Three decades. Six hundred and forty engagements. Here is what six of them got.

Profitable on paper, and the surety had held the bonding limit flat for two years.

Work in progress was a percentage-complete estimate nobody had challenged, carried forward month after month. Retention was sitting unclaimed across eleven closed contracts. The schedule going to the surety did not reconcile to the management accounts.

We rebuilt costing at job level on a cost-to-cost basis, made the winning estimate the monthly benchmark, put every variation on a priced register, and gave the surety a WIP schedule on a fixed rhythm with the same numbers as the board pack.

Profitable for three consecutive years, and drawing on the operating line every month.

Cash was in two places and both were measurable. Stock had stopped moving on a long tail of lines that were still being reordered against settings from three years earlier. Landed cost excluded duty and freight, so the margin on imported lines was whatever the system said it was.

We decomposed the cash conversion cycle, aged stock by SKU against real turns, recalculated reorder points on actual demand, rebuilt landed cost to the item and repriced the lines that had been sold on a cost that excluded duty and freight, and produced customer profitability with cost-to-serve allocated on orders, deliveries and returns.

Every variance was explained away rather than investigated, and capital kept getting approved on payback.

The standard cost had been set when the plant ran a different mix, so absorption was spread over a volume the factory no longer produced. The constraint governing output was not the thing the reporting was built around, so the business was optimizing everything except the machine that decided the month.

Standards rebuilt against current routings and rates. Variances split into usage, rate, mix and volume. Contribution reported per constrained hour, and product mix, pricing and capital all judged against it. Capital cases rebuilt on realistic ramp with maintenance capital separated from growth.

Everyone was busy, utilization looked healthy, and the firm was not making money.

Utilization only measured whether time was recorded. Realization was eroding quietly through write-offs at billing and scope that had grown without a change order. Lock-up was climbing because nobody billed promptly and nobody chased.

We separated utilization from realization and reported both by client, by matter and by team so write-offs concentrated visibly. Scope discipline was made mechanical and reviewed monthly. Lock-up was tracked as one number with billing and collection run as a process.

The group number was fine, so the group number was the only one anybody looked at.

Two sites were carrying the estate and two were draining it, and the average concealed both. Labor was reviewed as a percentage after the fact rather than scheduled against forecast demand. Two lease break dates had passed unnoticed in the previous eighteen months.

Every site onto a four-wall P&L on the same basis with central cost shown separately, ranked monthly with the bottom two discussed by name. Labor scheduled against forecast covers. Lease economics put on a calendar with break and renewal windows dated, and two underperforming sites renegotiated at renewal.

Growth was being bought and nobody had checked the price. A raise was eighteen months out.

Acquisition cost was calculated on marketing spend alone, so the real payback was nineteen months rather than the eleven the board believed, and the cohort curves that would have shown it had never been built. Revenue recognition on multi-year contracts and set-up fees would not have survived a diligence review.

Cohorts built by signing period so retention became a measured figure, then the two segments dragging it down were repriced and one was retired. Acquisition cost loaded with the sales cost that earns it, and spend moved off the two channels that were paying back past twenty months. Revenue policy fixed and documented before anybody else tested it. Runway modeled on committed cost with the decision date named nine months ahead of the cash-out date.