Where consultancy WIP leaks: fee burn, unclaimed variations, ageing claims

WIP is unbilled billable time valued at charge rates: work the firm has done that no invoice yet covers. On the balance sheet it looks like an asset. Operationally it is a perishable one — every week between doing the work and billing it, three things decay: the client’s memory of why the work was needed, the firm’s ability to evidence it, and the social ease of asking to be paid. Write-offs are rarely decided; they accrete, week by week, in that decay.

Which is why the useful question is not “what is our WIP balance” — every practice system can print that — but “where does our WIP go to die”. For consulting engineering firms the answer is remarkably consistent: it dies in overrun phases, in variations that never became claims, and in the gap between claim cycles. Three leaks, three mechanisms, three fixes.

Fixed fees fail at phase level, not project level — and that distinction is where most tracking goes wrong. A project 60% burnt against 60% claimed looks healthy while concealing a detailed design phase at 130% of its fee, subsidised on paper by a construction phase that has not started. The project-level view averages the problem away precisely until the healthy phases run out.

The mechanism of the overrun is usually not estimating error. It is the review cycle: the second and third design resubmissions to a council or water authority, each consuming senior hours the fee assumed once. Resubmission cycles have a signature in the timesheets — the phase keeps drawing hours after its planned drafting window closed — and that signature is visible weeks before the month-end report, if anyone is looking at burn per phase in real time.

• Budget hours and fee per phase, not per project — the phase is the unit that fails.
• Burn must update at timesheet entry, not at month-end journal — the gap between those two is the entire management window.
• A phase drawing hours after its planned window is the earliest overrun signal you will get; treat it as an alarm, not a curiosity.
• An overrunning phase is either absorbed (a priced decision) or converted into a variation conversation — the one unacceptable option is not noticing.

The unclaimed variation is the most human leak. The client asks for “a quick look at” an alternative layout mid-design; the engineer, who is helpful and busy, does it. Raising a variation feels like an act of conflict in a relationship business, so the work is done first and the paperwork intended later — and later, the leverage is gone and the memory is fuzzy. The work was real, the cost was real, and the fee never existed.

The fix is a register with a deliberately low-friction front door. The moment extra scope is identified — before the work is done — it becomes a one-line record: identified. Pricing it, submitting it, getting it approved are separate steps with their own states. The psychological point of the identified state is that it decouples capturing the variation from confronting the client; the commercial point is that a variation identified before the work starts is a negotiation, while one raised after is an invoice dispute.

The management number that matters is the funnel: variations identified versus approved versus claimed. A firm that only ever records approved variations has no funnel — it has a survivor list, and no way to see the goodwill work bleeding out at the top.

The third leak is pure latency. Work done early in month one, claimed at the end of month two, paid on thirty-day terms — the firm has funded that work for a quarter of a year before a dollar lands, and that is the well-run case. The badly-run case adds a claim assembled late because assembling it is a chore of reconstruction from three systems, then disputed because the narrative of the work has gone cold.

Late claims and disputed claims are the same defect at different ages. A claim raised promptly from a live record is small, recent and recognisable to the client; a claim reconstructed at quarter-end is large, stale and adversarial by its sheer size. The discipline compounds: firms that claim monthly from a live WIP position carry structurally less risk per claim than firms that claim heroically.

• The claim should be an output of the record — timesheets and WIP feeding it — not a document assembled from exports.
• Every claim carries a payment due date; overdue is a flag the system raises, not a discovery accounts makes.
• Approved timesheet weeks should lock — a billing record that can be silently edited after the fact is not a record, it is a draft.
• Reconciliation against the accounting system is a job for integration, not for a person with two screens open.

All three leaks are the same defect wearing different clothes: the system that knows the work and the system that knows the money are different systems, reconciled by humans, on a lag. Fee burn hides because hours and budgets live apart. Variations evaporate because scope lives in email. Claims age because assembling them means visiting every system the firm owns. Process reminders and better spreadsheets treat the symptom; the cure is structural — one record where the hours, the phases, the variations and the claims are the same data.

If you want to put numbers on your own leaks, the Engis ROI calculator estimates what under-billing, slow claims and missed variations cost a firm of your shape — it takes about two minutes with figures you already know.