Bring a Spoon.

The strategy presentation was beautiful. The typography suggested competence. The arrows suggested movement. A photograph of people pointing at a screen suggested collaboration, although nobody in the photograph appeared to have opened an actual spreadsheet.

The room approved the direction. Someone used the word “transformational.” Someone else requested the slides. Then the strategy encountered customers, employees, operating constraints, and the extremely inconsiderate behavior of the real world.

This is where an old English saying becomes useful: “The proof of the pudding is in the eating.” A pudding does not become good because the cook explains it persuasively. At some point, someone has to take a bite. The same basic obligation applies to a business strategy, a new technology, a marketing campaign, and a brand promising that it cares.

I like the saying because it can be delivered politely while asking a remarkably difficult question. You can admire the preparation, appreciate the effort, and still ask whether the thing does what it is supposed to do. The answer matters more than how confidently everyone agreed before the test.

The Cambridge Dictionary definition keeps the meaning close to practical experience. That simplicity is valuable. It brings the conversation back from “What do we want people to believe?” to “What can people actually experience?”

For this article, the United Kingdom offers a rich setting: maritime testing at Greenwich, engineering associated with Scotland and the Midlands, ceramic experimentation in Staffordshire, and modern public-service design. These are particular histories and practices, rather than proof that an entire population shares one personality.

The question running through them is universal. When a claim leaves the room where it was created, what happens next?

What the pudding is actually proving

In the proverb, “proof” concerns a test. The useful information arrives through trying the thing. The more compressed version, “the proof is in the pudding,” preserves the familiar sound while losing some of the action that makes the longer saying so good.

Without the eating, we could spend the afternoon staring at the pudding and complimenting its positioning. Perhaps it has a differentiated narrative. Perhaps it is the category leader in stakeholder-aligned custard. We still do not know whether anyone wants another spoonful.

A historical discussion of the phrase traces related wording through early English texts and records its appearance in William Camden’s 1605 work. That is evidence of a long linguistic life, not a reason to assign a single inventor or imagine that every historical use had exactly the same context.

The enduring insight is that descriptions and experiences are different forms of information. Descriptions help people understand what to expect. Experiences let them discover whether the expectation was justified.

Businesses need both. A useful product that nobody understands has a communication problem. A persuasive description attached to a disappointing product has a delivery problem. Solving one does not automatically solve the other.

A customer may believe your website enough to inquire, believe the conversation enough to buy, and believe the delivery enough to return. Those are separate moments of judgment. The first conversion is only one part of the relationship.

The saying also invites us to be specific about purpose. A pudding made for one occasion can be judged against that occasion. Likewise, a strategy should be evaluated against the problem it was meant to address. “It generated attention” is incomplete if the actual goal was profitable, repeat business from a defined customer group.

Before asking whether something worked, we need an honest account of what working would mean. Otherwise the definition will tend to drift toward whichever result looks most flattering in the quarterly review.

Greenwich: the clock has to work at sea

Royal Museums Greenwich preserves an extraordinary example of a claim meeting difficult conditions: John Harrison’s work on marine timekeeping. The problem was not simply to build a clock that looked impressive in a workshop. Reliable timekeeping could help navigators determine longitude, and the instrument needed to function during a voyage.

The museum’s history of Harrison’s timekeepers describes H1’s 1736 sea trial to Lisbon and back. Decades later, H4 underwent a voyage to Jamaica with Harrison’s son William in 1761. The trials, revisions, and disputes show how much more was involved than announcing a clever invention.

The museum’s H4 collection record identifies the watch as completed in 1759 and explains the development from Harrison’s earlier work toward the smaller timekeeper. The eventual form emerged through a long process of working on the problem.

For a business, the question is wonderfully direct: what is your version of taking the clock to sea? Where are the conditions that will reveal whether the idea works outside its most accommodating environment?

A website may look excellent on the designer’s large monitor. The customer may be using a phone in bright daylight, with one hand occupied and very little patience. A process may work when its creator is present to explain every exception. The employee using it on an ordinary afternoon may have none of that support.

A sales pitch may be persuasive to people who already understand the company. An unfamiliar buyer may not know which problem the company solves. Internal familiarity makes many things seem obvious that are not obvious at all from the outside.

The test conditions therefore matter. If you only test with people who know the answer, you may be measuring how kindly they compensate for the design. If you test with your easiest customer, you may learn something useful about that customer while learning very little about the broader audience.

Harrison’s history also reminds us that proof can become a contested social process. People must agree on what a test establishes, how it should be repeated, and what further evidence is reasonable. Even good results do not interpret themselves.

In contemporary business, settle as much of that as possible before the result arrives. Agree on the relevant conditions and the decision the evidence will inform. This reduces the temptation to declare the trial perfect when it supports your idea and inadequate when it does not.

From Scotland to industrial practice: understand where the energy goes

James Watt offers another useful route into the subject. The Science Museum describes the Scottish engineer’s improvement to steam-engine efficiency through a separate condenser. Condensing steam outside the working cylinder allowed the cylinder to remain hot, reducing the repeated heating and cooling that wasted energy. The museum’s account of the separate condenser also emphasizes the years of work required to turn the concept into an industrial machine.

The Science Museum Group’s surviving 1765 model gives the idea a material presence. It is a reminder that a consequential innovation can begin with a specific understanding of an inefficient mechanism.

That is a more useful starting point than deciding that everything needs to move faster. Speed applied to a wasteful process can produce waste at an impressive rate. The organization becomes very busy, which is helpful if busyness was the goal and unfortunate if it was meant to accomplish something else.

In business, energy is lost in less visible ways. Information is entered repeatedly. Customers explain the same problem to three departments. Teams wait for decisions that have no clear owner. Sales makes a promise that operations discovers after the contract is signed.

None of this necessarily appears as a dramatic failure. It can become the ordinary friction everyone has learned to work around. The workaround becomes familiar, then essential, then somebody’s full-time responsibility. Eventually a meeting is created to manage it.

A practical adviser asks where the effort goes and what useful result it produces. Which steps create value? Which exist because another step is unclear? Which delay protects the customer, and which delay simply preserves an old habit?

This is part of the thinking behind digital transformation and process improvement. Technology can help once the problem is understood. Buying a new system before understanding the process can make the inefficiency more durable and give it an annual renewal date.

The lesson I draw from Watt is to investigate the mechanism. “Our people need to be more productive” is a large claim. “Our people spend an hour reconstructing information that already exists elsewhere” identifies work we can examine and improve.

Staffordshire: keep the experiment book

In Staffordshire, Josiah Wedgwood’s ceramic work brings experimentation, product quality, and commercial judgment into the same conversation. The V&A describes nearly 5,000 glaze and body trials associated with refining his creamware, along with notes and ingredient lists recorded in an experiment book. Its introduction to Wedgwood also discusses customer-facing practices such as replacing goods broken in transit.

I like that combination. There is the discipline of improving the object, and there is the discipline of considering what happens when the customer receives it. A beautifully made item that arrives broken creates a problem the elegance of the original design cannot solve by itself.

Modern businesses sometimes separate those responsibilities too neatly. Product has its definition of success. Marketing has another. Fulfillment has a third. The customer, having failed to study the internal reporting structure, evaluates the whole experience.

The experiment book matters because memory is a surprisingly flexible business system. After a result arrives, people often remember having expected something close to that result. An assumption that was once confidently asserted becomes something “we always knew might need refinement.”

Write down what you expect before trying it. Record what changed, what stayed the same, and what happened. The purpose is to make learning possible, including when the result is inconvenient.

This need not become bureaucracy. A short record can preserve the question, the method, the observation, and the next decision. That is enough to make the second attempt more informed than the first.

There is also a humane consequence. When the organization keeps a usable record, the knowledge does not depend entirely on one person remembering everything. Someone new can understand why a choice was made. Someone returning to the problem can avoid paying for the same lesson again.

The history of industrial innovation is larger and more complicated than a parade of celebrated individuals. These examples do not establish an uncomplicated national success story. They give us specific practices worth examining: test under relevant conditions, understand the mechanism, record the learning, and consider the customer’s whole experience.

The boardroom’s missing question: what would show us we are wrong?

I have been in board meetings and senior leadership conversations where a CEO presents an idea that appears obviously terrible, and nobody says anything. The silence can be remarkably polished. Everyone looks engaged. People begin discussing timing and implementation as though the central assumption has already passed inspection.

In that situation, I speak up kindly and politely. I bring the evidence and experience that make me concerned, explain the risk as I see it, and leave room for information I may be missing. Challenging the idea is part of caring about the leader and the business.

The pudding saying gives that conversation a useful focus. We can move from a contest between personal opinions to a discussion of what would count as evidence. What do we expect customers to do? Why? What have we observed? What could we test before making a larger commitment?

A respectful challenge might sound like this: “I understand the outcome we want. I am concerned that the proposal assumes customers will accept a process they currently find difficult. Before we commit the full budget, could we test that assumption with the people who would use it?”

That question protects the relationship by making the concern concrete. It also protects the business by introducing a way to learn. The leader can examine the assumption without being asked to defend their intelligence in front of the room.

The same standard applies to my own advice. If I recommend something, I should be able to explain the reasoning, identify the uncertainty, and describe what would make me revise the recommendation. Experience gives a person useful judgment. It does not grant immunity from checking.

The difficult part often comes before the evidence arrives. People become attached to plans for understandable reasons. A proposal may represent months of effort, professional identity, or a promise already made to someone important. A test can feel like a threat to more than the idea.

That is why leaders should make revision normal before anyone needs it. State that a proposal can be thoughtful and still need testing. Distinguish the quality of someone’s contribution from the ultimate fate of their suggestion. Give people a way to change their minds without becoming the evening’s entertainment.

Independent executive strategy and senior advisory work can help create that space. The value is partly in bringing another perspective and partly in asking the question while the answer can still change the decision.

There is a useful sentence for almost any strategy meeting: “What result would cause us to stop, revise, or expand this?” If nobody can answer, the proposal may currently be a commitment looking for a measurement system willing to agree with it.

What counts as evidence, and what does it actually establish?

Evidence comes in different forms. A customer interview, a completed transaction, an observed usability problem, a controlled experiment, and a year of retention data can all be useful. They answer different questions.

A customer saying “I like the idea” tells you something about their reaction in that conversation. It does not establish that they will buy, use, recommend, or renew it. They may be encouraging you. They may like the concept at a price of zero. They may have interpreted the idea differently from how you intend to deliver it.

A purchase answers a more demanding question, but it still leaves others open. Was the customer a good fit? Did the sale depend on a discount that cannot continue? Did the promised experience happen? Did delivering it require effort that makes the economics unsustainable?

The task is to match the evidence to the claim. A claim about usability needs observation of use. A claim about demand needs behavior that meaningfully reflects demand. A claim about sustainable growth needs a longer view of revenue, cost, retention, and capacity.

Evidence also needs an honest boundary. Five interviews can uncover a serious misunderstanding. They cannot precisely estimate how common every preference is across an entire market. A randomized test can estimate an effect in the conditions studied. It cannot promise that those conditions will never change.

Being clear about those limits makes the evidence more useful. It tells the team what can reasonably be decided now and what remains uncertain. False certainty may simplify a meeting, but it complicates the work that follows.

This is where marketing strategy should connect research to decisions. We do not need an impressive pile of information. We need relevant information that helps us choose where to focus, what to offer, what to say, and what to examine next.

When the metric starts eating the business

A measure begins as a way to understand something. Over time, it can become the thing everyone is rewarded for producing. If the connection to the underlying goal weakens, the number can improve while the business becomes less healthy.

Imagine a team measured mainly on how quickly it closes support tickets. The team might resolve problems more efficiently. It might also close tickets before customers believe the problem is solved. Both actions can make the same report look better.

Or imagine marketing rewarded for lead volume. A broad offer attracts many inquiries, but most are poorly matched to the service. The dashboard celebrates growth while sales spends its time sorting people the company cannot help.

The answer is to examine the relationship between the measure and the purpose. Faster support should still resolve the issue. More demand should still connect to appropriate customers. A lower acquisition cost should be considered alongside the quality and value of the acquired business.

In KPI and dashboard design, I want each measure to have a reason to exist. Who uses it? What decision does it inform? What could make it misleading? Which other observation would help explain a change?

Suppose a campaign produces fewer inquiries but more qualified opportunities. Is that a failure? It depends on the goal, the economics, and what happens next. A single number cannot carry the entire interpretation simply because it fits nicely inside a colored box.

Timing matters as well. Some results appear quickly; others take time. A customer may inquire today, buy later, and reveal the quality of the fit through months of delivery. If the team changes direction every time the newest number twitches, it may never allow the process to become interpretable.

At the other extreme, “these things take time” can become a permanent shelter for poor performance. The practical response is to identify reasonable early signs, the later outcome they are expected to support, and the conditions that would justify reconsidering the plan.

A useful dashboard helps people investigate. It should be possible to move from “this changed” to “here are plausible reasons” to “this is what we will do next.” A beautiful dashboard that never changes a decision is, in effect, expensive wallpaper with refresh intervals.

Remember the people outside the measurement system, too. If abandoned inquiries never enter the CRM, the sales report may overlook the experience that caused them to leave. If only enthusiastic customers answer the survey, the answers describe the respondents, not everyone who encountered the business.

The customer experiences the whole pudding

A customer does not experience your company in the same categories used by the management team. They encounter a sequence: discovery, questions, decisions, payment, delivery, support, and whatever happens when something goes wrong.

They may never know which department caused the difficulty. They simply know whether the promise felt true. The fact that three teams met their individual targets will provide limited comfort if the customer still has to explain the problem again.

The UK Government Service Manual offers a practical approach through moderated usability testing: observe participants attempting relevant tasks and learn from what they do, think, and encounter. The method brings the service into contact with people who must actually use it.

There is a special kind of humility in watching someone struggle with something your team considers obvious. The button is right there. The wording made perfect sense in the meeting. The user, apparently unbriefed on how successful the design is supposed to be, continues looking in the wrong place.

The temptation is to explain. In a research setting, that can hide the very problem you need to see. In ordinary use, the designer will not be sitting beside every visitor offering gentle hints. If the experience depends on that assistance, the assistance is part of the cost of the design.

A useful task might ask a prospective customer to determine whether your company serves their location, understand the next step, or find what information is needed for an estimate. Watch where they hesitate and what they infer. Their interpretation is evidence, even when it differs from your intention.

The same approach applies beyond screens. How does someone arrive for an appointment? Can they understand the instructions? Does the confirmation answer the question they are most likely to have? Does a staff member have to repair the same misunderstanding repeatedly?

Research should include relevant differences in access and ability. The NHS service manual’s accessibility research guidance addresses the importance of considering disabled users and assistive technologies. A service that works smoothly for its creators may still create substantial barriers for other people.

For businesses in Florida, the same principle travels easily. A visitor navigating an attraction, a business buyer evaluating a technical service, and a homeowner requesting help may use the site under very different conditions. We should discover those conditions rather than imagine a single ideal customer sitting patiently at a desk.

The brand promise becomes credible through these details. If the promise is clarity, the process should be understandable. If it is care, a confused customer should receive a helpful response. If it is expertise, explanations should make the decision easier without requiring the customer to become an expert first.

Design a test worth trusting

“Let’s test it” sounds straightforward until the team has to decide what the test is. Are we comparing two approaches? Observing whether a process can function at all? Exploring which questions customers ask? Estimating whether an improvement is large enough to justify the cost?

Begin with the decision. If the result cannot change what you plan to do, be honest about whether the activity is a test or a demonstration. Demonstrations have uses. They become misleading when presented as an open investigation whose outcome has already been selected.

Then define a specific hypothesis. “The website needs to be better” is a complaint with excellent career prospects. “Customers may abandon the inquiry because the form asks for information they do not yet have” is something we can investigate.

Choose a method suited to the uncertainty. Observing people use the form can reveal confusion. Comparing versions under a well-designed experiment can help estimate the effect of a change. Talking with sales can reveal whether the removed fields were necessary to handle inquiries well.

Microsoft’s researchers explain in their work on trustworthy A/B-test analysis that online experimentation has statistical and implementation complications. Merely placing two versions into the world does not guarantee a reliable comparison.

For a small business, this means choosing an honest level of confidence. You may not have enough activity to support a precise statistical conclusion. You can still identify clear usability problems, run a limited pilot, and learn from carefully documented observations. Call those findings what they are.

Before starting, agree on the primary outcome and a few safeguards. If you make a form easier, check whether submissions remain relevant. If you automate a response, check whether it is correct and whether the customer can reach a person when needed. Improvement in one place should not quietly create a larger problem elsewhere.

Also establish a reasonable observation period and a review point. Looking repeatedly at early results and stopping when they happen to look attractive can produce a misleading story. Seasonal changes, unusual customers, and small samples can all make the first result less dependable than it appears.

Afterward, examine whether the test actually ran as intended. Were the right people included? Was the measurement functioning? Did another change occur at the same time? Microsoft’s post-experiment guidance emphasizes checking whether metric movements align with the test design before making decisions.

The final output should be a decision with a reason: proceed, revise, stop, or investigate further. An experiment that ends in a report nobody uses has produced an artifact, but it may not yet have produced learning that matters to the business.

A good decision can have a bad outcome

Here is where the proverb needs a careful reading. Actual results matter, but a single result does not reveal everything about the quality of the decision that produced it. Uncertainty remains part of business.

Jonathan Baron and John Hershey’s research on outcome bias in decision evaluation examined how knowledge of an outcome can influence judgments of the preceding decision. It is a useful warning against rewriting the quality of our reasoning after we know how the story ended.

Imagine two companies launching a modest pilot after relevant customer research. One encounters an unexpected disruption outside its control. The other benefits from a favorable event it did not anticipate. The different outcomes deserve investigation; they do not automatically establish that one team was foolish and the other brilliant.

Now imagine a company committing heavily to an unsupported assumption and getting lucky. If leadership treats that outcome as validation of the process, it may encourage the same recklessness on a much larger scale. Success can teach the wrong lesson when nobody asks how it happened.

Evaluate both the reasoning and the result. Given what was reasonably knowable at the time, was the decision sensible? Were the risks understood? Was the commitment proportionate? Once new information arrived, did the team respond appropriately?

Then examine the outcome on its own terms. What changed? What did the customer experience? What did the business gain or lose? Which assumptions now look stronger or weaker? This allows accountability without pretending anyone had access to tomorrow’s information yesterday.

The distinction is especially important for innovation. If every unsuccessful experiment becomes evidence of incompetence, people will learn to propose only things that are easy to defend. If every failure is celebrated as brave learning, standards can disappear. Good leadership makes room for thoughtful risk while requiring honest examination.

The pudding still has to be tasted. We should also know whether the oven malfunctioned, whether the recipe was followed, and whether one enthusiastic diner is enough to justify opening a restaurant.

AI demonstrations: impressive is the beginning of the conversation

A polished technology demonstration can make the future look wonderfully cooperative. The input is clear. The example is appropriate. The result appears quickly. Everybody sees what the tool could do under favorable conditions.

The business question begins afterward. What happens with incomplete information, unusual requests, contradictory records, and users who have not rehearsed the demonstration? What happens when the output needs checking, an integration fails, or someone must explain a mistake to a customer?

For an AI-assisted workflow, I would define the task before admiring the tool. Which part of the work is it intended to improve? How will correctness be assessed? Who reviews uncertain output? What happens when the system cannot complete the task responsibly?

Then compare total effort. A draft produced in seconds can still require substantial review and repair. The right question is whether the whole process becomes better, including the work shifted to other people. A stopwatch on generation alone may miss most of the job.

Use representative examples, including difficult ones. A proposal tool should be tested on the kinds of requests the business actually receives. A support assistant should encounter the ambiguities that support staff recognize. A reporting tool should be examined when records are incomplete, rather than only when every field is beautifully populated.

Define acceptable boundaries. Some outputs may be suitable for internal preparation but require review before reaching a customer. Some tasks may remain better handled by a person. A useful pilot can establish where the technology helps and where the process needs another approach.

There is no loss of ambition in asking these questions. They make a promising idea more concrete. They help a business distinguish between a compelling possibility and a dependable way of working.

The worst outcome is often neither adoption nor rejection. It is an expensive half-adoption in which nobody knows what the system is supposed to do, employees maintain the old process as insurance, and management reports that transformation is progressing nicely.

A practical example: a campaign with an attractive number and an awkward result

Consider a hypothetical commercial service company. Its campaign generates more inquiries than before, but the sales team reports that many are outside the service area or too small for the company’s operating model. Marketing points to volume. Sales points to wasted effort. Both are describing something real.

The leadership question is whether the campaign supports the business’s actual goal. If the goal is profitable recurring work in a defined territory, inquiry volume alone cannot answer it.

Define the useful outcome together

Bring marketing, sales, and operations into the same discussion. Agree on what makes an opportunity appropriate: geography, service need, scope, timing, and capacity. Write those criteria clearly enough that the teams can apply them consistently.

This is not an invitation to reject people rudely. A business can explain its fit respectfully and direct unsuitable inquiries elsewhere when possible. Clear positioning helps prospective customers avoid spending time on a service that cannot meet their needs.

Inspect the path that produced the inquiries

Review the message, landing page, form, and follow-up. Does the offer imply a wider territory than the company serves? Does the page explain the type of work the business handles? Are prospects receiving a timely, useful response?

Talk with the people who process inquiries. They may know which misunderstanding occurs repeatedly. Their observations can help define the next test instead of leaving the team to argue over a monthly total.

Change something with a clear reason

A limited test might make the service scope more explicit and clarify the next step. The expected benefit is better fit, even if total inquiries decline. That expectation should be recorded before the results arrive.

Watch for unintended consequences. Clearer language should help people understand suitability. It should not accidentally make appropriate customers believe the service excludes them. Review actual inquiries and customer questions alongside the numbers.

Judge the result across the business

Look at qualified opportunities, progression, sales effort, and the value of resulting work over a suitable period. Ask operations whether the new work fits available capacity. A campaign that wins business the company cannot deliver well has not completed the job.

The team may discover that fewer inquiries create more useful opportunities. It may discover that the original assumption was wrong and the real problem was slow follow-up. Either finding is useful if the organization allows it to change the plan.

This example is intentionally ordinary. Most businesses do not need a historic breakthrough every quarter. They need a clearer relationship between what they do, what customers experience, and what the results actually mean.

Build a culture that can accept the answer

Testing methods are only part of the work. The organization must be willing to receive the result. Otherwise evidence becomes something people curate to protect the preferred story.

Leaders set the tone when they respond to an inconvenient finding. A useful response asks what the team learned and which decision now makes sense. A damaging response searches immediately for the person who allowed the result to become visible.

People notice the difference. If a disappointing number leads to humiliation, the next number will arrive with more explanation and less clarity. Over time, the organization can become highly skilled at reporting progress while losing the ability to recognize its own problems.

Make the standard explicit: honest findings are useful; unsupported certainty is not required. A team should be able to say that a result is promising but inconclusive, that a pilot failed to answer the question, or that a well-liked idea needs revision.

Also distinguish persistence from protection. A promising strategy may deserve more time when the mechanism is plausible and the early evidence is consistent with expectations. A weak strategy should not receive indefinite extensions because stopping would embarrass someone senior.

A short decision record helps. State what you knew, what you expected, what you observed, and what you chose next. Months later, it can reveal whether the organization is learning or simply renaming the same unresolved issue.

Give people credit for improving the decision, including when their contribution prevents work from proceeding. Avoided waste rarely arrives with a ribbon-cutting ceremony. It can still be one of the most valuable outcomes of a thoughtful conversation.

That is the kind of confidence I respect: enough belief to act, enough curiosity to test, and enough humility to respond. It allows ambition to remain connected to the conditions that determine whether it will become useful.

Bring a spoon, and be willing to learn

The proof of the pudding is in the eating. The saying survives because it asks us to cross the distance between what we claim and what people encounter.

The British histories explored here give that idea practical texture. Greenwich asks us to test in the conditions that matter. Watt’s work asks us to understand the source of waste. Wedgwood’s experiment book asks us to preserve what we learn. Modern service research asks us to watch the people who must use what we create.

For a business, the obligation is straightforward and demanding. Explain the promise clearly. Define what would make it worthwhile. Gather evidence that fits the question. Examine the whole experience. Change the plan when the learning justifies it.

None of this guarantees a painless result. It gives you a better way to discover what is happening while there is still time to respond. It also gives customers, employees, and partners a reason to believe that your confidence rests on something more durable than a persuasive presentation.

If these questions connect with work you are doing, you can write to me.

Frequently asked questions

What does “the proof of the pudding is in the eating” mean?

It means that something’s quality is judged through actual use or experience. Applied to business, it encourages testing whether a strategy, product, or service delivers the intended benefit under relevant conditions.

Is “the proof is in the pudding” the same saying?

It is a common shortened form. The longer version makes the action clearer: trying or experiencing something provides the test. For this article, that emphasis on use is central to the business lesson.

Is this a British proverb?

It is a longstanding English-language proverb with documented early use in Britain, including wording in William Camden’s 1605 work. It is now used well beyond the UK. The article uses British examples without treating the saying as a description of every British person.

How do you know whether a business strategy is working?

Define the intended outcome, relevant measures, time frame, and conditions for reconsideration. Then examine customer behavior, delivery, economics, and other plausible explanations for the result. One attractive metric rarely answers the whole question.

Does a disappointing result mean the decision was wrong?

Not necessarily. A reasonable decision can have a poor outcome under uncertainty, and a weak decision can benefit from luck. Evaluate the original reasoning and the subsequent result, then update the approach using what is now known.

Can a small business test ideas without a large data set?

Yes. Customer observation, interviews, process reviews, and limited pilots can reveal useful information. Small samples do limit statistical conclusions, so describe findings accurately and choose a method suited to the decision.

What should leaders do when evidence challenges their preferred idea?

Examine the evidence and the test conditions, ask what remains uncertain, and decide whether to continue, revise, stop, or investigate further. Respond respectfully to the people who brought the finding forward.

What is the best way to assess an AI pilot?

Define the task and compare the whole workflow, including review, correction, exceptions, and customer effects. Test representative inputs and agree on acceptable quality and oversight. A polished demonstration is a starting point for evaluation.

Further reading