THE ADD-BACK · episode 04
SightingTwelve months out. The adjusted EBITDA schedule has an automation savings line that has been there three years without a hostile reader testing it.
The consensus readIt survived our own QoE, so it will survive theirs.
The mechanismYours was scoped to tie out what you presented. Theirs is paid to find what you didn't. An automation adjustment fails on the one criterion neither side can tie out from a schedule — whether the saving is offset by cost that moved elsewhere. There is no document that shows you a cost nobody named.
The exposureThe adjustment most likely to be disallowed is the one with no owner on the org chart, and it comes off the price after the LOI, when you have the least leverage you will ever have.
The testRun the offsetting-cost search yourself, twelve months early, while the answer is still a management problem rather than a retrade.
Exposure percentages are our own estimates applied to a composite schedule; the underlying pattern draws on the same composite case as Episode 01. Nothing here is a measured population statistic.
| Line | Moves | Running |
|---|---|---|
| Claimed — $1.4M of cumulative automation-related adjustments in the adjusted EBITDA schedule at T–12 | $1.4M | |
| 1. The three tests it passes changes the story Automation adjustments are usually well-prepared, which is why they get through internal review and why the disallowance is a surprise. Realised and committed: yes, roles were eliminated. Documented: yes, headcount schedules and termination records. Correctly characterised: yes, recurring savings presented as recurring. A seller's QoE is scoped to verify what management asserts; it ties out three of four criteria cleanly and reports the adjustment as supported. That report is honest, and it is not the test the buyer will run. | −$0 | $1.4M |
| 2. The fourth test, and why it is structurally different The fourth criterion is that the saving is not offset by incremental cost elsewhere, and it is different in kind: the other three are verification tasks against documents, this one is a search with no defined boundary. A buyer's confirmatory diligence runs it because they are paid to. The search is unglamorous — pull org charts across every adjacent function, look for headcount growth or reallocation in the twelve months following each automation, interview managers about what their teams actually spend time on now. What it finds reliably is the Episode 01 pattern: the eliminated work reappeared as absorbed load in departments that never asked for it and don't report it separately. Applied across $1.4M of adjustments, expect 30–45% exposed to this test. Nobody internally is paid to find an offsetting cost that reduces their own reported adjustment; the buyer is the first party in the chain whose economics reward looking. | −$400,000 to −$650,000 | $750,000 to $1.0M |
| 3. The attribution problem — 'influenced' savings A separate failure mode, and the one sell-side teams argue hardest and lose most often. Some portion of the adjustment will be characterised as savings the automation enabled or contributed to — downstream process improvements, error reduction, faster collections. The verb matters: an influenced number has no counterfactual, and a claim with no counterfactual cannot be tied to a document. Same structure as market sizing, where the widest definition produces the biggest number and gets quoted — here inside a single company's schedule, so as the boundary of 'attributable' widens the adjustment grows and its defensibility collapses. A buyer's team will not disprove these. They will decline to credit them, which achieves the same result and takes no work. The practical test, worth applying to every line yourself: could you write the sentence describing what would have happened without the automation, and support it with a document? | −$150,000 to −$250,000 | $500,000 to $850,000 |
| 4. Where I was wrong — I told a client to fight it where I was wrong A judgment error rather than an analytical one. On a sell-side preparation we found an automation adjustment we believed genuinely supportable — the offsetting cost was real but smaller than the buyer's team had estimated, and we had the time-allocation data to show it. My recommendation was to hold the line and defend the full adjustment with the analysis. It was analytically correct and strategically wrong. What it cost: The dispute consumed three weeks of a confirmatory diligence period, and the effect on the bidder was not that they accepted our number — it was that they began re-testing adjustments they had previously passed on, because a seller fighting hard on one line reads as a seller with something to protect. We won the line item and lost more elsewhere than the line was worth. The correction: at T–12, a defensible adjustment you have to argue for is worth less than a smaller adjustment nobody questions. Where offsetting-cost exposure is real but arguable, take the haircut in your own schedule before the process starts. You lose the delta and keep the diligence period short and quiet, which is worth more. That is not a conclusion I would have reached from analysis; it came from watching the second-order effect. | −$0 | $500,000 to $850,000 |
| 5. What survives residual $500,000 to $850,000 against $1.4M presented — and the range narrows substantially if you run the search yourself and reprice the schedule before the LOI rather than after. | +$0 | $500,000 to $850,000 |
| Realised | $500,000 to $850,000 against $1.4M presented |
Roughly 40–65% of presented adjustments exposed, split between offsetting cost and influenced savings. The range narrows if you run the search first.
Value at risk · multiple 9x, illustrative
$400,000–$650,000 of disallowed offsetting cost at an illustrative 9x is $3.6M–$5.9M off the price. Add the influenced-savings haircut and exposure runs higher. Substitute your own comps.
$4.5M to $8.1M off the price — and a retrade rarely stops at the value of the finding — one disallowed adjustment invites a re-test of the others
When it surfaces. Discovered at T–12 this is a schedule revision, a smaller number in the CIM, and a marginally lower asking price set by you. Discovered in confirmatory diligence after LOI it is a retrade, and the asymmetry at that moment is total: you are exclusive, your LPs have been told the process is closing, management has been told, the advisor's fee is contingent on completion, and the buyer has three weeks of sunk cost and no obligation. The cheapest possible time to find an unsupportable add-back is any time other than when a buyer finds it.
THE ADD-BACK · episode 04 · diligence pack
Artifact: Both schedules.
This is the offsetting-cost search and it is 80% of the exposure.
Artifact: Raw exports plus named list.
If the frequent responders sit outside the process's cost centre, that is the offsetting cost — and it is now documented by you, first.
Artifact: One sentence and one document per line.
Any line that cannot produce both is an influenced number; reprice or remove it.
Artifact: A schedule of the soft lines.
This is your haircut list.
Artifact: The invoices and the board materials.
A remediation is evidence the original implementation underdelivered, and a buyer will read it that way whether or not you present it.
Disqualifier
Sequencing: run 1 and 2 before your own QoE is scoped, so the provider can address offsetting cost in the report rather than being silent on it. A seller's QoE that doesn't mention offsetting costs on an automation adjustment is a report the buyer's team will treat as incomplete.
| Claim | Source | Sample | Class |
|---|---|---|---|
| Roughly 1.5–2.0 FTE of absorbed load across customer service and account management against a three-head reduction | Composite, built to the shape of a real quote-to-cash automation (same case as Episode 01) | n/a — composite | composite |
| QoE four-criteria test: realised or committed, documented, correctly characterised, not offset by incremental cost | Standard quality-of-earnings practice | n/a — practice standard | reported |
| 30–45% of automation adjustments exposed to the offsetting-cost test; influenced-savings haircut of $150K–$250K | Our estimate applied to a composite schedule | n/a — estimate, not a measured population statistic | illustrative |
| 9x entry multiple | Illustrative mid-market comp | n/a — illustrative, arithmetic exposed | illustrative |
One standing caveat. Every number in this show is somebody else's measurement, and I'll tell you whose, with the sample. None of it is diligence on your deal. Do that yourself.