Seven questions for the first two weeks

The seven questions are a card, not a survey. They are answered in the first two weeks on site, reviewed weekly, and the rule attached to them is blunt: if the cells are not filled, you do not quote. Each cell has four columns — the question as asked, what you are listening for, what to do when it cannot be answered, and the self-deceiving answer people give instead. The fourth column is what makes the card work, because every question has a plausible answer that means nothing.

There is a second rule, about timing. Answers are written down the same day. A day's delay costs roughly a third of the content — not because memory fades evenly, but because by tomorrow the answer has been quietly reconciled with the version the client would prefer you to have heard.

Question one is not "who asked for this"

The sponsor asked. That is rarely the interesting fact. The person most damaged by the current process is usually a middle manager: the one who owns the spreadsheet, who explains the delay upward, who has to clean up when the work fails. If the project collapses in three months, the sponsor loses a line item; that manager loses standing, and possibly the post.

So the first question is: if this fails completely in three months, who is hurt most? A good answer names a post, the specific way that person's numbers suffer, and when and to whom they report it. A self-deceiving answer is "the boss", "the client", or "nobody" — all three mean the person has not been found yet. If the answer cannot be produced, the cell is marked unanswered and the work stops there; the next move is to go and find that face. Do not ask the sponsor to make the introduction, because the sponsor's introduction changes what the person says.

Downstream, this answer is what makes the money map usable. It identifies the person who actually approves, as distinct from the person who signs.

The other six, and what each one moves

Two: has anything like this been tried in the past two years, and what happened? Listen for a previous vendor, a go-live date, whether the thing is still in use, and where the counterpart works now. "We did it and it went fine" and "never" are both self-deception: the first hides a failure, the second usually hides a predecessor nobody wants to mention. This answer converts the project from an opinion into a precedent, and it teaches you what "still in use" means inside this company.

Three: if this succeeds, whose work loses value? Listen for the specific actions replaced, the hours attached to them, and what share of that post's duties they represent. "Nobody loses", "IT will lose" and "junior staff" are the standard evasions; the last two usually err in the same direction, blaming a department or a level rather than a post. This is the first draft of the who-loses map, and a project with no losers is a project not yet seen.

Four: what is the first thing people in this department do each morning? Listen for a concrete action, a start time, and who it is handed to. "Meetings" and "clearing yesterday's backlog" are not answers. The follow-up is to watch one morning, the first fifteen minutes of it, because the real workflow starts there and that is where any instrumentation has to sit.

Five: which table is the most unpopular, who fills it in, and who reads it? Listen for a name, a cadence, where the table goes, and the last time a decision changed because of it. "There isn't one" is almost never true. Ask for the last three versions and compare: the columns that are always blank, or always written over by hand, show which parts of the form are theatre.

Six: if you had to make a decision for me tomorrow, what would you need to know, and where does it live now? Listen for the first tool they would open, the source, and which table the answer lands in. "It's in the data platform" and "it's in my head" are the two failure answers, and they point at the same thing: the knowledge exists but has no interface. This is the smallest possible version of the out-of-the-loop test.

Seven: has anyone internally opposed this, and on what grounds? Listen for the opponent's post, their actual argument, and whether they are still there. "No opposition" is the most expensive self-deception on the card, because opposition that is not surfaced does not disappear; it waits. If the opponent has not been spoken to, the answer is recorded as inference, not as a finding.

What the card feeds

Three answers go straight into reconnaissance artefacts. Question one determines who really approves in the money map. Question three produces the first list for the who-loses map. Question six produces the shadow-system inventory, because whatever the answer lives in is either a formal system or a spreadsheet pretending to be one. Questions four and five usually surface the same evidence from a different angle.

The discipline is the blank cell

Every answer carries a quotation and a date, so that a later disagreement is about the fact rather than about memory. A cell that cannot be answered stays blank and is marked unanswered; it is not filled with a reasonable inference. The blank is itself a finding, because it shows a part of the process the client cannot see about itself.

The card is finished when seven cells are filled or explicitly blank, question one names a post, question three names at least one concrete piece of work that goes to zero, question six comes with something you can look at, and the opponent in question seven has actually been contacted. Until that last condition holds, the seventh answer is a rumour with a date on it.