The Decision Nobody Logged

Everyone remembers the thread. Nobody remembers the decision quite the same way.

A client requests a change. The project manager acknowledges it. Someone in operations raises a concern. Another person offers a compromise. Seventeen emails later, the work proceeds. Three weeks afterwards, the client disputes the deliverable.

The agency believes the revised scope was approved. The client believes approval was conditional. The project manager remembers a conversation that neither party can reconstruct with certainty.

Every exchange exists. Nothing has been deleted. Yet nobody can establish precisely what was agreed.

The failure wasn't communication. It was the absence of a reliable decision record.

The hidden cost of an ordinary email thread

Businesses rely on email to authorise expenditure, negotiate deliverables, approve changes and assign responsibilities. Yet email was designed to exchange messages, not to establish the current state of a commercial agreement.

Consider an ordinary agency project worth £25,000. A client requests additional revisions. Several stakeholders discuss the implications. One agrees in principle. Another questions the timeline. The agency interprets the exchange as approval and begins work. When the invoice arrives, the client challenges the additional charge.

The agency now faces three unattractive options: absorb the cost, negotiate away part of its fee, or damage the relationship defending its interpretation. Even if it is eventually proven right, the investigation consumes time it cannot recover.

Multiply that exposure across dozens of clients, projects and approval chains. The vulnerability is not poor recordkeeping. It is the distance between a record of correspondence and a record of agreement.

Why the existing solutions don't quite solve it

Project management platforms track tasks. CRMs document client relationships. Email applications preserve correspondence. Meeting notes capture discussions. Increasingly, AI assistants summarise it all.

Each performs a useful function. None of them establishes whether a particular exchange produced an unambiguous, authorised decision.

A summary may accurately describe twelve messages without distinguishing a proposal from an approval. A task may appear in a project system without the conditions under which it was authorised. A CRM note may say the client agreed, while the underlying correspondence shows that agreement depended on an unanswered question.

The distinction is subtle until something goes wrong. Then it becomes expensive.

A summary tells you what happened. It doesn't tell you what you stand to lose.

An email summariser reduces reading time. That's useful, but reading time is rarely the expensive part of a misunderstood conversation. The real exposure begins when somebody acts on an agreement that was never clearly established.

An agency delivers work it cannot confidently invoice. A consultant completes services the client never explicitly authorised. A supplier absorbs costs because a change was discussed but never confirmed. These aren't failures of summarisation. They're failures of decision integrity.

Saving someone fifteen minutes is a productivity benefit. Preventing £5,000 of disputed work from becoming an unrecoverable cost is revenue protection.

The question shifts from "What did everyone say?" to "What are we about to spend, deliver or promise that nobody has clearly authorised?"

The most dangerous sentence in business

"I thought we had agreed."

Those five words reveal a structural weakness in how organisations manage decisions. An agreement is not simply information exchanged between two parties. It has a state. It may be proposed, questioned, amended, conditionally accepted, explicitly authorised or left unresolved, and that state changes as the conversation develops.

When those distinctions stay buried in correspondence, each participant builds a slightly different understanding. One remembers the latest message. Another remembers the original proposal. Someone else remembers the phone call in between. The business keeps operating because everyone believes they have enough clarity. Until their interpretations collide.

Searchability is not accountability

The usual response to an email dispute is to search the thread: find the message, identify who said what, reconstruct the sequence. But a searchable archive only answers one question: where was this discussed?

A reliable decision record must answer harder ones. What exactly was authorised? Who authorised it, and did they have the authority to? Were conditions attached? Was responsibility assigned? Did a later exchange supersede it?

And most importantly: what remained unresolved when the organisation decided to act? Sometimes the most consequential thing in a thread is not a decision that was made. It is a decision everyone assumed somebody else had made.

That means treating a thread not as messages to summarise, but as a developing record of commitments, conditions and open questions. And accepting that sometimes the correct finding is that no decision can be established. A useful system identifies uncertainty rather than manufacturing confidence.

The costliest mistakes are preventable before anyone acts

By the time an approval dispute reaches management, the damage is done. Work has been delivered. Invoices have been issued. Expectations have hardened.

The valuable intervention happens earlier: before the next reply, before the project moves forward, before a conditional statement becomes an assumed commitment. A short confirmation of what has actually been agreed can prevent weeks of reconstruction.

This is not about adding a checkpoint to every conversation. It is about recognising when an ordinary exchange has become commercially consequential, and making its decision status explicit while the people involved can still clarify it.

From observation to application

This problem led to the development of ThreadKiller, a decision-analysis tool from Mitchell Solutions.

ThreadKiller examines email correspondence to identify decisions, responsibilities, unresolved questions and critical context, then prepares a confirmation reply for you to review before proceeding. It needs no inbox access. You choose the correspondence to analyse, and nothing is sent automatically.

The principle is deliberately simple: before acting on an agreement, establish what the conversation actually supports.

Because a business can retain every email, document every task and archive every conversation, and still fail to preserve the one thing that matters when somebody challenges the outcome.

The decision nobody logged.

Explore ThreadKiller →