/Aristotl
Language
Demo

Operational drift: cost averages out across sites, consequence does not

Published Facts as of August 2026

Short answer

Operational drift is the slow gap between a documented procedure and what a site actually does. For cost-type variance such as labour or energy, gaps average out across a portfolio. For procedures with regulatory, legal or customer-facing consequences, they do not: the organisation's exposure sits at its worst site, not its mean.

A row of identical restaurant kitchens along one street; in the middle one, an old paper sheet hangs next to a newer one.
Every site runs the same policy. One of them is working from last month's sheet.

In a multi-site organisation, variance between locations is expected, budgeted and managed at portfolio level. One site runs three points over its labour target, another uses more energy, a third holds too much stock. The numbers move, management sees it and decides where to step in.

That logic breaks when the variance sits in procedures tied to regulatory, legal or customer-facing obligations.

What operational drift looks like

Operational drift builds up through small daily decisions and a slow disconnect between the documented procedure and the floor. It compounds from workarounds nobody logged, repeated shortcuts and improvised steps.

On the floor it looks ordinary: a manager prints a local reference sheet, a supervisor passes an update on verbally at handover, a team member works from memory during the rush because the current manual is out of reach. When the old and the new instruction exist side by side, both are treated as standard practice.

None of this moves a monthly summary. Each one creates exposure. Head-office tracking can prove that a procedure was distributed; it cannot prove what each location did with it.

Cost-type and consequence-type variance

Two kinds of variance live side by side in a multi-site network, and they do not produce the same kind of loss.

Cost-type variance is proportional: it scales with the size of the deviation. A site three points over its labour budget merges into a portfolio metric, and scheduling tools, budgets, dashboards and monthly reports exist to spot, investigate and correct exactly that.

Consequence-type variance does not spread or dilute:

  • An allergen incident at one branch is not reduced to a fraction of a risk because 29 other sites operated safely.
  • A contamination event at one plant is not offset by high compliance elsewhere in the company.
  • A working-time violation involving one temporary worker remains an exposure, whatever the thousands of compliant placements across the agency's network.
Two rows of eight site bars. On the left, costs scatter around a portfolio average. On the right, seven sites sit near zero and one spikes, while the average stays low.
Cost variance averages out across sites. Consequence variance does not: the exposure sits at the one site that drifted.

Belgian enforcement data shows what that means. The federal food chain agency, the FAVV, inspected about 58,000 supply-chain companies in 2024. The overall compliance rate was 85 percent. It still issued 9,931 fines, 1,673 seizures of goods, 467 temporary closures and 283 product recalls, up from 254 the year before, largely because allergen information was missing from the label. In hotels, restaurants and catering it ran about 22,000 inspections; accurate allergen information, hand hygiene and clean food-contact surfaces were among the most cited non-conformities.

If one location in a 230-unit chain fails to apply an updated standard, the compliance of the other 229 kitchens offers no protection against the enforcement action or the safety incident that follows. Portfolio averages stay useful for financial tracking. They cannot answer the safety question: which location is exposing the business right now?

Table comparing cost-type and consequence-type variance on what drifts, how it aggregates, management surface, visibility, failure cost, regulatory exposure, evidence lag and staffing.
Cost-type variance is managed by averages. Consequence-type variance is set by the worst performer and discovered on the regulator's schedule.

The case for tolerating variance, and where it stops

Operations directors have a valid point when they resist demands for full standardisation. Variance is built into planning: a regional director with twenty restaurants expects labour efficiency to differ by location, a manufacturer expects energy use to differ by plant, a retailer plans for local differences in stock, staffing and shrinkage. Trying to eliminate variance in those categories often costs more than the variance itself, and every extra point of consistency costs more than the last.

That principle holds as long as the risk scales with the performance gap. It stops holding when one location's failure produces a categorical consequence. A site over its labour budget does not create a regulatory breach. A plant running an outdated contamination protocol does. A staffing firm can run 9,000 compliant placements and still be fully exposed by one non-compliant placement on working time.

Where drift sits on the floor

The procedures themselves often look minor: a paragraph in a manual, a hygiene checklist next to a line, a scheduling rule in a staffing protocol, an allergen sheet behind a counter. Reading them takes minutes. The risk attached to them is set by law.

Food production. At the start of the morning shift, a supervisor works from a laminated hygiene sheet next to the equipment. Quality updated the policy in the system two weeks ago, but the copy on the floor was never swapped. Operators reasonably follow the instruction nearest their workstation, and two versions stay live at once. If something goes wrong, the question will not be whether head office updated the policy, but which version the line followed.

Hospitality. A cook needs to check an allergen before a dish goes out. Across three sites of the same group, that plays out three ways:

  • Site A: the cook opens current, accurate documentation.
  • Site B: an allergen sheet from the previous menu is still pinned at the prep station.
  • Site C: the cook has to stop and ask a shift lead, because the current version sits in a back-office folder.

Same policy, three different operating realities.

Staffing. A temporary worker arrives at a client site. The agency has updated its rules on working hours and mandatory training; the client's supervisor still works from an old site procedure. The worker sits inside two organisations, and neither sees the other's day-to-day execution. If hours or instructions are questioned later, an updated policy in the agency's system does not show what happened at the client site.

Retail and other frontline settings. Head office updates a procedure, the store manager sees it immediately, the shift team the next day, and the old printout stays on the staff board for weeks. That gap looks negligible until the procedure carries a customer-facing, safety or statutory obligation. The head-office policy is compliant on paper; the floor is not.

One placement, two organisations

Staffing adds a harder form of consequence-type variance, because legal accountability and day-to-day control do not line up. The agency is the employer, the client controls the workplace, and a failure on working hours, unsafe instructions or an unfollowed protocol creates liability for both at once.

The Netherlands is tightening the regime around this. Under the Wet toelating terbeschikkingstelling van arbeidskrachten (WTTA), from 1 January 2027 organisations that supply workers need an official authorisation, a EUR 100,000 security deposit and an entry in a public register; the Nederlandse Arbeidsinspectie enforces it in full from 1 January 2028. Its own figures show the scale: 2,722 accident investigations out of 4,306 reported accidents in 2024 and 3,044 out of 4,807 in 2025.

For a staffing provider, keeping internal procedures current is not the hard part. Seeing how they are executed at third-party client sites is.

The enforcement clock in Belgium and the Netherlands

WhenWhat changesWhere it bites
1 January 2024The FAVV fines directly instead of proposing settlements; unpaid fines can lead to seizure of assets and bank accounts by bailiffsEvery food-chain site in Belgium
2025Consumer complaints to the FAVV rose 20 percent to a record 6,268, more than half leading to enforcementRestaurants, bakeries, food shops
November 2025A public food-hygiene rating portal shows inspection outcomes to consumersAny site a customer can look up
1 January 2027Belgian employers need an objective, reliable working-time registrationEvery employer in Belgium
1 January 2027The WTTA authorisation applies in the Netherlands, with full enforcement from 1 January 2028Every organisation that supplies workers

Together, these timelines put one question to operations leaders: can you verify what was done at the exact location where compliance was required? Confirming that a procedure was written, filed in a quality system or sent to all 40 locations is not enough. The gap sits between distribution and the floor, and it is the subject of the delivery record is not the floor.

Sources

  1. Food Safety News, "Belgium sees outbreaks increase in 2024", 12 August 2025
  2. Lexology / Hogan Lovells, "Tougher enforcement of food law in Belgium", 13 February 2024
  3. Daldewolf, "Infringement of food law in Belgium: the new sanctions regime explained", 8 March 2024
  4. Brussels Times, "Food poisoning deaths double in Belgium despite 180,000 food safety inspections", 22 June 2026
  5. FAVV, Food Hygiene Rating portal
  6. Nederlandse Arbeidsinspectie, Jaarverslag 2024
  7. Nederlandse Arbeidsinspectie, Jaarverslag 2025
  8. Business.gov.nl, "Suppliers of personnel must be authorised (WTTA)"
  9. VBK Advocaten, "Dutch Provision of Personnel Accreditation Act (WTTA) adopted", 14 November 2025
  10. Baker McKenzie, "Belgium: higher fines for Belgian labor law offenses", 30 January 2026
  11. Partena Professional, "Compulsory working time registration in Belgium from 1 January 2027", 8 December 2025

Frequently asked questions

Operational variance is the expected spread in performance between locations. Operational drift is a gradual, unnoticed divergence between documented procedures and what a location actually does. Drift matters most where a failure has a regulatory, legal or customer-facing consequence that does not scale with the size of the gap.

The federal food chain agency, FAVV/AFSCA, inspects food-chain businesses nationwide. Since 1 January 2024 it can impose administrative fines directly instead of proposing settlements. In 2024, across about 58,000 inspected companies, it issued 9,931 fines and ordered 467 temporary closures. A public food-hygiene rating portal opened in November 2025.

Each placement can expose both the staffing firm and the client site. In Belgium the multiplier for social criminal fines rose to ten in February 2026. In the Netherlands the WTTA applies from 1 January 2027, with mandatory authorisation, a EUR 100,000 security deposit and a public register for organisations that supply workers.

Consequence-type variance disappears inside portfolio reporting and usually surfaces when an inspector arrives or an incident happens. Finding the outlier first takes evidence per location and per procedure, not a network average.

Read next

Ready when you are

Raise the standard at every location.

See how Aristotl fits into your operations stack in a 30-minute call.