The revenue was the cheap part

Ask an owner what it cost to stop selling something and you expect a number with a currency sign in front of it. Three of them gave the number and then said it was not the cost.

Money & World asked business owners, founders, product managers and freelancers what they had stopped selling, how long they kept it going after suspecting it was not working, and roughly what share of revenue it represented. Then what actually happened when it went. Twenty four people answered. Three answered from their own books rather than in general terms, and they were ending three unrelated things: a legacy maintenance service inside an information technology firm, a done-for-you video production line inside a software company, and half of a freelancer's income.

All three size the revenue they lost. None of them calls it the expensive part.

Nearly two years to decide, 90 days to leave

Answer: Kuldeep Kundal says it took him nearly two years to retire a service he had offered since the start of his career, that it was responsible for about 10 percent of his revenues by then, and that he tried cheaper support tiers and more junior maintenance staff before accepting that the line had to end.

Kuldeep Kundal is founder and chief executive of CISIN. He puts the objective of the exercise somewhere other than the revenue line, and he puts it first.

"There are more important things than revenue loss when it comes to sunsetting an out-of-date service; regaining the brain space previously occupied by technical debt is the main objective of the practice."

The delay he describes has a specific cause, and it is not indecision about whether the line was failing.

"I had a hard time letting go of a service I had been offering from the very beginning of my business career, and it took me nearly two years to come to my decision. Although by that time this particular service was just about responsible for 10 percent of my revenues, I was afraid to drop it because I had grown accustomed to keeping the revenue stream coming from long-term customers."

What filled those two years was a series of attempts to make the service survivable.

"While technology complexities grew at the time, and my margins were decreasing with every month due to rising engineering costs, I attempted to retain the service through the introduction of less expensive support tiers and younger maintenance specialists. However, being faced with real situations at that stage meant that we still had to rely on top-notch engineers for troubleshooting again and again."

Read that sequence again, because it is the mechanism rather than an anecdote about one firm. The cheaper tier and the junior staff were not a decision to end the line. They were a decision to keep it, at a lower price, and each attempt bought another interval in which the senior engineers were still being pulled in. The line was not costing him 10 percent of revenue. It was costing him the people he had hired to do something else.

He gave clients 90 days of notice. The reaction was not the one he had prepared for.

"Although we anticipated a wave of departures from our company among clients, only 20 percent of them shifted to a more costly service - our cloud services. At the same time, 60 percent of our customer base chose to trust other companies that we suggested to them, while only 20 percent expressed their dissatisfaction with the situation. However, we soon realized that this outstanding 20 percent accounted for the overuse of our services."

The dissatisfied fifth were the heaviest users. That detail recurs across this pool and it points the same way each time: the volume of complaint a business hears when it ends something is generated by the customers who consumed the most of it, which is a different population from the one that was paying for it.

"Therefore, apart from a temporary decline in revenue, the most important outcome of the decision was the uplift in our team's mood and the ability to concentrate on AI and SaaS innovations rather than on the same old technical debt issues we had been solving for years."

Six weeks to know, three months to act

Answer: Runbo Li says he suspected a done-for-you video production service was broken within six weeks of offering it, that it was maybe 15 to 20 percent of revenue, and that he kept it going about three months after he knew, telling himself the delivery could be systematised.

Runbo Li is chief executive of Magic Hour. His account is unusual in this pool because he dates the moment of knowing precisely, and it is early.

"Early on, we sold custom AI video production as a done-for-you service. Brands and creators would pay us to produce finished videos using our tools. It felt like easy revenue at the time, maybe 15-20% of what we were bringing in. But I suspected it was broken within six weeks of offering it. Each project ate 4-8 hours of hands-on time, the feedback loops were brutal, and it pulled our attention away from building the self-serve platform that was already growing faster organically."

Six weeks in, the case was closed. The line then ran for another three months.

"I kept it going for about three months after I knew it wasn't working. The reason was simple: those clients were paying real money, and when you're pre-seed with no salary, killing revenue feels reckless. I kept telling myself we'd systematize it, templatize the delivery, make it scalable. We never did. Every project was a snowflake."

The same shape as Kundal, at a twentieth of the timescale. In both cases the interval was not occupied by doubt about the diagnosis. It was occupied by a plan to rescue the line, and the plan was never executed in either firm.

What happened at the end is the part worth sitting with.

"When we finally cut it, here's what actually happened: almost nothing. Two clients asked if we'd reconsider. One was mildly annoyed. The rest didn't even respond to the email. That silence was the most informative data point of all. It told me the service wasn't solving a deep enough problem for them to fight for it. They liked it, but they didn't need it."

Then he prices it, and the price is not in dollars.

"The cost of finding out wasn't the lost revenue. It was the three months of divided focus I'll never get back. Every week I spent on a custom project was a week I didn't spend on the thing that actually scaled."

The line that is not being cut at all

Answer: Kharla Denura says SEO backlink outreach represented just under half her total income for roughly two to three years while she built a legal trust accounting practice, that she has not ended it but has stopped taking new engagements, and that the cost she did not anticipate was prospects arriving confused because her own profiles still advertised the service.

Kharla Denura is an independent legal trust accounting specialist trading under her own name. Her case is the one that does not fit the shape of a decision at all.

"SEO backlink outreach was the service that came first and helped sustain the freelancing services while legal trust accounting was still growing. For roughly two to three years I ran both, and during that period SEO represented just under half of my total income."

The reasoning she gives is about the structure of the work rather than about the revenue.

"The decision to stop was not sudden. Legal trust accounting engagements are longer, more specialized, and more stable. A law firm cleanup or ongoing monthly reconciliation is not a one-off project. The client relationship continues, the work compounds in value, and the niche is specific enough that the right clients are not shopping on price."

There is no cut in her account. There is a tap being closed.

"I stopped pursuing new SEO engagements entirely. The clients I had been working with long term stayed, and I continue supporting them, but no new ones have come in and none are being sought. In practice the service is winding down."

And then the cost, which is one nobody would budget for.

"The consequence I did not fully anticipate was the confusion from still carrying SEO backlink outreach as a secondary description on my professional profiles. Prospects would arrive unclear on what I actually specialized in. That friction came directly from not cleaning up the public picture fast enough as the shift happened, and it is something I am actively correcting now by pushing legal trust accounting to the front of everything."

That consequence was still visible on the day this piece was reported. Her own site was retrieved on 25 August 2026, and the page title it served described her as a legal trust accounting bookkeeper and SEO backlink specialist, carrying both services at once. She is describing a lag between what a business has stopped doing and what the world can still see it offering, and the lag was checkable.

What the accounts can see, and when

Two instruments deal with a deteriorating line of business, and they take opposite views of when it becomes real.

American securities law attaches to knowledge. Item 303 of Regulation S-K, at 17 CFR 229.303(a), says the discussion and analysis "must focus specifically on material events and uncertainties known to management that are reasonably likely to cause reported financial information not to be necessarily indicative of future operating results or of future financial condition." The operative requirement sits a little further down, at 229.303(b)(2)(ii), and it is short.

"Describe any known trends or uncertainties that have had or that are reasonably likely to have a material favorable or unfavorable impact on net sales or revenues or income from continuing operations."

Known. Not decided, not announced, not executed. On that test Li's obligation would have begun at week six, in the same period he was telling himself the delivery could be templatised.

The accounting standard runs the other way. IFRS 5, in the text the European Union has endorsed, will not let a business present an exit until the exit is committed. Paragraph 6 asks whether the carrying amount will be recovered principally through a sale rather than through continuing use. Paragraph 8 then defines what makes a sale highly probable, and the bar is a list of things a business must already have done: management committed to a plan, an active programme to locate a buyer initiated, the asset actively marketed at a reasonable price, completion expected within one year, and actions that "indicate that it is unlikely that significant changes to the plan will be made or that the plan will be withdrawn."

None of these three sold anything. All three closed a line, and paragraph 13 deals with that case separately.

"An entity shall not classify as held for sale a non-current asset (or disposal group) that is to be abandoned. This is because its carrying amount will be recovered principally through continuing use. However, if the disposal group to be abandoned meets the criteria in paragraph 32(a)-(c), the entity shall present the results and cash flows of the disposal group as discontinued operations in accordance with paragraphs 33 and 34 at the date on which it ceases to be used."

The same paragraph says in terms that abandonment includes assets "that are to be closed rather than sold". So for a line that is shut rather than sold, the accounts recognise nothing until the day it stops. Kundal's two years, Li's three months and the whole of Denura's wind-down are ordinary trading in every set of accounts that would ever be drawn up.

The gap between the two instruments is the interval all three of these people are describing. One requires the business to say what it knows. The other refuses to record anything until the business has finished acting. Between them sits a period, in this group running from three months to two years, in which somebody knows a line is failing, has not decided to end it, and is spending senior attention on a rescue.

Where this is thin

None of these three businesses files under either instrument. Item 303 binds registrants filing with the Securities and Exchange Commission. IFRS 5 in this form binds companies reporting under EU endorsed international standards. A private information technology firm, a pre-seed software company and a sole trader are outside both, and nothing here suggests any of them was obliged to do anything.

The documents are doing a different job. They show that the formal system has already thought hard about when a failing line becomes reportable, arrived at two different answers, and left the same gap between them that these three describe living in. What neither instrument has is a line for the thing all three call the actual cost. Divided focus does not appear in a statement of comprehensive income. There is no disclosure item for the senior engineers who were still being pulled onto a service that had already been decided against.

Three accounts are three accounts. They are not a sample, the figures in them are self-reported and appear in no public dataset, and each contributor is describing a business whose books nobody else has seen. What they have in common is worth naming anyway, because it is not the thing the question asked about. The question asked how much revenue the line represented. All three answered that question, and then told us it was the wrong one.