When a clinic’s technology invoice jumps and nobody can explain it, the symptoms tend to be the same: a surprised finance lead, a defensive technical team and a vague promise to look into it. The underlying cause is rarely mysterious. Variable technology spending needs a method, and the method has a name, FinOps. This piece takes eight common symptoms in turn, names the likely cause behind each and sets out a fix, in the problem-and-remedy format an operations manager can use as a checklist. It covers administration and platform costs only; there is no clinical content here.
The diagnoses draw on the Ops Atlas, whose long entry, the Ops Atlas reference on FinOps, catalogues the standard anti-patterns. Cases below are generic illustrations, not reports about any real organisation.
Symptom 1: the bill rose and everyone assumes waste
Likely cause: judging the invoice by its size instead of its unit cost. A practice that handled more appointments naturally used more booking and messaging capacity.
Fix: choose one business output, such as consultations booked, and divide spend by it each month. The Atlas notes that a bill can rise thirty percent and still be healthy if orders rose sixty. Only if the cost per unit climbs is there a real problem to investigate. Follow that figure over time inside one service; comparing it with unrelated services misleads.
Symptom 2: nobody can say who owns a charge
Likely cause: allocation was left to labels alone. The Atlas points out that labels alone cannot finish the job: support plans, levies and charges for shared platforms have nowhere to carry a tag.
Fix: keep the labels, and add a written rule for shared items. One workable rule splits the unlabelled remainder pro rata to what each team has already labelled. Where the large provider’s compliance reporting is used, remember it skips resources that were never tagged at all, so a clean report may hide exactly the items you most need to see.
Symptom 3: a surprise appears on the invoice weeks late
Likely cause: the only review happens when the invoice arrives. An invoice may show up 3 to 12 days after month end, per the Atlas, so any error may be days old before it is spotted.
Fix: use the daily figures, which the Atlas calls adequate for decisions, and set up anomaly alerts early. One provider’s alert service needs about ten days of history and can lag by up to a day, so the time to enable it is before you need it. Add a five-minute daily scan by a named person.
Symptom 4: the team bought a discount and the savings never showed
Likely cause: the commitment was sized to a number the vendor suggested, or to the pre-cleanup footprint. The Atlas names one trap in particular. Size a machine-specific commitment to the estate as it stood before the clean-up, and you are paying for capacity that the clean-up has since reshaped away. A related trap is shrinking machines after the purchase, which strands the pledge.
Fix: trim first, buy second, and buy against the steady hourly floor of what remains. Subtract planned changes. Purchase in tranches and check daily utilisation in the first weeks. Judge the result by the effective savings rate, not by the share of usage covered. And do not add the two types of recommendation together, since that counts some savings twice.
Symptom 5: the forecast keeps missing
Likely cause: the model extends history and ignores events. A new service line, a seasonal campaign or a software launch is invisible to a trend line.
Fix: add a short list of known upcoming events to every forecast, each with an estimated effect. Hold a reserve sized roughly to your recent forecast error, and review misses by splitting them into volume, rate and scope. Over time, move toward models driven by real business drivers.
Symptom 6: the cost conversation turns into a blame session
Likely cause: a central group acting as cost police, issuing orders without sharing data. The Atlas files this under anti-patterns, adding that the people who run the systems carry out most usage reductions regardless.
Fix: send each team a statement of its own spend, which the Atlas calls showback, and let the team decide how to respond within agreed service guardrails. Showback is always needed; whether to charge costs back into budgets is a policy decision, and neither is a more advanced stage than the other.
Symptom 7: the dashboard exists but nothing changes
Likely cause: data without decision rights. Someone built a handsome report, yet nobody in the room has the authority to act on it, or the room never meets. The Atlas expresses accountability as data combined with decision rights and a cadence; with any of the three at zero, the product is zero.
Fix: name who may change what. List five typical decisions, such as switching off an unused environment, resizing a database or adjusting a log retention period, and write beside each who can approve it and how quickly. Then book the meeting. A fortnightly thirty-minute slot with the three right people beats a monthly gathering of twelve.
Symptom 8: someone bought software and expects it to solve everything
Likely cause: treating a purchase as the practice. Reporting tools, policy engines and estimators all have a place, and the Atlas names a few, but it also lists the tool-as-solution belief among its anti-patterns.
Fix: after the purchase, hold a review and ask which decision each screen is meant to support. If a screen supports no decision, retire it. Tools earn their keep when they shorten the loop between spotting a problem and fixing it.
A differential diagnosis table, in words
- Bill up, unit cost flat: growth, usually healthy.
- Bill up, unit cost up: price change or waste; split usage from rate.
- Bill flat, savings promised but missing: check commitment sizing and utilisation.
- Large unexplained remainder: allocation gaps; apply the sharing rule.
- Repeated surprises: cadence is too slow; move to daily checks.
The protective rule that applies throughout
Whatever the symptom, the Atlas is firm that savings must not erode service quality. For a facility whose patients rely on its booking and records systems, that means the guardrails come first: agree which availability and response measures are protected, alter just one setting per pass, let a complete working cycle pass before judging, and reverse the change if any measure degrades.
Related reading from this blog
Money discipline at a personal scale is covered in a festive-season spending plan built around a budgeting app, which follows the same rhythm of plan, track and review. Patients researching procedures can also browse an informational article on stretch mark removal options; it is unrelated to the cost topic here, and nothing in this article is medical advice.
Run through the eight symptoms once, tick the ones that apply and repair the cheapest first. In many organisations, two of them account for most of the surprises, and fixing those buys the credibility to tackle the rest.