A strategy consultancy charges a fortune for slide decks and meetings — that’s the oldest joke about the industry. The joke misses the point: what justifies the fee isn’t the slide, it’s the discipline of never writing an answer before formulating the right question. Most companies do the opposite — they jump straight to a solution and only later, if there’s time left, check whether it solved the right problem.
The most common symptom in an unstructured leadership meeting isn’t lack of data — it’s an excess of opinions competing for attention, with no criterion for which one to test first. In those meetings, whoever speaks loudest or last usually wins, not whoever holds the most likely hypothesis.
MECE: splitting a problem without overlap or gaps
MECE (mutually exclusive, collectively exhaustive) is the first tool any junior consulting analyst learns, and the easiest one to copy without paying for it. The rule is simple to state and hard to practice: break a large problem into parts that don’t overlap (mutually exclusive) and that, added together, cover the entire problem (collectively exhaustive). Without it, a meeting discusses random fragments of a problem — a bit of marketing, a bit of operations, a bit of competition — with no guarantee that, together, they explain the whole.
Revenue decline, for instance, always splits into three branches that don’t overlap and that, together, exhaust the explanation: fewer customers coming in (acquisition), the same customers buying less (average ticket), or customers leaving for a competitor (retention). Any real cause of revenue decline lives in one of those three branches — never outside them, never in two at once.
An example: why revenue dropped 12% this quarter
Picture the owner of a mid-sized company looking at the quarter’s close: revenue 12% below target. The most common instinct is to react on the most visible branch — cut the marketing budget, switch agencies, or blame “a weak market.” None of those reactions come from a question; they come from a guess dressed up as a decision.
Applying the MECE split before any action: the three branches — acquisition, average ticket, retention — get checked separately against historical data. Acquisition is stable, the number of new customers per month hasn’t changed. Retention is stable too, the rate of returning customers is the same as always. Average ticket, on its own, dropped 14% — more than enough to explain the entire 12% revenue decline by itself. The question stopped being generic (“why did revenue drop?”) and became specific (“why is the customer who keeps buying spending less per purchase?”) — and only that specific question led to the real cause: an anchor product, historically sold alongside the rest of the catalog, had been discontinued two months earlier with no equivalent replacement.
Hypothesis before evidence, not after
The second tool is even more counterintuitive: state the most likely hypothesis before looking for data, not after. It sounds like an invitation to confirmation bias, but it works the other way around — a written, explicit hypothesis is falsifiable (“if this is true, data X should show Y”), while analysis without a hypothesis turns into an endless dig, always finding one more variable to check before deciding anything.
As João Paulo Batistella, an innovation executive, argues when analyzing why so much corporate AI investment fails to deliver a return: the question that separates real investment from simply “following the wave” is a single one — what result is actually expected? If a company can’t answer that before spending, it’s probably investing because the market is investing too, not because it tested a specific hypothesis. The same discipline applies to any business decision, not just AI: without an explicit hypothesis about what you expect to find, there’s no criterion for knowing when to stop analyzing and start acting.
The 80/20 test: which hypothesis to check first
Not every branch of a MECE split deserves equal analysis time. The practical question is: which hypothesis, if confirmed, explains the largest share of the problem at the lowest verification cost? In the example above, checking all three branches against data the company already had (sales report, CRM) took less than an hour — far cheaper than any of the instinct reactions (switching agencies, cutting budget) and far more accurate than any of them. Only after isolating the right branch (average ticket) did it make sense to spend more time digging specifically there.
It isn’t about having more data. It’s about structuring the question before spending any time on any data — most companies invert that order, gathering data first and only later trying to figure out which question it answers.
Next time a problem lands on your desk, before opening any spreadsheet:
- Write the problem as a MECE split — branches that don’t overlap and that, together, exhaust the explanation.
- For each branch, write down the most likely hypothesis before looking for any data.
- Test the hypothesis that, if confirmed, explains the largest share of the problem first — and only dig into the others if it doesn’t add up.
None of those three steps requires a consulting contract or new software. The most valuable tool a firm like that sells was never proprietary — it’s a discipline of thought any company, of any size, can copy for free starting today.
Follow Eleva Tecnologia for more analysis on technology applied to business: follow @ElevaTechno on X or @elevatechnologies on Instagram, or learn about the group at elevatec.net/about.
