Evidence Over Assurance
A vendor once sent us a report on their own operations. It was well laid out. Quality was described as strong. Controls were described as in place. The team was described as confident. We asked for the data behind a single line, the one about defect rates holding steady. There was no data. There was a sentence, and the sentence had been written by the person whose job was to be reassuring.
We did not think the vendor was lying. That is the point. They had confused assurance with evidence, and the confusion is one of the most expensive habits in organisational life.
Assurance is a feeling. Evidence is a fact.
Assurance is what someone tells you about themselves. It can be sincere and still worthless. It answers the question "do you believe this is fine" when the useful question is "how would we know". Evidence answers the second question. It is the thing you can check.
The difference matters because the two are often mistaken for each other in the room. A confident presenter, a clean slide, a firm voice: all of these read as evidence and none of them are. We have sat through briefings where every claim was delivered with total conviction and not one claim could be traced to a document, a number, or a record.
What counts as evidence
Evidence is specific and it is verifiable by someone other than the person who produced it. A register with entries and dates. A log that shows what happened and when. A number with a source. A document that can be matched against reality. A record of a decision and the reason for it. The test is simple: could a stranger, given the same file, reach the same conclusion without you in the room.
Assurance fails that test. A statement that controls are in place is not evidence of controls. The evidence is the control itself, operating, visible in a record. A statement that quality is strong is not evidence of quality. The evidence is the defect data, the rework rate, the customer complaints, over time.
Why an unchecked finding is only an opinion
Here is the uncomfortable version. A finding you cannot check is an opinion, however well dressed. It may be a good opinion, formed by an expert, and it is still an opinion, because it rests on trust rather than verification. Trust is fine between people. It is not a foundation for a decision that carries money, risk, or someone's career.
This is not a technical point. It changes what you do. If a finding is evidence, you can act on it, defend it, and revisit it. If a finding is an opinion, acting on it is a bet. Most organisations do not realise they are betting. They think they have facts because the facts were stated confidently by someone credible.
How we test a finding
We run every material finding through the same three questions.
First, what is this based on. If the answer is a person's impression, we go looking for the record underneath. Sometimes there is one and the impression was right. Sometimes there is not, and the finding quietly dissolves.
Second, who else could check it. If only the person who produced the finding can confirm it, it is weak. Evidence should be checkable by someone independent, and it should survive the check.
Third, what would change my mind. A finding that no possible fact could contradict is not a finding. It is a belief. A good finding comes with the conditions under which it would be wrong, and the data that would show it.
We apply these to our own work before we apply them to anyone else's. A report that cannot survive its own test has no business leaving the desk.
The cost of running on assurance
The cost shows up slowly and then all at once. A compliance function that runs on assurance discovers the gap only when an inspector arrives. A quality process that runs on assurance discovers the defect only when a customer does. An exit process that runs on assurance learns what people really thought only after the good ones have left and said so elsewhere.
There is also the cost to the people inside. When an organisation rewards assurance, it teaches everyone to sound confident rather than to be right. The most assured voice in the room wins, and the person with the awkward data learns to keep quiet. Over time the organisation loses its ability to hear bad news, which is the one thing it most needs to hear.
Two questions that sort one from the other
When a claim arrives, two questions do most of the work. The first is: who produced this, and do they benefit from it being believed. A claim from someone with an interest in the answer needs more support than a claim from someone without one. The second is: what would I see if this were false. If nothing would look different, the claim cannot be tested, and an untestable claim is a belief dressed as a finding.
These two questions are quick and they change the conversation. They move a meeting from how confident everyone sounds to what is actually known, and they do it without anyone having to accuse anyone of anything. The person making the claim simply learns that claims here come with their basis attached, which is a habit worth teaching early.
What good looks like
Good looks unglamorous. Claims come with their basis attached. A statement about quality is followed by the number and the source. A statement about compliance is followed by the register and the date. A statement about culture is followed by the exit data and the pattern it shows. Nothing is asserted that cannot be checked, and nothing important is left as an impression when a record could settle it.
This is slower to produce and faster to act on, because a checked finding can be used. It also changes the tone of a meeting. When everyone knows that claims will be tested, the conversation moves from who sounds most certain to what is actually true, which is where the value was all along.
The habit underneath
Evidence over assurance is not a technique. It is a habit of asking, gently and every time, "how would we know". Asked early, it saves money. Asked late, it saves a career. Asked as a matter of course, it changes how an organisation thinks, because it stops treating confidence as proof.
A finding you cannot check is only an opinion. Check it, or call it what it is.
We don't guess. We investigate, and the first thing we investigate is the basis of the claim in front of us.