Should we buy this, and at what price.
Edition 3, October 2026
The failure rate for corporate acquisitions has been reported at somewhere between half and three-quarters for forty years. The number is not contested. Deals keep getting done anyway, because a live target generates momentum that a spreadsheet cannot stop.
By the time a target reaches the board, four advisers are being paid to close it and nobody is being paid to kill it. The screening that mattered happened months earlier, informally, in a conversation nobody minuted.
Targets are read at three tiers, and each tier ends in a verdict rather than a score. The four lenses apply as they do everywhere: does it pay, does it build recognition, does it open future options, and can this buyer actually integrate it. The fourth lens can kill a deal the first three support.
The verdict is recorded with its confidence range before the negotiation moves, which means the reading cannot be quietly revised upward as the deal gathers speed. Twelve months after completion the verdict is marked against what the asset actually did.
Acquisitions runs five Calls, not the seven used elsewhere, because the decision economics are a gate rather than an allocation. Kill: terminate the target entirely, do not revisit. Park: hold a target outside the mandate against a named trigger, a price move, a mandate change or a specific data item, and a review date, or it becomes a Kill. Walk: stop at this tier on the current evidence. Hold: the case stands but the evidence has gaps, so the next step is a defined data request. Proceed: progress the target to the next tier. The difference between Walk and Kill is the difference between not now and not ever, and confusing the two is how dead deals come back.
You bring the target and yourself: the teaser or information memorandum, whatever financials the seller has released, and a plain statement of what you can distribute, operate and pay. The same business is worth different amounts to different buyers, so the reading is yours from the first minute.
The reading works through four lenses, four angles on the one target. Your investment posture frames the question. The financial and operational record carries the most weight, roughly forty per cent, which is a stated assumption rather than a measured result and is also reported at equal weights as a sensitivity check. What customers say about the brand comes next, read from reviews and market signal. The last lens asks what the target could earn in your hands specifically, given your channels and capability. Each produces a range, not a single number, and the ranges combine into one view.
What arrives is one Call: Kill, Park, Walk, Hold or Proceed, with a confidence level, a named owner and the date the call gets checked. The work is sized to the stakes, from a thirty-minute first pass to a forty-hour read before a letter of intent. You spend the cost of the decision, not the cost of the deal.
Illustrative composites. None is drawn from a client engagement: the figures, readings and instructions were constructed to show how the method behaves, and no audit of them has taken place.
The acquisitions literature is unusually honest and unusually ignored. Event studies have shown for decades that acquirers capture little of the value and vendors capture most of it, and the field responded by producing better estimates of the same finding rather than an instrument anyone would use on a Tuesday. Volume I is not a contribution to that literature so much as an attempt to get the verdict written down before the momentum makes it academic.
And commercially: it is cheaper to kill a deal at tier one than to explain it at year three.
Twelve months after a reading, the instruction is compared against what happened and the result is recorded, whether or not it is flattering.
The evidence base for Edition 3 is sixty anonymised 2026 buy-side screens, each scored blind under a decision rule frozen before any outcome was known. Edition 3 makes no accuracy claim. It reports how reproducibly the rule scores and how decisively it resolves, and it sets up a prospective test that tracks every frozen call to its outcome, with the first review in January 2027.
A reading under Volume I settles one question about one target: proceed, hold it for a defined data request, park it against a trigger, walk at this tier, or kill it for good. The output is a single instruction with a confidence level, a named owner and a date it gets checked. It starts small. You describe the decision, Rob reads whether the method fits it, and you hear back either way. Send one sentence: what is the decision you are trying to make?