matrix with one accountable owner for the whole policy and one for each rule that teams share. Then add the diagnosis's habit of watching the total: a regular look at the customers who got the most promotional messages across every channel, with one history rebuilt in full, and a rule that no new sending tool or team sends until it reads the caps, the order, the topics and the holds.
Filled example
Here the policy is filled in for a fictional business. Its fields appear as rows; in the working document the channel caps would sit in adjacent columns. Every figure, date and rule is illustrative and was chosen for this business from its own records.
Klempraxi Trampoline Parks: customer contact policy
Klempraxi Trampoline Parks is an invented operator of 3 indoor trampoline parks with a booking app, an email list of 41,000, texts for reminders and offers, and a membership scheme. Its parties team was about to move party enquiry follow-up into the CRM as an automated email sequence, the fourth sender able to reach the same parent. The term before, a refund dispute had shown that holds were set only in the email platform: the parent kept receiving offer texts and app messages until the dispute closed. The marketing operations lead wrote the policy with the CRM lead, the app product lead, the parties manager and the finance director, who owns data protection, and it took effect before the parties sequence went live.
Owner: marketing operations lead. Latest change: 3 November 2026. Built as shared conditions in the email platform and the CRM, a send condition in the text tool and global settings in the app messaging tool.
| Field | Klempraxi's policy |
|---|---|
| Senders in scope | Email platform (CRM lead); text tool (CRM lead); app messaging tool (app product lead, also used by membership); parties follow-up sequence in the CRM (parties manager); customer lists uploaded to 2 ad platforms (paid media freelancer, for holds and opt-outs only) |
| Service or marketing | Test: a message is service if the parent would still need it with any offer taken out. Service: booking confirmations, session reminders, safety forms, receipts, membership payment notices, complaint replies. The "book again" panel was removed from session reminders, which had made them marketing |
| Cap for each channel | Email: 5 promotional emails per person in any 7 days, broadcasts, flows and the parties sequence together, checked before every send by one shared condition named shared-email-cap. Owner: CRM lead. Set 3 November 2026 from the sign-up forms' promises and the complaints and unsubscribes after the busiest weeks |
| Texts and WhatsApp | 2 promotional texts per person in any 14 days across every flow, matching the opt-in line "up to 2 offer texts a fortnight". No WhatsApp. After 5 promotional texts with no reply, click or booking: service texts only. Owner: CRM lead; set 3 November 2026 |
| App messages | 4 promotional push or in-app messages per person in any 10 days, shared by marketing, product and membership; offers last in the queue. Owner: app product lead. Set 3 November 2026 from notifications switched off after the busiest weeks |
| Priority order and exclusions | 1 service messages; 2 party enquiry follow-up; 3 membership renewal offers; 4 abandoned booking; 5 new sessions and events; 6 membership "join now" sequence; 7 weekly newsletter. A person in the party follow-up receives no abandoned booking reminder or newsletter while it runs; a message that loses waits 2 days, then drops. Enforced by entry conditions and a flag set while the party follow-up runs; each workflow's rank sits in its inventory row |
| Quiet hours | No promotional texts or push between 10 pm and 7 am in the recipient's time zone, set once in each tool as a system rule, with service messages let through; held messages go at 7 am unless the offer would have ended, in which case they drop. Hard limits for the one market served confirmed by its legal review and recorded by the finance director on 3 November 2026 |
| Topic choices | Your bookings (service, on, operations lead); Membership payments (service, on, marketing operations lead); New sessions and events (promotional, counts, app product lead); Offers, including the weekly newsletter, abandoned booking reminders, membership renewal and "join now" offers (promotional, counts, marketing lead); Party ideas (promotional, counts, parties manager). The same 3 promotional topics are in the email preference centre |
| Holds while an issue is open | Open complaint: set by the complaints log, released when the log records it closed, owner the operations lead. Refund dispute: set by the finance team, released by the finance director on its outcome. Accident report: set by the park manager, released by the operations lead when the report is closed. Every hold reviewed after 21 days; reminders and receipts continue; review requests stop |
| Suppression and requests to stop | Unsubscribed, bounced, complained and do-not-contact checked before every send; members kept out of the "join now" sequence; any request to stop goes to the opt-out procedure, and broad words mean all marketing in every tool and both ad platforms |
| Inactive contacts | Email: no click or booking in 9 months, then one re-permission email, then suppression for those who do not answer. Texts: the rule above. App: none beyond the topics |
| Survey and research invitations | No more than 1 survey invitation per person in 3 months across marketing, operations and parties, recorded in the CRM; invitations carry no offer; people who opted out of marketing are not invited, as the finance director decided; invitations are not counted in the caps because the spacing rule already limits them |
| Changes, tests and review | Owner: marketing operations lead. Any change to a cap, the order or a hold reviewed by a second person, normally the marketing operations lead, and logged. 7 seed contacts, 1 for each of the 4 senders that message people and 1 per hold reason, run each month. The 25 most-messaged customers read each month. Last review 3 November 2026 |
The first monthly read of the most-messaged customers found the parties sequence sending past the email cap: it ran in the CRM and checked the shared condition only when a contact entered. The CRM lead moved the check to before each send, and the marketing operations lead reviewed and logged the change.
Common mistakes
- Rules with no owner. A cap or a hold that belongs to nobody drifts without anyone noticing. Name an accountable owner for each rule, not only for the document.
- A cap with no reason or date. The next person cannot tell a decision from a leftover. Write why each limit was set and when.
- A new tool, team or topic added outside the policy. Its first send runs on its own rules. Make the policy change part of the launch.
- Hard limits recorded without their source. Hours and consent rules differ between markets and change over time. Note who confirmed each legal limit and on what date.
- Quiet hours as a campaign setting. One campaign forgets. Make it a system rule in local time.
- Topics nobody owns. Promotions drift into the topic most people keep on. Give each topic an owner who approves what goes into it.
- A hold used for a request to stop. A hold expires; an objection never does. Send requests to the opt-out procedure.
- A policy no tool reads. Rules that exist only in the document change nothing. Build each as a condition and test it with seed contacts.
- Borrowed numbers. A cap copied from another business fits neither its customers nor yours. Set it from your own signals and write the reason.
Other documents that run a programme, among them the email marketing calendar and the email sign-up source register, sit in the templates library.
