Our packages

When it works: what breaks next, and in what order

Growth breaks a business before it breaks a campaign. The constraint does not disappear, it moves — and never to where you are watching.

Published on 7 August 2026 — Algeria Agency

The article beside this one explains how to make the machine work: the leaky bucket, the two numbers that multiply, content as stock, the form as a toll. It addresses a business whose acquisition is not yet producing.

This one begins where that one stops, and it deals with the case nobody discusses because it looks like good news: it is working. Cost per enquiry has fallen, volume has risen, and something is starting to grind somewhere else.

What grinds is never the campaign. It is the telephone nobody answers at two in the afternoon, the store room that runs out of the reference that is selling, the carrier who had not planned for this volume, and the cash position discovering that selling more consumes money before it returns any.

This article gives the usual order in which those things break, what to watch to see the next one coming, and why we will never tell you at what volume your business breaks.

Success moves the constraint, it does not remove it

A business is a run of stages of unequal capacity: attract, answer, sell, prepare, deliver, collect, repeat. At any moment one of them is narrower than the others, and it is that one which decides what the whole thing produces.

While acquisition is the narrowest stage, all the useful work is widening it — which is exactly what the article beside this one describes. When it stops being the narrowest, the useful work changes in kind and ought to change location, which almost never happens because attention stays where it paid off.

That lag is what produces the disagreeable period following a success. You carry on optimising a stage that is no longer the constraint, the acquisition numbers keep improving, and revenue stops following — because it is now set somewhere else.

The most reliable symptom of that moment is not a fall. It is a widening gap between two curves you were watching together: enquiries rise, sales stop rising as much. Nothing looks broken, and yet each additional dinar of advertising returns less than the one before.

So the right question then is not "how do we get more enquiries". It is "what at our end has stopped keeping up" — and the answer is nearly always a stage nobody was watching, because it had never needed watching.

The first thing to break is the telephone

In almost every Algerian business we have worked with, the first stage to give way is answering. Not selling, not delivering: simply replying to somebody who writes or calls.

The reason is arithmetic and nobody does the sum before it bites. A campaign that doubles enquiries doubles conversations, and a conversation takes the same time as before. What was absorbed by one person between other tasks becomes a full-time job without anybody having decided to create one.

The first sign is a response time stretching by a few minutes, then an hour, then half a day. None of those slips shows on an acquisition dashboard, and cost per enquiry carries on looking healthy — which is the trap: your favourite indicator does not move while the business loses customers.

The second sign is quieter and worse: the quality of replies drops. Somebody handling three times the messages replies more briefly, follows up less, forgets to call back. They are the same conversations and they turn into sales less often.

There is a simple measure to keep as soon as volume rises, and it needs no tool: the time between an enquiry arriving and the first reply, recorded by hand for a week. It is the first stage to watch because it is the first to break, and it is also the cheapest to widen.

Stock breaks at the moment it costs most

The second stage to give way is the store room, and it has a disagreeable property: it does not break at random. It breaks on the reference that is selling, at the moment it is selling — which is when running out costs the most.

The mechanism is simple. A campaign concentrates demand on a few items, and the turnover of those items becomes unrelated to the history your replenishment is set against. Stock sized on a normal year is wrong by the third week of a good campaign.

The consequence is twofold and the second is the more expensive. You lose the sale, which is visible; and you carry on paying for advertising that sends people to an unavailable product, which is not — unless somebody connects the store room to the ad account, which is nobody’s job.

The rule that avoids most of the problem is organisational rather than technical: whoever runs the campaigns has to know, every week, what is out of stock or close to it. A ten-line list sent on Sunday is enough, and it beats an integration that will not exist for six months.

The mirror rule is to cut advertising on an out-of-stock item the same day. It sounds obvious and is rarely done, because cutting a line that is "working well" takes a decision, and at that moment nobody wants to touch what is working.

Delivery: volume changes the carrier’s job

The third stage is shipping, and it breaks in a particular way: it does not give way, it degrades. The parcels keep going out, and they arrive later, more often on the second attempt, with more returns.

What changes is not the carrier’s competence, it is the regime. A carrier taking five parcels a day handles you at the edge of their round; the same carrier at forty a day has to build you into their planning, and moving from one regime to the other does not happen by itself because nobody announced it.

There is a conversation to have beforehand and it costs a phone call: tell your carrier the volume you expect, a month before you reach it. It is the only measure in this section, it is free, and it is the one almost nobody takes.

The second effect of volume is on returns and failed deliveries. At five parcels, a wrong address is an incident somebody resolves on the phone; at forty it is a queue, and that queue consumes the time of the same person who answers enquiries — so the first stage breaks a second time, by an indirect route.

That is the clearest case of a principle running through this whole article: the stages are not independent. Widening acquisition loads answering; loading delivery reloads answering. A constraint moved often comes back to the same place through another door.

Cash: selling more consumes money before it returns any

The fourth stage is not operational and it is the one that stops the most businesses: cash. Growing consumes cash, and selling more makes the position worse before it makes it better.

The order of the flows explains it in a sentence. You pay for advertising now, for stock now, for delivery now, and you collect later — later still if part of your sales is paid on delivery, and much later if you sell to businesses paying at thirty or sixty days.

The result is a counter-intuitive and perfectly ordinary situation: a business whose campaigns work, whose sales are rising, and which cannot fund the next replenishment. Nothing failed; everything succeeded faster than the cash conversion cycle.

The measure to keep is not a marketing indicator, it is the delay between the dinar spent and the dinar collected on an average order. Multiplied by the order rate it gives the sum that must be permanently available, and that sum rises in proportion to success.

The practical consequence is that a growth plan has to include the question of funding the cycle, however modestly. It is entirely reasonable to slow a profitable campaign deliberately because the cash cycle cannot follow — and that is a business decision neither we nor any supplier can take for you.

Enquiry quality falls before volume does

There is a degradation that is neither operational nor financial and that misleads a great many people: as a campaign widens, it reaches people further and further from your offer.

That is mechanical and nobody’s fault. The people most likely to buy are reached first; widening means going down into a less qualified audience. Volume rises while the proportion of useful enquiries falls, and the two movements can offset each other long enough for nothing to be visible.

The symptom appears among your people before it appears in the numbers. Whoever answers starts saying the enquiries are "less serious", and they are right — but since the number of sales has not yet fallen, the remark gets filed as a complaint rather than as a measurement.

It is also what makes cost per enquiry misleading past a certain volume. A less qualified enquiry sometimes costs less to obtain while being worth much less, so the indicator improves while the situation worsens. That is the point at which to stop steering by cost per enquiry and start steering by cost per sale, which presumes somebody at your end returns the information about what enquiries became.

None of this justifies stopping widening. It justifies knowing what to expect: widening is paid for in quality, the only question is how far the trade stays good, and that is read in the one figure the supplier does not have.

The channel that works becomes the channel you depend on

When a channel works, everything pushes towards putting more into it: it is profitable, it is measured, it is comfortable. And as its share rises, a disagreeable property appears without anybody deciding it — your business becomes dependent on a platform it controls nothing about.

The risk is not theoretical and it does not take the form of a catastrophe. It takes the form of a rule change, a gradual rise in auction prices, an account suspended for three days pending a check, or a season when costs double without explanation.

It should be measured rather than feared, and it fits in a fraction: what share of your enquiries would disappear if that channel stopped tomorrow. Past a certain share the answer is "most of them", and that sentence should trigger a decision rather than a shrug.

What the article beside this one calls content as stock is the cheapest answer to this problem, and this is the moment to reread it: what has been published carries on producing when advertising stops, which is precisely the property advertising does not have.

The rule that works is not to diversify early — that is waste while the first channel is still learning. It is to diversify when the first channel works, which is exactly when nobody wants to, and to do it with a small share rather than a second full budget.

Prices: the moment to stop being the cheapest

There is a point in almost every period of growth where the problem stops being demand and becomes margin. You sell more, you work more, and the result rises more slowly than the activity.

The cause is almost always the same: a price built for low volume and a light structure, kept after volume revealed costs that price did not contain — response time, returns, failed deliveries, preparation, and the cost of the advertising itself.

The honest calculation is done on one order, not on a month. Take an average sale and subtract everything attributable to it: the product, preparation, delivery, the share of advertising, the share of returns, and the human time it consumes. What is left is what the sale actually returns, and the result is often surprising.

This section is not an invitation to raise prices, and it is not a supplier’s place to say what your prices should be. It says something narrower: a price is a decision with a date on it, and a business that has tripled its volume while keeping its launch-year price has not decided, it has inherited.

It is also the limit of what acquisition can do for you, and it should be said plainly. No campaign, no page and no content corrects a sale that does not earn enough to pay what it costs. Past that point our work makes the problem larger rather than smaller.

What to stop doing yourself

There is a stage of growth where the owner becomes the constraint, and it is the hardest to see because it looks like commitment rather than a bottleneck.

The sign is not tiredness, it is waiting. Things are waiting for your approval, your call, your decision on a price, your check of a message. Each is short and together they set the pace of the business, exactly as a narrow stage sets what a chain produces.

The useful sort is not by importance but by frequency. What comes back every day has to leave your hands even if it is important; what happens once a month can stay even if it is minor. That is the opposite of instinct, which delegates the minor and keeps the frequent.

Three things almost always go first and return the most time: replying to routine messages, preparing orders, and chasing deliveries in progress. None needs a rare skill and all consume daily hours.

And one thing does not go, whatever the volume: the decision about what you sell, at what price, with what promise. That is what the article on team arrangements says, and it holds at every level — it is the only decision nobody else can take for you.

The page’s ceiling, and why it comes last

The article beside this one says a page that converts better makes everything else cheaper. That is right, and what it does not add is this: that lever has a ceiling, and you reach it after widening everything else.

The reason is that a page’s conversion rate is not solely a property of the page. It depends on who arrives on it, and as a campaign widens towards a less qualified audience the same page converts worse without having changed by a comma.

That produces a frequent and expensive confusion: people conclude the page has degraded and rebuild it, when it is the audience that changed. The work is real, it is well done, and it shows nothing — because it addresses a stage that is not the constraint.

The second ceiling is simpler: at some point the page does what it can. The remaining gains become small, tests need more traffic than before to show anything, and the effort spent there would sit better on answering, on stock, or on price.

The way to know where you are is to compare two periods at constant audience rather than two periods full stop. If conversion falls while the traffic source is changing, the page is not at fault; if it falls at constant source, it is. That distinction is the only thing that prevents rebuilding a page that had nothing wrong with it.

Why we do not say at what volume it breaks

The question we are asked is reasonable: from how many enquiries a day does it start to jam. We have no figure to give and know of no honest one, for a reason worth explaining because it applies to all data of this kind.

A business that meets a constraint fixes it. It hires somebody to answer, it changes carrier, it raises its safety stock. That is the normal and desirable behaviour, and it has a consequence for measurement: at the moment you ask a business what broke, it describes a repaired state.

In other words, the subject responds to the observation, and the lag always runs the same way: the available data describes the previous constraint, never the current one. An average taken across businesses that have grown reports the bottleneck they have already solved — precisely the one you do not need warning about.

A more ordinary fact sits on top of that: the threshold depends on your weakest stage, and nobody — you included — knows which that is until it gives way. Two businesses identical in revenue break at very different volumes depending on whether they have one more person on the phone or three more days of stock.

What we look at instead is measured at your end, this week: in front of which stage is there a queue. Unanswered messages, orders picked but not shipped, returns waiting, a replenishment invoice waiting on a collection. The stage with a queue is the constraint, today — and in six months it will be another one, which is exactly why a published figure would be no use.

What we do, and what we refuse to do here

Every month we look at where the queue is, not only at what acquisition produced. It is a two-minute question in the meeting and it changes what gets discussed: once the constraint has left the campaign, carrying on optimising the campaign is work that shows up nowhere.

We send the out-of-stock list before spending, we cut advertising on an unavailable item the same day, and we give warning when the volume ahead is going to change your carrier’s regime — one call, a month before.

We refuse to say at what volume your business will break. The previous section explains why that figure cannot exist honestly, and an invented one would have a precise effect: it would have you watching the wrong place with confidence.

We refuse to raise a budget when the constraint is elsewhere. If enquiries wait half a day before being handled, spending more produces enquiries that will also wait — cost with no result, invoiced by us, which is the worst possible service to a client whose campaigns are working.

And we will tell you when the problem is your price, knowing that setting it is not our trade and that the conversation is unwelcome. Past a certain point no campaign corrects a sale that does not cover what it costs, and our work then makes the problem larger. Better to hear it from a supplier losing budget by saying it than to discover it over a full year.

Frequently asked questions

At what volume does it start to jam?

We do not give that figure. A business meeting a constraint fixes it, so all available data describes a repaired state and reports the previous bottleneck rather than the current one. Look instead at where the queue is this week: unanswered messages, orders picked but not shipped, returns waiting.

What usually breaks first?

Answering. Doubling enquiries doubles conversations, and a conversation takes the same time as before. First-response time stretches by a few minutes, then an hour, and none of it appears on an acquisition dashboard — while cost per enquiry carries on looking perfectly healthy.

Our sales no longer follow our enquiries. What should we look at?

Two things, in this order. The queue: is there a stage where things are piling up. And quality: widening a campaign reaches an audience further from your offer, so volume rises while the useful proportion falls. At that point steer by cost per sale rather than cost per enquiry.

Should we deliberately slow a profitable campaign?

Sometimes yes, and it is a business decision no supplier can take for you. Growing consumes cash — advertising, stock and delivery are paid before collection — and a campaign that works can drain a cash position without anything having failed.

When should we diversify channels?

When the first one works, which is exactly when nobody wants to, and with a small share rather than a second full budget. Diversifying too early is waste while the first channel is still learning. Measure the dependence as a fraction: what share of your enquiries would disappear if that channel stopped tomorrow.

Our page converts worse than before. Should we rebuild it?

Not before comparing two periods at a constant traffic source. A widening campaign brings a less qualified audience, and the same page converts worse without having changed. Rebuilding a page that had nothing wrong is real work, well done, on a stage that is not the constraint.

Where we come in

What breaks first is almost never acquisition: it is the person who answers, the item out of stock, or the parcel waiting on a shelf.

We stop the advertising on the day an item is no longer available.
We send you the shortages before the spending date, never after.
We count the waiting enquiries before proposing a larger budget.

We will not tell you how many people to hire: we can show you the queue, and deciding is a trade that is not ours.

See the Pro page Start with the free audit

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.

Request the free audit Write to us