Before diagnosing: suppress the person and read the consent record
The message arrives as a reply to a newsletter, a complaint through customer service or a note passed on from a shop counter, and it asks for two things: stop the mail, then find out how the address came to be on the list. As general practice, keep that order, because the diagnosis takes time and the person should not receive another message while it runs. The FTC's CAN-SPAM guide and the ICO's guidance on direct marketing each describe honouring an opt-out promptly, which is a reason to stop the mail first and does not decide what this message is.
For the stop, the marketing opt-out request procedure is the owner, and this page does not repeat its steps. As general practice, treat a never-signed-up message as a request that procedure covers: its trigger is any person who asks to stop, objects to marketing or withdraws a permission, and where the scope is unclear it suppresses across all marketing meanwhile. The person therefore goes onto the suppression listA list of addresses or contacts excluded from every send, such as unsubscribes, hard bounces and spam complaints. until the data protection owner decides what follows. Whether such a message counts as an objection, a withdrawal or neither, and whether any send was lawful, are questions for that owner. This page answers neither, and it is not legal advice.
Then read the record. A consent recordThe stored evidence of a marketing permission: the channel, the wording the person saw, the date and the source of each opt-in, kept with the contact record. is the evidence kept for one permission: where the address arrived, when, under what wording, with which choices and with what confirmation. The consent map names the system that holds it for email, which may not be the platform that sent the message. The GDPR, as the site's guides summarise it, asks that consent be demonstrable, and a record is what makes that possible. What the business relies on for each address, including the lawful basisOne of the six grounds the GDPR accepts for processing personal data, such as consent, contract or legitimate interests, chosen for each purpose. recorded for the purpose, is the data protection owner's decision, so the diagnosis supplies facts to it and does not make it.
The reply. As general practice, keep it short and factual. Say that marketing to the address has stopped. Name where and when the address arrived only if the record shows it. Do not ask the person to stay, do not argue the point and do not describe any handling of opt-outs other than the one the procedure sets out. Not everyone writes: a person who does not recognise a message may report it as spam instead, which the spam complaint rateShare of delivered emails that recipients marked as spam. counts, so each reply prompts a look at the other addresses that came by the same route.
Diagnosis table
Read the record first, then match what it shows to a row. The rows are routes by which an address reaches the list without a choice the person remembers; they are not ranked against each other, and one person can fit two of them. The five whys method suits each row: ask why the address is on the list, then why that route kept no evidence of a choice.
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| The record is complete and names a source, but the address had not been mailed for a long spell before this send | The record's date against the previous send to the address, and the wording shown at that source | A real sign-up the person no longer remembers because the list went unmailed for a long time | As general practice, keep the person suppressed, answer from the record, and restart the rest of the list in groups sorted by record quality |
| The address belongs to someone else or looks invented, and the record carries no confirmation time | The opt-in method of the source form, and whether the record holds any confirmation | An address typed in by someone else, or by a bot, under single opt-in | Confirm addresses from exposed forms before they join; hold unconfirmed addresses from that source |
| Contacts from a form that should confirm carry no confirmation time, or confirmations changed after an update | Where each form posts, whether the confirmation message arrives, and when the change happened | A confirmation step lost when a form, platform or integration changed | Rebuild the step, point the form at the current account and test it with a real sign-up |
| A customer who bought says they never agreed to marketing, and the record holds no refusal offered at purchase | Whether the checkout offered a refusal, whether the record shows it, and whether the mail matched what was bought | A soft opt-in relied on without a recorded chance to refuse at purchase | As general practice, record the source, wording, choice and time at checkout, and hold marketing to those addresses until the data protection owner decides |
| The source is a counter, a phone call or a paper sheet, and the person says they gave an address for a receipt, booking or card | The source value, who typed the address, whether the spoken line or sign is on file and whether a confirmation followed | A till, counter, phone or paper sign-up with no marketing choice of its own | Give marketing its own question, keep the wording with the record and confirm addresses that staff typed |
| The source is a checkout, a download, an account or a competition, and the person remembers that, not a newsletter | The wording shown, and whether the marketing box was pre-ticked or tied to the terms, the download or the entry | Marketing consent bundled into something else: a pre-ticked box, a download, an account or a competition entry | Offer an unticked, separate marketing choice and move addresses from the bundled source to re-permission or suppression |
| The source reads import, sync or event, or is empty, and the person has no dealings with the business | The date and origin of the file or sync, what its people were told and whether the contact had a record before it crossed | Contacts brought in from a file or another system: an event or partner list, a sales export or a CRM sync | Hold the file in a staging segment, keep contacts with no record out of the marketable set, then re-permission or suppress |
Four neighbouring diagnoses start elsewhere. The unsubscribe spike diagnosis starts from exits piling up on one send, flow or source; the pages on emails going to spam and a high bounce rate start from placement and bounces; unsubscribed contacts still receiving email starts from a person who opted out and is mailed anyway. This page starts from the person who writes that they never signed up.
Fix, verify and prevent each cause
The complaining address belongs to the opt-out procedure; the real work is the other addresses that came by the same route.
A real sign-up the person no longer remembers
The record here is complete, yet the person is not necessarily wrong to be surprised: a list left unmailed through a long silence carries consent records that may have been lost and subscribers who have forgotten joining, and the dormant list restart tutorial owns that situation.
Fix: as general practice, keep the person suppressed and reply from the record. Do not mail the rest of the list all at once: sort addresses by the quality of their record and by what the people last did, mail the strongest group first, and send a re-permission messageA short email sequence asking contacts with doubtful consent or long inactivity to confirm they still want email; non-responders are mailed no further. to the groups whose record is partial or old, so that anyone who does not confirm is suppressed.
Verify: after the first group is mailed, read the replies and exits from that group before widening the audience; more never-signed-up replies from that group can point to a weaker record than it looked, and the group goes back to re-permission or suppression.
Prevent: keep to the rhythm the sign-up promised, so that no group goes unmailed long enough to forget, and write the promise at the form so that a later message can be connected to it. A line such as "Get our newsletter" commits the business to nothing specific; a sentence that names the content and its rhythm can be tested against what arrives.
An address typed in by someone else, or by a bot
Under single opt-in a form accepts whatever is typed into it, so mistyped, invented, borrowed and bot-submitted addresses enter the list as a real subscriber's does, as the opt-in comparison explains. Double opt-inA sign-up process in which the subscriber confirms through a link sent to the address given. adds a confirmation click from the address's owner, proof that the mailbox is real and that its holder acted deliberately a second time. Such an address shows an open form as its source and no confirmation time.
Fix: set the method source by source, as the comparison's decision table does, and keep the validation it lists on any form that stays on single opt-in. As general practice, treat the backlog like any doubtful group: hold unconfirmed addresses from an exposed form in a staging segment, send one re-permission message and suppress those who do not answer.
Verify: type a test address you control into the form and confirm that it does not become marketable until its confirmation link is clicked, and that the welcome message waits for the click.
Prevent: write the method for each source, with the reason, in the register. Decide a period after which an address that has not confirmed is dropped. And make the first message easy to leave, so that someone who joined by mistake goes quietly instead of reporting the mail.
A confirmation step lost when a form, platform or integration changed
A change can drop the confirmation step without anyone deciding to. The platform migration tutorial lists forms and the confirmation step among the things a move can lose, and a form still pointed at the old account is one such loss. In the record, a step that was not rebuilt shows as sign-ups with no confirmation time; if the message stopped arriving, people who meant to join do not finish, which reports read as people choosing not to join.
Fix: point each form and integration at the current account, rebuild the confirmation message and the cut-off for unconfirmed addresses, and keep the consent wording as it was unless the change is deliberate, recording any new wording from its first day. As general practice, send a confirmation or re-permission message to addresses that arrived between the change and the repair with no confirmation time.
Verify: push a genuine sign-up through each form and trace it. The contact appears with its source and time, the confirmation message lands in an inbox, and the welcome sequence begins only afterwards.
Prevent: add the confirmation flow to the checks that follow any change to a form, a platform or the sending domain.
A soft opt-in relied on without a recorded chance to refuse
The soft opt-in is a UK term, described in the ICO's guidance, for an existing customer, similar products and a refusal offered at the sale and in each message. It is decided at checkout, needs evidence of the refusal offered and is narrower than anyone who has bought; whether the business relies on it for an address is the data protection owner's decision. The complaint arises when the record shows no refusal offered at the sale, or the mail concerns something unlike what the customer bought.
Fix: the guide to sign-up flows says what to keep (where the sign-up happened, what the person was shown, what they chose and when) and keeps the marketing choice separate from accepting the terms. Put the refusal into the checkout and record it. As general practice, hold marketing to addresses from that checkout until the record shows what was offered and the data protection owner has decided.
Verify: buy as a test customer through the live checkout and read the record that results. It should hold the wording shown, the choice and the time, and the next marketing message should offer the refusal again.
Prevent: give the checkout its own row in the register, carrying the basis as the data protection owner decided it, and re-read that row when the product range changes, because a new line may no longer count as similar to what customers bought.
A till, counter, phone or paper sign-up with no marketing choice of its own
At a counter the question can read as part of the payment, a spoken offer says less than a written one, and a notebook entry shows nothing of what the customer was told. The in-person sign-up tutorial owns this cause. The record points at it when the source reads counter, phone or paper, with no wording on file and no confirmation.
Fix: move capture to a form the customer completes on a tablet or screen, with the marketing question kept apart from the receipt, the booking and the card. Where staff type an address, apply the tutorial's read-back and confirmation rule before any marketing. For addresses already taken on paper with no wording on file, hold them in a staging segment and re-permission them.
Verify: complete a sign-up at each counter as a customer would, and one by phone as staff would, then read the records: each should show its source, location, wording and time, and the staff-typed one a confirmation.
Prevent: let each counter's form fill in the source value from a fixed list, keep each wording version and the date it started, retire the notebook, and tell staff that a spoken request to stop is an opt-out request for the procedure to log.
Marketing consent bundled into something else
A pre-ticked box, a download treated as a subscription, an account that carries marketing along with it and a competition entry that doubles as a sign-up share one fault: the person did one thing and was subscribed to another, which the list building and consent guide lists among the things to avoid. The record shows a download or an entry as the source, with wording that does not mention marketing or a box ticked before the person saw it.
Fix: offer marketing as its own unticked box, kept apart from the terms and describing the mail honestly, and treat a resource request and ongoing mail as two choices. For addresses already admitted through the bundled source, send one re-permission message, which turns an unknown group into a smaller known one, and suppress those who do not confirm.
Verify: go through each form as a visitor. The marketing box should be unticked and separate from the terms, and submitting without ticking it should not add the address to the marketing list.
Prevent: re-read the wording whenever a form, an offer or the checkout changes, and file each wording in the register with the date it started.
Contacts brought in from a file or another system
Three routes sit under this cause: a partner's list, an event list or sales export, and a CRM sync. The CRM contact import tutorial owns the first two: before anything loads, it asks where the file came from and what the people on it were told, and the trade show tutorial adds that a scan alone does not amount to marketing permission. For the sync, the CRM and marketing platform sync guide leaves contacts that lack a consent record outside the set that can be marketed to. The record points here when the source reads import, sync or event, or was never written.
Fix: park the file in a holding segment, as the list building guide advises for any import. Change a sync rule so contacts without a consent record do not cross into the marketable set until one exists. Then re-permission the groups whose record supports it, as under the first cause, and suppress the rest.
Verify: a contact with no record stays outside the marketable set after the next sync, a test row from the file lands in the holding segment, and the check against the suppression list runs before the load.
Prevent: write the file's real origin and what its people were told as the source value instead of a generic "import", and hold each file in staging before it joins regular sends.
Prevention: a source register that can answer the next complaint
The remedy that outlasts any one fix is the email sign-up source register, which gives each source a row. The template owns the fields; what this page adds is the use, because a complaint becomes a lookup instead of an investigation.
| What the person says | What answers it in the register | Cause it separates |
|---|---|---|
| "I have no idea who you are" | The source, its owner and the wording shown at the time | A forgotten sign-up, or one from a bundled or imported source |
| "That was not my address" | The opt-in method and the confirmation record | A typed-in address from a single opt-in source |
| "I only gave my email for a receipt" | The offline capture row, the spoken line and the source value | A counter, phone or paper sign-up without its own choice |
| "I did not ask for offers after buying" | The checkout row, the refusal offered and the basis decided | A soft opt-in without its record |
| "I did not confirm anything" | The opt-in method, and the confirmation message recorded for that form | A confirmation lost in a change |
Two habits keep the register useful.
- Write the row before the source goes live. A new form, pop-up, counter script, partner promotion or import is a new row first. A source with no row is the first thing to find when a complaint cannot be traced.
- Sample the records on a routine. The email list hygiene procedure checks a sample of recently added contacts for a recorded source, date and wording, and suppresses any contact that has none.
A basket-maker's week of complaints (illustrative)
Fradelvo Basketry
Fradelvo Basketry, an invented basket-making workshop with a shop counter, a stall at craft fairs and an online shop, receives 3 replies in one week from people who say they never signed up. Each is suppressed the same day under the opt-out procedure, and its consent record is read. The first address came from a counter sheet that staff keyed in, with no marketing choice separate from the receipt. The second came from a list of 210 badge scans imported after a fair, with no wording on file. The third came from a pop-up form on the website whose confirmation step was lost when the form tool was changed 9 weeks earlier. Counter addresses wait in a staging segment until each has its own marketing question and a confirmation email. The fair list goes into a staging segment and receives one re-permission message, after which 64 people confirm and the other 146 are suppressed. The pop-up's confirmation step is restored and tested with a real sign-up, and the 38 addresses that joined without a confirmation time get a confirmation message. Each source's wording and method is written into the register, so the next reply can name the source from the record.
The troubleshooting collection lists the neighbouring diagnoses, each built around a symptom that a sender can observe.
Common mistakes
- Diagnosing before suppressing. A message that arrives while the diagnosis is under way can add to the complaint; as general practice, stop the mail first, under the opt-out procedure.
- Treating the reply as a verdict on the send. Whether a message is an objection or a withdrawal, and whether a send was lawful, belong to the data protection owner.
- Defending the address with a receipt, a booking or a card. None of them is a marketing choice, so none of them answers the complaint.
- Citing a confirmation as lasting interest. A click shows a deliberate step by the address's owner at one moment; it does not show that the interest lasts.
- Fixing the one person. The route that admitted them may have admitted others, and they are found from the source on the record.
- Deleting the person to close the matter. Suppress instead; a deleted contact can come back.
