.png)
Perfect Data Is Where Good Decisions Go to Die Waiting.
A peculiar superstition in modern organisations is the belief that better decisions will arrive as soon as the data becomes perfect. This is comforting because it makes inaction sound intelligent. “We need more data.” “We need one source of truth.” “We need to validate the numbers.” These phrases are often reasonable. They are also sometimes the corporate equivalent of hiding under the duvet until the weather improves.
Perfect data has a seductive quality. It promises certainty, removes embarrassment, and allows everyone to postpone the awkward business of judgment. Unfortunately, most organisations do not operate with perfect information. Customers are moving, systems are changing, teams are adapting, competitors are acting, and reality keeps behaving rudely while the steering committee waits for version 17 of the dashboard.
The uncomfortable truth is this: imperfect data used consistently is often more valuable than perfect data that arrives too late. A compass that is slightly inaccurate but always points in roughly the same direction is more useful than a satellite system that becomes available after the ship has hit the rocks.
The Problem: We Confuse Data Quality with Decision Readiness
In the previous articles, we explored how organisations improve when they design around human behaviour, cognitive ease, learning, and embedded feedback. Imperfect data sits right at the centre of that challenge because it is not only a technical issue. It is psychological.
Human beings dislike uncertainty. The brain prefers clarity, closure, and patterns it can trust. Imperfect data creates discomfort because it forces people to act without the full story. It threatens status if the decision goes wrong. It triggers risk aversion because losses feel more painful than equivalent gains. It creates cognitive dissonance because leaders want to be Data-Driven but do not want to admit that the data is partial, messy, delayed, or politically inconvenient.
So, organisations do something very human. They wait. They commission another report, add another field, debate definitions, challenge the sample, question the dashboard, and build a more elaborate temple to certainty. Meanwhile, the customer problem continues, the rework continues, the process friction continues, and the data quality initiative becomes a waiting room where improvement slowly loses the will to live.
At CoEcosystem, we help organisations use imperfect but meaningful data to drive better decisions, continuous improvement, and progressive data maturity. Learn more at CoEcosystem’s website.
A Better Framework: The Good-Enough Data Test
The answer is not to lower standards. Bad data can be dangerous. But the answer is also not to wait for perfection. The better question is: “Is this data good enough for the decision we need to make now?” That question changes everything.
1. Is the Data Directionally Useful?
Not all decisions require laboratory-grade precision. Some require enough signal to choose the next sensible move. If customer complaints have doubled, you may not need a perfect taxonomy before investigating. If rework is concentrated in one handover, you may not need six months of validated trend data before observing the work.
The first question is whether the data points in a useful direction. Is the signal strong enough to justify exploration, a small test, or closer attention? If yes, the organisation can move without pretending it has achieved certainty. This is how mature improvement works: not by making giant decisions from weak data, but by making small, reversible decisions from useful signals.
2. Is the Measure Consistent Enough to Learn From?
Consistency often matters more than perfection. A measure that is imperfect in the same way each week can still reveal whether things are getting better or worse. This reduces cognitive load because people learn the pattern, understand the limitation, and build confidence through repeated use.
The danger is constantly changing definitions in search of purity. Every new definition may be more accurate in theory, but it can destroy comparability in practice. Leaders then spend meetings debating whether the movement is real or merely a measurement change, which is a marvellous way to convert data into fog.
Good improvement systems often start with stable, transparent measures. “This is not perfect, but we know what it includes, we know what it excludes, and we will use it consistently while improving it over time.” That sentence is more useful than many dashboards.
3. Does the Data Reduce or Increase Decision Anxiety?
Data should help people decide. Too often, it helps people worry more precisely. A dashboard with twenty-seven metrics, six colours, three filters, and no obvious decision attached is not insight. It is a cognitive obstacle course.
Good data design reduces decision anxiety. It shows what matters, why it matters, what has changed, and what action is available. It separates signal from noise. It uses plain language. It makes trends easier to see. It gives people enough context to act without requiring them to become amateur statisticians during a Tuesday performance meeting.
The test is simple: after looking at the data, does the team know what conversation to have, what question to ask, or what action to test? If not, the data may be decorative rather than useful.
4. Are We Improving the Data While Improving the Work?
This is the most important shift. Data quality should improve through use, not before use. When teams use data to make decisions, they discover what is missing, what is unclear, what needs better definitions, what should be automated, and what no one actually uses. Data becomes cleaner because it has consequences.
This is behavioural design. If data hygiene is treated as an abstract compliance duty, people will do the minimum. If better data helps them solve real problems, reduce rework, protect customers, or make their own work easier, engagement improves. The best data improvement process is not a lecture about data quality. It is a feedback loop where people can see that better data leads to better decisions.
A Practical Example: The Rework Metric Nobody Trusted
Imagine an operations team trying to reduce rework. Everyone knows rework is a problem. Customers are frustrated. Staff are annoyed. Managers suspect certain handovers are causing the issue. But the data is messy. Some rework is logged properly. Some is hidden in email. Some is corrected quietly by experienced staff. Some is recorded under vague categories such as “other,” which is where useful information goes to disappear.
The perfection-first response is predictable: “We cannot act until the data is clean.” So the organisation launches a data quality project, debates definitions, adds fields, and waits. Three months later, the data is slightly better, the rework remains, and staff have learned that improvement begins with admin.
A good-enough data approach starts differently. It says, “We know the data is imperfect. What can we safely learn from it?” The team reviews four weeks of available records, samples real cases, speaks to frontline staff, and identifies three recurring rework causes. The numbers are not perfect, but the pattern is strong enough to test.
They run a small experiment. One handover checklist is simplified. One unclear input field is redesigned. One upstream team receives faster feedback when rework occurs. At the same time, the team improves the rework categories because people can now see why better classification matters. Within weeks, rework begins to fall in the tested area, and the data improves because it is connected to action.
One approach waits for data confidence before improving the work. The other builds confidence by improving the work and the data together.
The Real Shift
The real shift is from data perfection to decision usefulness. It is from treating imperfect data as an excuse for delay to treating it as a starting point for disciplined learning. It is from asking, “Can we trust this completely?” to asking, “What decision is this good enough to support?”
This requires psychological maturity. Leaders must create safety around uncertainty. Teams need permission to say, “The data is imperfect, but the signal is useful.” Analysts need to explain limitations without making the data sound worthless. Managers need to reward action based on transparent assumptions rather than punish every decision made without perfect certainty.
It also requires better design. Use simple visuals. Show trends, not just snapshots. Make definitions visible. Separate known facts from assumptions. Use confidence levels. Link metrics to decisions. Celebrate small improvements in both performance and data quality. Make data hygiene rewarding by showing its effect on real work.
The aim is not reckless action. It is intelligent movement. A business that waits for perfect data may look disciplined, but sometimes it is merely frightened in a more sophisticated way. A business that acts on imperfect data without understanding its limits is careless. The sweet spot is progressive confidence: act where the signal is strong enough, learn quickly, improve the measure, and adjust.
Continuous improvement has never required perfect information. It requires honest information, used consistently, in a culture safe enough to learn. This is why the future belongs not to organisations with perfect data, but to organisations that can make better decisions before certainty arrives.
Further Reading
If this idea resonates with you, these books are useful companions for going deeper.
1. How to Measure Anything, Douglas W. Hubbard. A practical book on measurement under uncertainty. It is useful because it challenges the belief that important things cannot be measured and offers ways to reduce uncertainty enough to make better decisions.
2. Superforecasting, Philip E. Tetlock and Dan Gardner. A valuable book on making better judgments in uncertain conditions. It reinforces the importance of probabilistic thinking, updating beliefs, and learning from imperfect information.
3. Factfulness, Hans Rosling, Ola Rosling, and Anna Rosling Rönnlund. A clear and engaging book on interpreting data more wisely. It helps leaders recognise how instinct, fear, and outdated assumptions distort the way we read the world.
4. The Signal and the Noise, Nate Silver. A useful book on distinguishing meaningful signals from random noise. It is especially relevant for organisations drowning in metrics but struggling to know what deserves attention.
5. Calling Bullshit, Carl T. Bergstrom and Jevin D. West. A sharp guide to spotting misleading numbers, weak claims, and data used without context. It is useful for building healthy scepticism without falling into data cynicism.
6. Data Feminism, Catherine D’Ignazio and Lauren F. Klein. A thoughtful book on power, context, and ethics in data. It is relevant because data is never neutral in how it is collected, interpreted, and used inside organisations.
Discussion Question
Where is your organisation currently waiting for perfect data when “good enough, consistently used, and transparently improved” would help you make progress sooner?
Comments