Digital marketing
A mailing list ages, and nobody sees it
A list does not degrade visibly: it grows while the share that actually receives shrinks. What to do each quarter.
A mailing list is presented as an asset that accumulates, and the figure describing it only ever goes up. That is what makes its degradation invisible: every month it contains more addresses and, proportionally, fewer and fewer people who actually receive anything.
The two movements are independent. Addresses are added by collection; they are lost through job changes, abandoned mailboxes, closed services, and people who stopped opening without unsubscribing. Only the first movement is counted.
This article is about maintenance: addresses that no longer exist, addresses that exist and no longer read, the cleaning that frightens people, the segmentation that actually works, the list somebody hands you, and reactivation. The companion article covers the channel itself — what it does here, collection, the spam folder, frequency, what the law requires — and none of that is repeated here.
One figure appears in it, and it is about the structure of Algerian businesses: it changes what a professional list actually contains.
Two movements, only one counted
The number shown by a sending tool is a cumulative total, and a cumulative total does not fall. It adds everything that came in and removes only what was explicitly removed — unsubscribes and invalid addresses. Everything else stays, including what no longer works.
The quantity that matters is different and nobody displays it: the number of addresses that opened or clicked at least once in the last six months. It is the only one corresponding to what the word “list” means to whoever owns it.
The gap between the two installs itself silently and is cumulative. A list built over three years contains a significant share of addresses that no longer serve any purpose — not because anybody made a mistake, but because three years have passed.
The practical consequence is to adopt a second figure and follow it. The first is for reassurance, the second is for deciding. Showing them side by side once a quarter is enough to make visible a phenomenon that otherwise never gets noticed.
What an Algerian list actually contains
The usual advice on professional lists assumes company addresses: a name, a dot, a domain the company owns. It draws rules from that — segment by company, expect an address to die when the person changes job, target by role.
That assumption does not hold here, and the structure of the economy says why. The commercial register counts 2,419,913 registered operators, of which 2,145,557 are natural persons — roughly nine in ten. Most of the businesses you are writing to are one person, and that person uses a free address.
Two consequences follow, running in opposite directions. The good one: the address does not die when the business changes, because it belongs to the person rather than the company — an Algerian professional list therefore decays more slowly than an equivalent list elsewhere.
The bad one: segmenting by company means nothing, and tools that infer the company from the address domain return the same answer for almost everybody. The only division that works is behaviour — who bought, who opened, who asked for a quotation — and that is section four.
CNRC, position at 10 December 2025
Addresses that no longer exist
Part of a list no longer corresponds to anything: mailbox deleted, domain expired, service closed. Sends to those addresses come back, and that return is recorded by the tool under a technical name nobody looks at.
Not removing them costs more than the obvious waste. Mail services observe the proportion of sends that fail, and a sender with a notable share of messages going to dead addresses is treated as a careless one — which degrades delivery of the messages aimed at valid addresses.
That is what makes this section important rather than administrative: keeping dead addresses is not neutral, it damages the reception of everything else. You do not clean a list out of tidiness, you clean it so that messages arrive.
The action is mechanical and takes ten minutes a quarter: open the hard-bounce report and delete the addresses concerned. Hard bounces have to be distinguished from soft ones — full mailbox, server unavailable — which do not justify deletion and resolve themselves.
The segmentation that actually works
Segmentation is presented as sophisticated work and nearly always fails for the same reason: it is done on criteria you do not possess. Sector, company size, role — that information was never collected, and guessing it produces a false division.
The division that works rests on what you know for certain because you saw it: this person bought, this person asked for a quotation and did not follow up, this person signed up and never did anything. Three groups, derived from facts, with no additional data entry.
Those three groups want different messages and that is what makes the exercise pay. An existing customer receives information about what they already own; an unfollowed quotation receives what was missing from the decision; an inactive subscriber receives a message recalling why they signed up. The same message sent to all three is optimal for none.
A fourth distinction is nearly always useful here and costs little: the wilaya, where it was given. It avoids announcing an in-store event to somebody a thousand kilometres away, which is the fastest way to make an otherwise interested person unsubscribe.
Cleaning, and why it frightens people
Removing addresses from a list is something almost nobody dares do, for an understandable reason: the number falls, and that number was presented to somebody as a result. Nobody wants to announce that the list has shrunk by thirty per cent.
The opposite reasoning is the right one, though, and it is checkable. An address that has opened nothing for two years will produce nothing, and its presence degrades delivery for all the others — meaning that keeping it costs real readers rather than preserving potential.
Two populations do have to be distinguished, and cleaning treats them differently. Invalid addresses are deleted without hesitation. Valid but silent addresses are only removed after a reactivation attempt, which is the next section, because a share of them will come back.
How the operation is presented matters for it to happen at all. It is a fall in the displayed number and a rise in the number that counts, and showing both side by side turns bad news into a result — which is exactly what the first section’s second figure is for.
The reactivation message
Before removing a silent address, it is worth sending a message designed for that, and this message resembles no other. Its objective is not to sell, it is to obtain a reaction — any reaction, including an unsubscribe.
What works is short and direct: recall when and why the person signed up, say what they will receive, and explicitly offer to let them go. That last element looks counter-productive and is what makes the message work: it turns a non-response into a decision, and a decision, whatever it is, cleans the list.
The proportion that comes back is small and that is not a failure. The point of the exercise is not to recover the silent, it is to separate those who still exist from those who do not — and to do it through an act of theirs rather than a decision of yours.
This message is sent once, and those who do not reply are removed. A series of three reactivation follow-ups is a contradiction: it insists at people who have just been shown not to be listening, and it produces exactly the careless-sender signal described above.
The list somebody hands you
At some point somebody will offer a list: a file from a trade fair, an export from a former employer, a purchased directory. The offer always comes with the same argument — these are professionals in the sector, it cannot do any harm.
It does harm, and the mechanism is measurable rather than moral. People who do not remember signing up mark the message as spam, and that report is the heaviest signal a mail service records. A single operation of this kind is enough to durably degrade delivery of your ordinary messages.
There is also a legal consequence, which the companion article develops: prospecting requires consent, and a purchased file contains none. We mention it without detailing it here, because the practical argument is already sufficient.
The rule is therefore simple and has no useful exception: you write to people who gave their address to your business, knowing what they would receive. A list without that property is not a list, it is a risk disguised as an asset.
Six months without writing
Prolonged silence is treated as an absence of risk — you send nothing, so you disturb nobody. It has a precise cost, though, and it is the most frequent situation among small businesses using this channel in bursts.
Two things degrade during that silence. Recipients’ memory: somebody who signed up eight months ago and received nothing no longer recognises your name, and their first reflex on seeing your message is to report it. And sender reputation, which is built by regularity and lost through absence.
That produces a counter-intuitive situation: the first campaign after six months of silence is the most likely to go badly, and it is precisely the one sent to the whole list because there is something important to announce.
The precaution is not to restart with a mass send. Begin with the most recent and most active addresses, a few hundred, then widen once delivery is verified. It is two extra days, and it avoids burning the list at the moment it is finally needed.
The sending address
The address in the sender field is not a presentation detail: it is what recognition and part of delivery rest on. It deserves a decision, and that decision is taken once and for all.
Sending from a free address is possible and is paid for in two ways. Recognition is weaker, because nothing distinguishes your message from a personal one. And mail services apply stricter rules to bulk sends from those domains, which they do not control.
Sending from the company domain is preferable and requires a technical configuration the sending provider documents: three records to add wherever your domain name is managed. It is an hour of work, once, and it separates correct delivery from random delivery.
What must not be done is changing the sending address. Every change restarts recognition among recipients and reputation with mail services, and businesses do it without thinking — a different provider, a new tool, a “contact” address replaced by an “info” one.
Backing up the list
The list is the only asset of this channel, and it lives at a provider. That is acceptable as long as a copy exists elsewhere, and the copy almost never exists.
The ways of losing access are banal and involve no fault: a subscription not renewed during a trip, an expired payment method, an account created in the name of somebody who has left, a service that closes. None of those situations is spectacular and all produce the same result.
The export takes two minutes and is filed where your other documents are. One file a quarter is amply sufficient, and it is worth checking once that it contains what it should — the address, the sign-up date and the source, failing which you are exporting a list you will not be able to justify.
That last column is the one that gets forgotten and the most important in the long run. Knowing where each address came from is what allows you to answer a challenge, clean intelligently, and avoid finding yourself in two years with a file nobody knows was collected or acquired.
Taking over an inherited list
The ordinary case is a list whose origin nobody knows: a few thousand addresses, in a tool whose access is in a former employee’s name, with no sign-up date and no source.
The first question is not technical: do we know where these addresses came from? If the answer is no for all of them, the list must not be used as it is, whatever its size — for the reason developed in section seven, which is practical as much as legal.
What is recoverable in that case is the identifiable part: the addresses matching existing customers your management system knows. Those are legitimate, their origin is known, and they constitute a smaller and genuinely usable list.
For the rest, the only honest route is a single reactivation message framed as re-establishing consent: recall the context, explicitly ask for confirmation, and delete everything that does not reply. It shrinks the list considerably and it is the only path that does not mortgage the delivery of every subsequent send.
What we do, and what we refuse
What we take on is bounded: the double count from the first section, configuring the sending address, the division into three behavioural groups, the quarterly cleaning and the reactivation campaign that precedes it.
We do not send to a list the client did not collect, including when it is provided in good faith and contains exactly the right people. It is not only a question of rules: one send of that kind durably degrades delivery of every subsequent message, including transactional ones, and that damage is not quickly repairable.
We do not keep addresses inactive for two years in order to maintain a number. It is a refusal that is sometimes uncomfortable, because the number has been reported to somebody. We show both figures side by side — the cumulative total and the actives — precisely so that the conversation is about the right one.
Finally, we do not promise an open rate. That measure has become unreliable for technical reasons the companion article explains, and it is the wrong question in any case. What we track with you is the number of replies and orders attributable to a send, which is smaller, slower to obtain, and genuinely yours.
Frequently asked questions
Our list is growing. Is it healthy?
The displayed number is a cumulative total and does not fall: it adds arrivals and removes only unsubscribes and invalid addresses. The quantity that matters is the number of addresses that opened or clicked at least once in the last six months, and no tool puts it forward. Track both side by side each quarter — that is what makes visible a degradation that otherwise never gets noticed.
Should we segment by sector or company size?
No, and particularly not here. Roughly nine in ten registered operators in Algeria are natural persons, so a professional list is made of personal addresses: inferring the company from the domain returns the same answer for almost everybody. Divide on what you have seen — bought, asked for a quotation without following up, signed up and did nothing — plus the wilaya where it was given.
Why delete addresses that cost nothing?
Because they cost the delivery of everything else. Mail services observe the proportion of sends that fail, and a sender with a notable share going to dead addresses sees reception degrade for their valid ones. You do not clean out of tidiness, you clean so that messages arrive.
How should silent addresses be handled?
With a single reactivation message, before removing them. Recall when and why the person signed up, say what they will receive, and explicitly offer to let them go — that last point is what makes it work, by turning a non-response into a decision. Once only: three follow-ups insist at people who have just been shown not to be listening.
We have been offered a file of professional addresses.
Refuse, and the practical argument suffices before the legal one: people who do not remember signing up mark the message as spam, and that report is the heaviest signal a mail service records. A single operation of this kind durably degrades delivery of your ordinary messages, including transactional ones.
We have not written for six months. Can we restart?
Yes, but not with a mass send. Two things degraded during the silence: recipients’ memory, so they no longer recognise your name and report it, and your sender reputation. Restart with the most recent and most active addresses, a few hundred, then widen once delivery is verified. Two extra days, and you do not burn the list at the moment it is needed.
Where we come in
The gap between the total and the share that opened anything in six months is your diagnosis. It widens every quarter with nothing to announce it.
- We take that count again every three months and send you both numbers.
- We remove addresses silent for two years, telling you before we do.
- We add the signup date and the source to addresses carrying neither.
A list under five hundred addresses needs no upkeep at all: export it twice a year and go and do something else.
Read next
Email marketing: three cases where it works here
A list of addresses is the only portable marketing asset. What remains is being in one of the three cases where this channel works in Algeria.Your book is your best customer list: the renewal calendar
A signed policy gives you a guaranteed appointment a year in advance. Almost nobody uses it.Approaching twenty named businesses: the method without advertising
If you can name your possible customers, you should not be buying them. How to build the list, what to write, how many times to follow up, and when to stop.
Let us talk about your project
A free audit, no commitment: we look at your online presence and tell you what is holding it back.