Reading their data state in plain words

Take the Monday report and go down it one number at a time. For each, ask the bank's question g08: "Where do the stock numbers in it come from, step by step?" with its teach-back said first: "A number that is typed by a person on a Friday behaves differently from one a till records. I need to know which I am dealing with." She does not need to know what a source system is to answer that. She knows that sales come off the till and that stock is what the outlet managers put in the shared Excel on Friday, because she has chased them for it.

The three answers carry three different builds. A number recorded by a system arrives on its own, with a timestamp, in the same shape every time; at Harbourline that is the POS, one CSV per outlet, three years deep. A number typed by a person arrives when the person remembers, in whatever shape they typed, and at Harbourline there are three versions of that shape in circulation and an "adjustments" column nobody can explain. A number that is not recorded cannot be built on at all until somebody starts writing it down; supplier orders live in WhatsApp threads and there is no order table anywhere.

Say the middle case out loud when you hear it, because it is the one that quietly decides the project. A stock number typed on Friday cannot feed a Monday decision any fresher than Friday, whatever the dashboard refreshes at. It also cannot be trusted more than the person typing it is checked, which at Harbourline is not at all. That one answer tells you the first build is a single agreed stock sheet with a checker, not a chart of the sales.

Two sources are usually run together into one number, and they should not be. In 2014 Steve Lohr reported in The New York Times that data scientists spend 50 percent to 80 percent of their time collecting and preparing data before it can be explored. In 2016 CrowdFlower's data science survey reported that respondents spent 60 percent of their time cleaning and organising data and a further 19 percent collecting it. The two sources measured two different things with two different methods and gave two different shapes of answer: one a range for collecting and preparing together, the other two separate figures for cleaning and collecting. Quote whichever you use as it stands, and never the "80 percent cleaning" figure that circulates as if it were one measurement, because it is not one anyone made.

The point for door 4 is the reason behind both figures. Preparation time is mostly the cost of typed and absent numbers: reconciling three versions of the stock sheet, working out what the adjustments column meant, discovering the order data does not exist. Every minute you spend at door 4 asking "typed or recorded" is a minute you are not spending in week three finding out the hard way.

DAMA-DMBOK, 2nd edition, chapter 15, names five levels of data management maturity: Initial/Ad hoc, Repeatable, Defined, Managed, Optimized. Five, not six: DMBOK has no level 0, and a "not started" level you may have seen belongs to a different framework. The names are the whole of what is borrowed. The questions beside each rung are ours, written so a person who runs pharmacies can answer them about her own Monday.

Rung 1: "when the number is wrong, who fixes it, and how do they know?" If the answer is "Priya notices, usually" the habit is ad hoc. Rung 2: "does the same person do it the same way every week?" Yes, Priya, six hours, from the same export, which is why Harbourline is past the bottom rung. Rung 3: "is any of it written down?" No. There is a spreadsheet with an adjustments column and the method is in Priya's head, which is also why her two weeks of leave in November matter. Rung 4: "does anyone check afterwards whether it was right?" Nobody, which is why nobody trusts it. Rung 5: "has it got better on its own in the last year?" It has got a column longer.

Write the rung beside each source in your notes and never tell her the number. "You are a 1.5" is a grade, and a grade ends a conversation that was going well. What she hears is five questions that each made her notice something about her own shop, and what you carry out is a per-source reading that tells you where to start: the source on the lowest rung that feeds the biggest decision. The real assessment, with the practices behind each level, is a governance discipline, taught in the Governance 101 course; in a 45-minute discovery meeting the ladder is an interview aid and nothing more.

Joe Reis and Matt Housley, in Fundamentals of Data Engineering (2022), describe companies as starting with data, scaling with data or leading with data, reported here from summaries rather than a page. Harbourline is starting with data: one recorded source, one person who assembles it, and a request for a dashboard that is really a request to be able to trust a number. Everything you propose should fit a company at that stage, which rules out most of what a vendor would propose.

Leandro DalleMule and Thomas Davenport, in "What's Your Data Strategy?" (Harvard Business Review, 2017, reported), ask whether a company needs a single source of truth or several fit-for-purpose versions of the truth. Harbourline has multiple versions by accident: three stock spreadsheets in circulation, none of them the agreed one. Their frame is useful because it names the choice. A company can decide to run several versions; it should not discover it has.

"Who owns this data" is a question a business person hears as "whose department is it", which is not the answer you need. The bank's question g13 asks it the usable way: "What does the loyalty app hold about a customer, and who owns that data?" with the teach-back "Owned means who can say yes to an export. That is usually a contract, not a login." Said that way, Adeline's answer is exact: the vendor holds it, an export costs extra and needs a clause, and anything with Malaysian customer data goes past the head-office lawyer, who takes four weeks.

Ask the same question of every box. Who can say yes to changing a product code? Nobody, which is the answer that explains the renamed SKUs and the adjustments column in one go. Who can say yes to a new column in the stock sheet? Whichever outlet manager opens it first. An unowned source is not a gap in your notes to fill in later. It is the first thing to propose an owner for, and Priya, two hours a week, is the obvious candidate.

Question g11 is short and has to be asked with a name in it: "Who can log in to the POS back office and pull an export?" Teach-back: "Access is the first blocker on most data projects, so I ask early and I ask by name." The name is the point. "IT can" is not an answer; "Kenneth can, and only Kenneth" is, and it is the answer that makes you ask what happens in January.

That second fact, that Kenneth is leaving and she has not told the team, does not come out on the first question, whatever you ask. On the bench it sits behind the highest trust gate on the sheet: she only volunteers it once four or five open questions with their reasons said aloud have convinced her you are trying to help her and not to audit her. That is why door 4 is the fourth door and not the first. Walk in and ask who holds the admin login before she knows why you are asking, and you get "our IT guy". Ask it after the Monday walk-through and the stock sheet, with the teach-backs in between, and you get the name, the date and the fact she has been carrying alone. Session 3 shows the same thing as a number: the fact is worth 3 points, and most question orders never reach it.

The map above is the first thing you send Adeline after the meeting, before any chart, before any proposal. It has four boxes, an owner or the word "nobody" on each, a key, a lock and one target. It contains no word she would have to ask about. She can look at it on her phone on Monday and say "the stock sheet is not three versions, it is four" or "actually Kenneth's deputy also has a login", and that correction is worth more than a week of your building. She could never make the same correction to a schema diagram, because she would not know which box was wrong.

The map is also your own honesty check. If a box on it says "nobody", the project has an ownership task before it has a data task. If the only key is on one person, the project has a continuity task before it has a chart. If an arrow is dashed, the number at the end of it cannot be built on until someone starts recording it. A map with those three marks on it is a project plan that she drew, and that is why it goes first.