A trained pear tree carries new growth while showing where an older branch was removed. Original editorial painting created for this article.
After launch, repeat use, customer departures and the work behind each sale can help you decide what to improve, change or stop.
By Arbab Naseebullah Kasi, Chief Executive Officer - CEO, Feel Worldwide Foundation Inc.
A business can keep receiving orders while becoming harder to sustain. Customers buy once and disappear. A popular package requires so many exceptions that the owner spends evenings correcting it. Another service attracts little attention but brings dependable repeat work. The sales total puts these different stories into one number.
Once a business is operating, the question becomes more demanding than whether someone will buy. Will the right customers keep receiving enough value to return? Can the team deliver that value reliably? Will the money left after delivery support the work and the obligations around it?
Answers may justify a better version of the current offer, a substantial change in direction or the retirement of one product. None requires pretending that the earlier work was worthless. The purpose of learning is to let experience affect what happens next, while treating customers, workers and suppliers fairly through the change.
Follow the people who have already bought
New sales can conceal a shrinking relationship with existing customers. Separate the two before deciding that growth proves the business is healthy.
Start with a cohort: a clearly defined group whose experience you follow over time. This might be customers who first paid in January, organizations that began a maintenance contract in one quarter or households that received their first delivery during a particular month. Keep the original group identifiable. Do not quietly replace customers who leave with new arrivals.
Choose a meaningful measure of return. For a monthly service, renewal may matter. For a consumable product, another purchase within a reasonable replenishment period may be useful. For a durable item, a second purchase within 30 days could be an unreasonable expectation. Successful use, low avoidable returns and appropriate referrals may tell you more.
Here is a deliberately hypothetical subscription example, not a company result or Foundation case study. Begin with 100 paying customers and $10,000 in monthly recurring revenue. After a defined period, 75 of those original customers remain. Cancellations remove $1,500 of the original monthly revenue, reduced plans remove another $500, and expansion among remaining customers adds $3,000. No new customers enter this calculation.
Customer retention is 75 divided by 100, or 75%. Revenue retained before expansion is $8,000 divided by $10,000, or 80%. With expansion included, revenue from the same original group is $11,000, giving net revenue retention of 110%.
All three numbers describe the same imagined business. Revenue from the group has grown even though one quarter of its customers have left. That could reflect a promising move toward customers who need more of the service. It could also conceal an offer that is failing many smaller buyers and becoming dependent on a few large accounts. The calculation tells you where to investigate.
Slack's April 2019 registration statement provides a real example of careful metric definition. It reported net dollar retention of 143% as of January 31, 2019, comparing monthly recurring revenue from the same paid customers across twelve months. Its measure included expansion and deducted contraction and attrition, while excluding new paid customers. That figure was not the percentage of customers who stayed. It is a historical company measure, not a target every small business should copy. Slack's original registration statement.
Compare cohorts at the same age. January buyers observed for six months cannot fairly be compared with June buyers observed for one month. Note discounts, acquisition channels, seasonality and changes in customer mix. Show counts beside percentages: losing two of ten accounts and losing 200 of 1,000 both produce 20%, but support different levels of confidence about the underlying pattern.
Find out what a departure actually means
A cancellation is an event. Its explanation requires more work.
Some customers leave because the task is complete. Others lose their budget, move away or change their own operations. Some never learn how to use what they bought. Others encounter poor service or discover that the offer solves a less important problem than expected. Treating all departures as a request for a lower price can send the business in the wrong direction.
Invite recent departures to describe their actual experience, without making feedback a condition of cancellation. Ask what they hoped to accomplish, what happened during their last use, when the service stopped being useful and what they chose instead. Speak with repeat customers and people who reduced their purchases too. Keep a record of who was invited and who responded. A handful of unusually enthusiastic or angry respondents cannot represent everyone.
Netflix's 2011 experience shows why separating explanations matters. Its October shareholder letter reported negative domestic net subscriber additions of 0.81 million in the third quarter. Management attributed the upset principally to DVD-related pricing changes and, to a lesser extent, the proposed and then cancelled DVD rebranding. The letter also acknowledged moving too quickly and explaining the change poorly. These are management's contemporary explanations, not a controlled estimate of each decision's effect. Netflix's October 2011 shareholder letter.
Earlier that month, Netflix had announced that DVD service would remain on Netflix.com, emphasizing the simplicity of one website, account and password. The company reversed that part of its plan. Netflix's October 10 announcement.
The useful lesson is the specificity. A change can advance an internal strategy while adding work for the customer. Before redesigning an offer, understand whether the problem lies in price, usefulness, delivery or the transition itself. Several can be true at once.
Put the hidden work beside the revenue
Customers may love an offer that is exhausting the people delivering it. That is evidence about the business model, even when complaints are rare.
Track preparation, delivery, administration, corrections and support. Notice work that happens outside recorded hours and costs that the owner absorbs personally. Examine which requests create repeated exceptions, which promises depend on one person and whether staff can raise these problems without being blamed for disappointing customers.
Consider another explicitly hypothetical example. A small recurring service has 20 clients paying $300 monthly, producing $6,000 in revenue. Materials, travel and other direct cash costs total $1,800. The owner spends 160 hours delivering and administering the service. Valuing that time at an illustrative $25 per hour adds $4,000, leaving only $200 before other overhead, taxes and financing costs.
The $25 figure is a planning assumption, not a wage recommendation or an accounting rule. It helps reveal whether the apparent margin depends on nearly free owner labor. Avoid counting the same labor twice if it is already included in the recorded costs.
Now suppose the owner proposes a narrower service for 18 clients at $350 each. The forecast is $6,300 in revenue, $1,500 in direct cash costs and 120 hours at the same assumed rate, leaving $1,800 before the same excluded costs. The arithmetic improves despite fewer clients. The customer acceptance, reduced workload and delivery quality remain unproven assumptions.
This is where a useful improvement can be quite ordinary: fewer custom options, clearer ordering deadlines, better onboarding, a different service area or a price that reflects the work. Record the effect on customers as well as the hours saved. Removing a time-consuming accessibility arrangement, safety check or contractual obligation is not an acceptable shortcut to a cleaner spreadsheet.
Make your explanation testable
Research offers support for a more disciplined way of changing course, with limits worth keeping visible.
Arnaldo Camuffo, Alessandro Cordova, Alfonso Gambardella and Chiara Spina studied 116 Italian startups in a randomized trial with 16 observation points over about a year. Their paper, published in Management Science, examined training founders to frame and test business assumptions scientifically. The authors reported improved performance and a greater likelihood of moving to a different idea among the treated entrepreneurs. The setting was early-stage entrepreneurship, not a representative sample of established businesses everywhere. The original startup study.
A 2024 replication and extension by Camuffo and colleagues pooled 759 firms across four randomized trials, including the original 116. It found an increase in idea termination and evidence of a more selective pattern of major changes. In its full-sample analysis, the intervention increased the probability of pivoting radically once by 8.3 percentage points. The findings did not endorse an endless sequence of pivots. The replication and extension.
These studies test particular training interventions and measured outcomes. They do not tell a specific owner which offer to keep, establish a universal number of pivots or guarantee that a change will improve profit. Applying the approach after launch still requires judgment about the business, its customers and the risks of experimenting with live services.
Write a prediction before making the change. For example, if repeated cancellations appear to follow difficult onboarding, predict what a clearer onboarding process should improve, over what period and without what unacceptable trade-off. Decide how the comparison will be made. Keep other changes limited where feasible, and note differences between the groups you compare.
When numbers are small or random assignment is inappropriate, the result may remain suggestive. Customer accounts, delivery records and repeated observations can still improve a decision. Honest uncertainty is more useful than presenting a busy month as proof of causation.
Choose the scale of change the evidence supports
Improvement keeps the central offer and fixes a specific weakness. Customers need the service, but booking is confusing or delivery is inconsistent. A focused repair may be enough.
A larger change becomes plausible when the evidence challenges the intended customer, the main problem being solved or the way the business gets paid. Perhaps a tool designed for occasional individual use proves more useful to organizations with recurring needs. That possibility deserves investigation, including whether the company can serve those organizations responsibly. A few inquiries alone do not establish a new market.
Retiring an offer becomes reasonable when its continuing demands cannot be justified, or when another service uses the same limited resources more productively. Stopping one offer can preserve the wider business. It can also protect customers from another cycle of promises that the team cannot reliably deliver.
Ask what would change the decision in either direction. If nothing could persuade you to stop, the review is ceremonial. If one disappointing week can trigger a complete reinvention, the evidence has too little time to become informative.
Count the cost of getting from here to there
A better future model may still be unaffordable to reach. Separate the ongoing economics from the transition budget.
List development, retraining, contract changes, customer support, refunds, unusable stock, temporary duplicate systems and time away from ordinary sales. Distinguish cash that must be paid from accounting write-downs and from the value of people's time. Check when the costs arrive, not just their total.
Shopify's 2023 logistics exit makes this distinction concrete at a much larger scale. Its financial statements record the sale of its logistics businesses, mostly to Flexport, and an impairment of US$1.34 billion on those sales. That accounting impairment reflected carrying amounts exceeding estimated fair value; it should not be described as a US$1.34 billion cash payment to exit. Shopify's 2023 financial statements, Note 4.
Shopify's public explanation emphasized concentrating attention on its core mission and acknowledged the impact of the associated changes on employees. That explanation belongs to management. The record documents a consequential change and its costs; it does not prove that every business should narrow its scope or reduce staff. Shopify's May 2023 announcement.
For a smaller, hypothetical decision, a $4,000 cash transition cost divided by an expected $500 monthly increase in net cash contribution gives a simple eight-month payback. At $250 a month, it becomes sixteen months. This rough calculation ignores financing costs, taxes, timing within each month and the risk that the improvement never arrives. It is a prompt for a cash forecast, not a promise of recovery.
Money already spent cannot be recovered simply by continuing. Equally, money available today is not automatically spare. Preserve the ability to meet existing obligations and obtain appropriate financial or legal advice where contracts, employment or solvency are involved.
Leave a clear record for the next decision
Before acting, write a short decision note that a colleague could understand without hearing the founder's enthusiasm. Record the offer being reviewed, the customer groups observed, the period covered, the important findings and the plausible alternative explanations.
Then describe the proposed change, the spending and workload limits, who is responsible, what customers will be told and when the evidence will be reviewed. Use a review period long enough for the relevant buying or renewal cycle. Identify conditions that require an earlier pause, such as a safety problem, serious service failure or an inability to meet obligations.
If an offer will close, plan its final delivery rather than merely removing it from the website. Check notice requirements, prepaid balances, refunds, warranties, data access and supplier commitments. Offer a workable transition where possible, and be candid where no equivalent replacement exists. Workers and customers should not have to discover a decision through a missing service or an unanswered message.
After the change, compare the outcome with the prediction. Include work that moved elsewhere instead of disappearing. Record any customer group that benefited less than the headline average suggests. Keep the unsuccessful findings available so the next version of the business does not repeat the same expensive lesson under a new name.
The next version may be more focused, easier to use, less demanding to deliver or smaller than originally imagined. Its strongest justification is a traceable connection between what happened and what changed. That gives an owner something more dependable than confidence alone: a business whose decisions can be examined, corrected and explained.
If your organization wants to explore a practical contribution to entrepreneurship learning or responsible livelihood development, submit a partnership inquiry to Feel Worldwide Foundation Inc.. Describe the proposed contribution, intended participants and scope so its fit and feasibility can be reviewed. To offer financial support, visit the Foundation's donation page and review the available giving options and terms.
