Asher's Blog

There are businesses in which an average service is merely disappointing, and there are businesses in which an average service can genuinely damage the customer. If a restaurant serves an ordinary meal, I may decide not to return. However, if a payroll company processes salaries incorrectly, a payment provider holds money without explanation, or a compliance platform repeatedly misses important obligations, the customer does not simply have a disappointing experience. The customer inherits a new problem from the company that was supposed to remove the original one.

This is why being almost good can be more dangerous than offering no service at all, especially when you operate in a business built on trust. When you do not offer a service, the customer knows that the responsibility still belongs to them and can make another arrangement. Once you accept that responsibility, however, the customer changes their behaviour around your promise. They stop checking certain things themselves, organise other activities around your delivery and assume that the problem has been transferred into capable hands. If you then fail repeatedly, you have not simply failed to perform a task; you have broken the confidence that allowed the customer to depend on you.

The cost of being almost good is therefore much greater than the value of the feature that failed. It appears in the customer’s lost time, the emergency work their team must now perform, the explanations they must give to employees or regulators, and the anxiety of not knowing whether the next transaction will also fail. It also appears inside your own company through refunds, support calls, manual interventions, damaged morale and the increasing amount of energy required to persuade dissatisfied customers not to leave.

Trust changes the standard

Customers do not judge every product by the same standard because the consequences of failure are not the same. A minor inconvenience in an entertainment product may be tolerable, while the same level of unreliability in payroll, banking, healthcare, security, logistics or regulatory compliance may be completely unacceptable. The closer your product is to people’s money, livelihoods, safety, identity or legal obligations, the less room you have for repeated failure.

Imagine that an employer hands payroll to a provider because paying hundreds or thousands of employees manually has become stressful and risky. The employer is not merely purchasing a dashboard with buttons, reports and approval flows. The employer is purchasing the confidence that salaries will be calculated correctly and delivered when promised. Employees may have rent, school fees, loan repayments and family obligations tied to that date. If the platform fails repeatedly, the consequences spread far beyond the payroll administrator who clicked the button.

This is what founders sometimes miss when they become too focused on features. A feature can work during a demonstration and still fail to deliver the outcome for which the customer bought it. The customer does not care that ninety percent of the process worked if the remaining ten percent prevented salaries from arriving. In a trust business, the product is not complete merely because the software performed most of its technical functions. It is complete when the customer receives the promised outcome with the reliability the situation requires.

Being almost good is particularly costly because it encourages dependence without being dependable enough to carry it. The customer has moved away from the old method, trained employees on your system, transferred sensitive information, changed internal processes and perhaps persuaded senior management to approve the purchase. When your service then fails consistently, the customer must either continue carrying the risk or undertake another difficult transition. Your mediocrity has effectively imposed a switching cost on someone who trusted you.

Repeated promises are not improvement

One of the most damaging habits a company can develop is repeatedly telling customers that everything will become better next month. A sincere promise may calm a customer once, particularly when the company explains what happened and demonstrates what it is changing. However, when the same failure returns and the same promise follows it, reassurance gradually begins to sound like avoidance.

Customers can tell the difference between a difficult problem being solved and a company merely buying time. Genuine improvement leaves evidence. The number of incidents falls, response times improve, recurring causes are removed, communication becomes clearer, and the customer no longer has to report the same problem repeatedly. If none of those things changes, asking for more patience is not a recovery strategy. It is asking the customer to finance your learning with their time, money and reputation.

Feedback is also not a substitute for action. Many businesses have become good at requesting feedback because surveys, interviews and support tickets create the appearance of customer-centredness. Yet customers eventually notice when they have explained the same frustration several times without seeing any material improvement. At that point, asking for more feedback can become irritating because the problem is no longer a lack of information. The company already knows what is wrong but has failed to prioritise it.

When a customer reports a serious recurring issue, the correct response is not merely to apologise and forward the complaint to another department. Someone must own the problem from discovery to resolution, understand its operational and emotional consequences, identify the underlying cause and ensure that the fix prevents a recurrence. The customer should also be told, in plain language, what changed. That is how a company converts a failure into evidence that it can be trusted to learn.

Mediocrity creates hidden operating costs

Average service often looks cheaper from inside the company because the business sees only the cost it has avoided. Perhaps it postponed rebuilding an unstable system, hired less experienced people, retained an unreliable provider or delayed investing in customer support. On a spreadsheet, those decisions may appear prudent because expenditure remains low. What the spreadsheet may not immediately reveal is the cost of all the problems the company now manages every day.

Every recurring failure produces work. Customer-success employees must apologise, engineers must interrupt planned development, finance teams may process refunds, account managers must defend renewals, and senior leaders may become involved in issues that should never have reached them. If the founder is constantly called upon to reassure customers, the company is paying for poor quality with the founder’s attention, which is one of its scarcest resources.

There is also the cost of mistrust. A customer who no longer trusts a provider checks every transaction, asks for more reports, copies more people into emails and escalates matters more quickly. What used to require one interaction may now require five because the company must keep proving that it has done what it promised. The resulting bureaucracy is often blamed on a difficult customer, when it may actually be the customer’s rational response to repeated disappointment.

Reputation compounds this cost. In many African business communities, buyers speak to one another, particularly within the same industry. A dissatisfied customer may not publish a complaint online, but they may quietly warn several other businesses against using the product. This kind of reputational damage is difficult to measure because the founder rarely sees the opportunities that never arrived. A company may believe it has a customer-acquisition problem when it actually has a trust problem travelling through private conversations.

AI has raised the minimum standard, not removed responsibility

The age of artificial intelligence makes average work increasingly difficult to defend. Companies now have access to tools that can help them analyse support conversations, identify recurring incidents, test software, document processes, monitor systems, personalise communication and discover patterns that a small team might previously have missed. This does not mean that AI will automatically make every product excellent, but it does mean that founders have fewer excuses for ignoring obvious, repetitive problems.

The important question is not whether a company has added an AI feature to its product. The question is whether it is using every suitable tool to make the customer’s experience more accurate, reliable and effortless. An AI chatbot that responds instantly but cannot resolve anything may make bad service faster. Automated messages that continue assuring a customer while their money remains missing do not create trust. Intelligence should improve the substance of the service, not merely decorate its communication.

AI can help a company become more responsive, but leaders must still decide what standard they are willing to accept. They must define which failures are intolerable, invest in eliminating them and give employees the authority to act when customers are at risk. Technology can reveal the pattern, recommend the next action and sometimes execute it, but responsibility still belongs to the organisation that made the promise.

Reliability must be designed

Great service is rarely the result of asking people to try harder. It comes from building a system that makes the desired outcome repeatable. If a critical service depends on one exceptional employee remembering every detail, the business does not yet have reliability; it has an individual carrying an organisational weakness.

Designing reliability begins with identifying the promises customers consider essential. In payroll, these may include accurate calculations, secure handling of information, successful payments, timely statutory remittances and immediate visibility when something goes wrong. Once those promises are clear, the company can establish controls around them: automated checks, approval limits, reconciliations, alerts, fallback providers, clear escalation paths and named owners for serious incidents.

The company must also distinguish between ordinary errors and failures that threaten trust. A spelling mistake on a non-critical page should not receive the same response as a delayed salary payment or an unexplained debit. Teams need severity levels that reflect the customer’s risk, not merely the technical difficulty of the issue. Sometimes a problem that is easy for an engineer to fix is extremely serious for the customer, while a technically complex issue may have almost no immediate effect on them.

Reliability also requires honesty about capacity. A young company should not promise a service it cannot yet deliver consistently merely because a competitor offers it or a customer has requested it. Saying, “We do not support this yet,” may lose a transaction, but it preserves the customer’s ability to make an informed decision. Saying yes and failing at a critical moment may lose the customer permanently and damage the company’s reputation with others.

There is nothing wrong with beginning small. A startup can limit a service to a particular customer segment, transaction volume, geography or use case until its processes are mature enough to expand. What matters is that the promise matches the company’s actual capability. It is better to be exceptional within a narrow scope than broadly available but repeatedly unreliable.

Almost good can be a stage, but it cannot become a culture

Every product begins imperfectly. Startups learn from customers, uncover edge cases and discover weaknesses that were impossible to see before real usage. The argument against average service is therefore not an argument that a company must attain perfection before launching. If founders waited until nothing could ever go wrong, very little would be built.

The real issue is the company’s relationship with its imperfections. Does it treat recurring failure as urgent evidence, or does it normalise it? Does it improve after incidents, or does it become skilled at apologising? Does leadership know the problems customers encounter most often, and can it show that those problems are declining? A company may be temporarily imperfect while moving rapidly towards excellence. What it cannot afford is to make mediocrity part of its identity.

This distinction matters because excellence is not the absence of every mistake. It is the refusal to remain comfortable with preventable mistakes, especially those that can damage a customer. Excellent companies still experience incidents, but they detect them quickly, communicate honestly, reduce the consequences, learn from what happened and strengthen the system afterwards. The customer sees not only that the company failed, but also that it deserves another opportunity.

Founders must therefore create an internal intolerance for recurring failure without creating a culture of fear. Employees should be able to report problems early rather than conceal them, while also understanding that the same preventable issue cannot continue indefinitely without ownership. Blaming people for every error will make information disappear. Accepting every error without demanding improvement will make standards disappear. Good leadership protects openness and accountability at the same time.

Decide what your promise is worth

Every company makes a promise, whether it states it explicitly or not. The moment you accept a customer’s money, information or responsibility, you are telling them that they can rely on you for something. The seriousness with which you treat that promise will eventually define the value of your brand.

If the service is not ready, narrow it, delay it or decline to offer it. If a failure occurs, acknowledge it quickly, protect the customer and correct the cause. If the same complaint continues to appear, stop treating it as an isolated support issue and recognise it as a product or operational priority. Above all, do not keep borrowing trust with promises of improvement that your actions do not support.

Customers are often generous with businesses that are honest, responsive and visibly improving. They understand that young companies will encounter difficulties, particularly when solving complicated problems in difficult markets. What customers find much harder to forgive is being repeatedly exposed to the same risk while being told that a solution is always one month away.

Being almost good may produce revenue for a while because the product appears to work and the customer wants to believe the promise. Over time, however, the hidden costs begin to surface in churn, escalations, reputational damage, exhausted employees and a founder who spends too much time rescuing relationships. The business discovers that mediocrity was never cheap; its costs were simply delayed.

In a trust business, no service can sometimes be better than an unreliable service because absence does not pretend to carry responsibility. Once you offer to carry that responsibility, you must build the competence, systems and character required to carry it well. Customers do not need a company that is almost dependable. They need one that understands what their trust costs—and behaves accordingly.