Take an audit firm that specializes in media.

Every year, it works with a dozen groups: regional dailies, magazines, TV channels, radio stations. The sectors differ, but the same questions keep coming back:

• should online news be paid for?
• how do you organize a newsroom that produces for print, web and video?
• how do you reduce dependence on advertising?

A consultant is preparing a new engagement for a news channel.

He remembers a recommendation made last year to a regional daily. It fit in three lines, on page twelve of the report:

Switch to a strict paywall for local news.

The engagement had been a success. The client was happy. The recommendation had been applied, and subscriptions had gone up.

The consultant is tempted to reuse it.

After all, the question is the same.

Three lines at the end of six weeks

Those three lines had taken six weeks of work.

Behind them, there were:

• interviews with the editor-in-chief, the sales leadership and several journalists;
• the site's audience figures and how they had moved over three years;
• an analysis of advertising's share of digital revenue;
• an option seriously considered and then ruled out: letting readers have a few free articles each month before asking them to subscribe;
• a tense trade-off with management, who were afraid of losing their audience.

All of that led to the recommendation.

But the report keeps only the conclusion.

That's normal. A report is made for a client. It has to be clear, readable, actionable. Nobody wants to receive the full log of a six-week investigation.

The problem isn't that the report simplifies.

The problem is that it becomes the only trace of the work.

Every tool did its job

When you try to recover the reasoning, you find that nothing has been lost.

The interviews are in the consultants' notes. The figures are in a spreadsheet. The trade-off with management is in an email thread. The report is in the client's case.

Each tool kept the object it had been given.

None of them was there to keep the whole.

The link between these pieces, what made them justify the recommendation together, existed only in the consultants' heads. During the engagement, it was perfectly clear. A year later, one of them has left the firm, the other has gone through six more engagements, and nobody knows anymore exactly why the strict paywall won out.

What's left is a conclusion.

Without its why.

A recommendation is only true under conditions

And that why was decisive.

The strict paywall held for this daily because three conditions were met:

• its readership was local and had been loyal for decades;
• the local news it produced was available almost nowhere else;
• digital advertising made up only a marginal share of its revenue.

Under those conditions, charging for news was reasonable. Readers were willing to pay for what they couldn't find elsewhere, and the paper had little to lose on the advertising side.

The news channel is in the opposite situation.

Its news is widely available elsewhere. Its digital revenue depends on audience volume, and therefore on advertising. A strict paywall would drive away a large share of its visitors, and with them most of its revenue.

The same recommendation, right in one case, becomes wrong in the other.

And nothing in the report lets you see it.

That is the real danger of a deliverable: the conclusion survives, its conditions disappear. The risk isn't just starting from scratch. It's copying a good answer into the wrong place.

What you're not allowed to reuse

You might think the fix is simple: just reopen the daily's case to recover the figures and the interviews.

That would be a breach.

The audience figures, the revenue, what people said in interviews all belong to the client. They were entrusted for a specific engagement. Going back to them to serve another client, possibly a competitor, would betray the trust the entire consulting profession rests on.

So the firm is stuck between two bad options:

• copying conclusions without their reasons, at the risk of applying them where they become wrong;
• starting from scratch on every engagement, at the risk of never learning from its own experience.

Either way, the firm produces reports.

It doesn't get any smarter.

What you keep: the decision and its why

This is precisely the problem I'm trying to solve with what I call a Context.

A Context is a lasting workspace, separate from conversations and deliverables, that keeps what a piece of work needs to retain over time: the material received, the ideas drawn from it, the decisions made and where things stand. It can fit in a few readable files that an AI rereads at every work session.

For an audit engagement, it starts with a simple habit: every structuring recommendation becomes a decision record.

A decision record is a short sheet that keeps a decision together with what produced it:

• the decision itself;
• the alternatives considered;
• the reason for the choice;
• the conditions under which it holds;
• what would make it obsolete.

For the regional daily, it would look roughly like this:

Decision — Switch to a strict paywall for local news.

Alternatives — A model with a few free articles per month; staying free, funded by advertising.

Reason for the choice — Local news is hard to find elsewhere and the readership is loyal: free articles would mostly have gone to readers already willing to pay.

Conditions — Loyal readership, little competition for its news, marginal digital advertising.

What would make it obsolete — A free competitor entering the local news market, or a sharp rise in advertising's share.

This decision record stays inside the client's engagement, with its data.

But it finally keeps what the report had lost: the why.

What circulates: ideas detached from the client

The decision record never leaves the client's case.

What can come out of it are ideas.

At the end of the engagement, you draw atomic notes from it: one idea per note, phrased to stand on its own, with no figures and no client name. For example:

A strict paywall holds when the readership is loyal and the news is hard to find elsewhere.

When advertising weighs little in digital revenue, charging costs little in lost audience.

These notes say nothing about the daily. They contain neither its figures, nor its interviews, nor its name. They say what the engagement taught.

Then, engagement after engagement, thematic notes emerge. They connect several atomic notes to bring out a broader understanding. For example: what makes a paywall succeed or fail depending on the type of media, the nature of the news, the weight of advertising.

These notes belong to the firm.

They grow richer with every engagement.

And they betray no client.

The next client

Back to the consultant preparing the news channel engagement.

He doesn't reopen the daily's case. He doesn't need to.

He looks up the firm's notes on paywalls. He finds the recommendation that had worked, but above all the conditions that made it hold: a loyal readership, news hard to find elsewhere, marginal advertising.

He checks those conditions against his new client's situation.

None of them is met.

The dependence on audience volume overturns the recommendation, and he sees it on day one, before writing a single line. The option ruled out for the daily, free articles before a subscription, may turn out to be the right one for the channel. Or something else entirely.

The decision will belong to this new engagement.

But it will start from the right place.

And when the engagement is over, it will in turn produce a decision record, which will stay with the channel, and a few atomic notes, which will enrich the firm's. The note on paywalls will gain a nuance: what happens when news is abundant and advertising dominates.

The firm gets smarter with every engagement

The report remains the product sold to the client.

It has to be clear, concise, suited to its reader. Nothing changes on that side.

What changes is what the firm keeps after delivery.

Before, it was left with a pile of reports, reopened in the hope of finding some inspiration, without knowing whether their conclusions still held anywhere else.

Now, it keeps two distinct things:

• inside each engagement, decision records that keep the why behind each recommendation, safely stored with the client's data;
• at firm level, atomic and thematic notes that keep what each engagement taught, detached from any client.

The former protect the memory of each engagement.

The latter build the firm's knowledge.

That's where the real capital lies: not in the reports, but in the reasoning you managed to keep, and in what you managed to draw from it without carrying anything away.

A report tells you what to do.

It doesn't tell you when to stop doing it.

---

A note on the examples

The examples in this article draw on more than thirty years of practice in IT and team management, and on many conversations with peers. I have modified, generalized or recombined them: none describes a real client, product or organization, nor reflects my employer's data or positions. The opinions expressed are mine alone.