A weekly operations meeting works when the status is already known before anyone speaks. Meetings spent collecting status are meetings that could have been a dashboard. What most of the hour goes on The discussion that matters is usually squeezed into the last ten minutes. * Each area reporting what a
To onboard someone into a process, the goal is to make the system explain itself. Onboarding that depends entirely on shadowing a colleague does not scale and does not survive that colleague leaving. Why onboarding takes weeks New people are shown the interface and left to infer the process. * Training
To close the month faster, you move work backwards into the month rather than compressing it at the end. Almost everything that makes closing slow could have been captured when it happened. What the week is spent on Very little of it is accounting. * Chasing people for numbers they hold
To plan an operations budget well, you need last year's actuals broken down by cost centre — not the budget that was set, but what was really spent and on what. Why budgets repeat last year's mistakes Most planning starts from the previous budget rather than from
A quarterly process review works when it starts from data. Reviews that start from opinion produce a list of complaints; reviews that start from stage timings produce a list of changes. Why reviews produce nothing Without evidence, the loudest problem wins rather than the largest one. * Complaints prioritized by who
To prepare for an annual audit, the evidence has to exist before anyone asks for it. Most audit pain comes from reconstructing controls that were applied but never recorded. What auditors actually ask for The questions are predictable, which is what makes preparation possible. * Proof that a control was applied,
Confusing permissions or filters is one of the more consequential mistakes in an internal build. A filter changes what is displayed; a permission changes what is reachable. Why the distinction matters A filter is a convenience. Treating it as a control is how data gets exposed. * A filtered view assumed
Choosing between an indicator chart or gauge is a question about what the viewer needs to conclude. Each form makes a different claim about the data. Why panels become unreadable Every element added is individually justifiable and collectively overwhelming. * Single figures presented as charts * Proportions shown as bars, where the
Choosing between duplicate clone or template depends on what you want to carry across: the structure, the data, or an entirely external starting point. Why reuse is underused Teams often rebuild because they do not know the options exist, or cannot tell them apart. * The same app rebuilt for a
Choosing between conditional fields or multiple forms depends on whether you are hiding information or organizing it. Conditionals hide what does not apply; multiple forms separate what applies differently. Two different problems They are often treated as alternatives when they solve distinct things. * A field irrelevant to this case should
Choosing between manual or AI classification depends on the cost of getting one wrong. High-volume, pattern-following, cheap-to-correct fields are exactly what an agent should fill. Why classification fields go empty They are genuinely useful and genuinely tedious, which is why they are the first thing skipped.
Choosing between viewing a document inline or downloading it sounds trivial until you count how often it happens. On an approval process running dozens of times a day, the cost compounds. What a download actually costs Every download is a context switch out of the process. * The approver leaves the