Decline

When revenue starts sliding, the explanations inside a company multiply. Sales says a competitor cut prices. Marketing says the ad budget was trimmed. Engineering says the features customers asked for still aren't built. All three sound reasonable, so the meeting ends the way it always does, with a little more money going somewhere.

Every one of those explanations assumes the same thing: fix the cause and the numbers come back. Some declines don't come back. Document tools that customers installed on their own servers stopped attracting new buyers once tools that open from a web address arrived, no matter how many features got added. Fax software disappeared along with the offices that still had fax machines. In neither case did revenue fall because the product got worse.

Decline exists to separate those two situations by name. A slump you can fix and a drop that won't reverse call for money to move in opposite directions — push harder on the first, spend less and collect what's left on the second. With a name for it, one question opens that discussion: is this product in decline?

Fewer new customers in a quarter isn't enough to call it decline. The drop has to hold across several quarters and survive an attempt to spend the numbers back up. Three signals usually get read together.

1. Acquisition doesn't recover. You run the same ad budget as last year and get 60 percent of the signups, while customer acquisition cost (CAC) — the money spent to bring in one customer — climbs on that same channel from $60 to $145. CAC rises in maturity too, but there the total customer count holds. In decline, departures outrun arrivals and the total itself falls.
2. The customers left behind change character. With almost nobody new arriving, the base is people who have used the product for years. They rarely leave, and they rarely touch new features either. Satisfaction scores often go up as a result, because everyone who was unhappy has already gone. Churn inside that group is low enough that a dashboard can look healthy for a year after growth has ended. Read those scores on their own and you'll misjudge the state of the product.
3. Price moves get weak responses. Cut pricing by 20 percent and new contracts barely move. If price were the obstacle, that cut would show up. When it doesn't, people aren't declining to buy because it costs too much — they've stopped looking for this kind of product at all.

The cause is usually one of two things. Something else does the same job better, or the job itself went away. Asking churned customers what they use now tends to separate them. A competitor's name means the market is still there and only the position moved, so there is something to win back. "We don't do that anymore" means there's no market left to win. Self-hosted document tools losing to web tools is the first kind of answer, and fax software is the second.

Nothing follows decline automatically. Unlike the move from introduction to growth, no condition carries the product forward. The team picks one of three paths.

1. Harvest — stop feature work, keep the smallest crew that can maintain the product, and go on collecting from the customers who stay, who at this point leave very slowly. Drop a six-engineer product to one engineer and revenue can halve while the money left over goes up.
2. Rebuild — remake the product and send it back to an earlier stage. Not a few added features but a fresh answer to who you sell to and what you sell them, which means confirming product-market fit — the state of a built product actually matching demand in a specific market — from scratch.
3. Shut down — end the service. That process is called sunsetting, and it usually runs in order: stop new sales, honor existing contracts to expiry, provide data export, turn off the servers. Customers need the export step early enough to act on it.

Two questions decide which path. Does the money the remaining customers pay exceed the cost of keeping the product alive, and could the people attached to it do something bigger elsewhere? If both money and staffing have room, harvest runs for years. If the money is already short, the choice narrows to rebuild or shut down. Postpone the call and nobody chooses, which leaves six engineers attached to the product by default.

"This isn't a soft quarter, it looks like decline. Three quarters pointing the same way."

One odd number would be noise; a structural change needs a different response.

"I'm against adding four more engineers to a product in decline."

Don't put people where the return won't come back.

"Let's call it decline and move to maintenance mode, but honor the contracts we have."

Stop investing without cutting anyone off abruptly.

Pagedeck still sells a self-hosted version built in its early days. Eleven customers stay on it because their security rules block outside services. At last quarter's leadership meeting Grace raised the fact that the product had signed zero new contracts in two years, and Mia put its renewal rate next to its maintenance payroll. It was 4 percent of company revenue with two engineers on it.

The decision in the room was harvest. New sales closed, one engineer stayed for security patches and incident response, and the other moved to the web product.

The most common misreading treats decline as evidence that a team failed. The numbers go down, so it looks like somebody's fault. But the four stages of a life cycle describe what the market around a product does, not how a team performed. Nobody's mistake made fax machines disappear. Read it as failure and the person closest to the product stops saying the word, and a call delayed by a year costs real money.

The other misreading hears "decline" and assumes immediate shutdown. The name invites that, but this stage often runs longer than any other. A product with its costs cut can lose half its customers and still return more cash than before, which keeps it running for years. Shut down on arrival and you throw away the revenue those years would have paid, and you push out customers who have nowhere to move yet. The moment to stop isn't the day the stage starts — it's the day maintenance costs more than the product brings in.