The delivery record is not the floor: what a completion rate proves
Published Facts as of September 2026
Short answer
A completion rate proves that a procedure reached people: who was assigned it, who opened it and who finished it. It does not prove that anyone followed it on the floor. For that you need evidence tied to a location, a procedure and a date, such as a checklist, a reading or a photo of the finished task.

Multi-site organisations track which procedures went out, who attended training and who completed mandatory material. That record is useful: it shows what was sent, to whom and when.
The trouble starts when the same record is used to answer a different question: are people following the procedure at every site? A completion record shows that an instruction reached an employee. It does not show that the task was done correctly on Tuesday's late shift. Treat one as proof of the other and management ends up confident on the strength of data that measures communication, not work.
What a delivery record measures
SOP tracking starts with individual activity: whether material was assigned, opened, attended or marked complete. That answers one question: was the procedure communicated? It leaves another open: was it followed when the work was done?
Take a hygiene protocol. A learning system can confirm that every employee at Site A completed the hygiene module. It cannot show that anyone washed their hands during service, sanitised a food-contact surface properly or ran the procedure on a given shift.
The record-keeping job is real, and often required. Under Regulation (EC) No 852/2004, food business operators must make sure food handlers are supervised and instructed or trained in food hygiene suited to their work. The Belgian Act of 4 August 1996 on the well-being of workers obliges employers to inform and instruct their staff. The European Commission's HACCP guidance goes a step further and expects staff to show they understand the hazards, the critical control points and the corrective actions.
So training records belong in the control environment. They are not evidence that the procedure was carried out. A log showing that a kitchen team completed instruction in the second quarter proves that session took place. It says nothing about the third quarter.
A delivery record confirms that you distributed a requirement. An execution record shows what happened when the requirement met the work.
Where the inference breaks
The error happens when a delivery metric leaves the training system and enters an operations review, where it is read more broadly than it was built for.
A dashboard shows: SOP completion, 97 percent. That is correct if 97 percent of the assigned staff finished the material. But the review wants to know which sites are actually following the procedure, and the dashboard cannot say without evidence tied to the work at each site. This is not a badly designed KPI. The organisation changed its question and kept the old measurement.
Why the delivery metric stays
Delivery metrics survive because they do real work and are wired into the organisation.
- They meet a need. Companies have to show that procedures and training reached the frontline, and in regulated sectors that record is part of compliance.
- They add up easily. Head office can compare completion across hundreds of sites on one baseline. Evidence of what happened on the floor is harder to collect, because it has to be tied to physical work.
- They are embedded. Schedules, worker records and training files often depend on the existing platform. Replacing it touches integrations, data migration, file formats and four departments: IT, operations, HR and legal.
A Benelux food producer described exactly this. Its software was tied into internal systems and its instruction material was locked in platform-specific formats. Operations knew what the numbers could not show, but did not control every dependency needed to change them.
The barrier is technical, not a lack of understanding. The failure only starts when delivery data is presented as proof of compliance.
What inspectors look at instead
Put internal records next to what regulators find on site and the gap is visible. The FAVV ran about 22,000 inspections in hotels, restaurants and catering in 2024. Almost two-thirds ended favourably. The main concerns were accurate allergen information for consumers, hand hygiene and dirty food-contact surfaces. Those findings describe practice and physical conditions on the floor. No training platform sees them because employees completed a module.
| What the operations review asks | What the delivery record answers | What execution evidence would answer |
|---|---|---|
| Is this procedure followed at site X? | Material for it was assigned to staff at X | The required task was done at X on a stated date |
| Which sites are not compliant? | Sites with low registration, access or completion | Sites where recorded work misses the standard |
| Did the latest rollout reach the floor? | Staff opened or completed the material | The changed procedure was applied and checked at each site |
| Are specific sites exposed? | Sites with low delivery figures | Sites with recent failed checks or missing evidence |
| What changed after the update? | Delivery rates rose | The task or control changed at site level |
| Can we show compliance to a regulator? | Required instruction took place, where that is the rule | The control was performed, where the rule is about practice |
Inspection and internal reporting serve different purposes. The FAVV does not ask whether an employee has seen a procedure; it looks at compliance at the establishment. In the Netherlands the NVWA publishes inspection results for individual hospitality, craft and retail businesses. Your internal system does not have to copy an inspector. You do have to know what your own data can prove. A 98 percent completion rate can sit next to poor execution at individual sites; there is no contradiction, because the two numbers describe different events.
From communication to control
A procedure creates value only when it controls a process or changes behaviour on the floor. Communication is the first input, not the proof.
That is why more training data does not close the gap. Raising completion from 90 to 99 percent improves your view of participation. It tells you nothing new about the floor. Quizzes and digital acknowledgements improve engagement and accountability, but they still measure activity around the work, not the work.
Five requirements for evidence from the floor
1. The unit is the site
Employee completion answers whether people received the instruction. Adherence is about the place where the work happens. The reporting question becomes: what evidence do we have that site X performed procedure Z in this period? That needs site, procedure and date in the record.
2. The evidence shows the required work
The record should match the control, not whatever the software makes easy to capture:
- a cleaning procedure: a dated photo or a completed checklist;
- temperature control: the recorded reading;
- an opening routine: the checklist for that site and that day;
- maintenance: a completion record linked to the asset and the check.
More is not better. A thousand irrelevant photos add admin without adding control. The target is specific evidence of the required action, at the site where it happened.
3. The evidence is recent enough to act on
Site performance moves. A module finished six months ago cannot show the procedure was followed today, and a photo from last year does not show current practice. Management should be able to separate old evidence from recent evidence and see where it is missing or overdue. The question shifts from "have employees completed this SOP?" to "when was this control last evidenced at each site, and what did it show?"
4. It adds to the compliance record, it does not replace it
You still need proof that instruction and training took place; Regulation 852/2004 sets training duties for food handlers and for the people running HACCP procedures. So keep two linked records:
- the instruction record: who received which instruction or training, and when;
- the execution record: what work was done at the site, when, and what shows it.
The first protects your ability to show you communicated the requirement. The second shows whether it is being applied. Neither does the other's job.
5. It works with the procedures you already have
Better execution evidence does not justify rebuilding a whole procedure library. If your procedures, schedules and instruction material live in established systems, the execution layer has to work with them. In a large network the cost of changing the documentation stack can quickly exceed the value of the new measurement.
Questions for the next operations review
Ask one diagnostic question: what event produces the "SOP adherence" figure on the dashboard?
| If the figure counts that a worker... | it measures |
|---|---|
| was assigned the procedure, opened it, attended a session or marked it complete | distribution and training activity |
| performed procedure Z at site X on a stated date, backed by a dated photo, checklist, reading or verification log | execution |
KPIs should carry that distinction in their names. Renaming a completion rate "adherence rate" does not turn distribution tracking into adherence measurement.
What each record is for
A delivery record has real value: it shows that you sent the requirement and met your instruction duties. It becomes a problem only when it is read as proof of execution.
This matters for where management spends its attention. Steering on completion data alone points you at the sites with low digital participation, not the sites that are struggling on the floor. A site at 100 percent completion with repeated failed checks needs more attention than a site at 80 percent whose work is consistently evidenced. The first proves training was received. The second is stronger evidence that the procedure is being carried out.
Keep the compliance records you have and add one measure: what happened at each site after the procedure went out. Why that gap costs most at the one site that drifts is the subject of operational drift.
Sources
- Regulation (EC) No 852/2004 on the hygiene of foodstuffs, Annex II, Chapter XII (EUR-Lex)
- Commission Notice 2016/C 278/01 on food safety management systems, prerequisite programmes and HACCP (EUR-Lex)
- Act of 4 August 1996 on the well-being of workers in the performance of their work (Justel)
- FAVV/AFSCA, activity report 2024
- Food Safety News, "Belgium sees outbreaks increase in 2024", 12 August 2025
- Hogan Lovells, "Tougher enforcement of food law in Belgium", 13 February 2024
- NVWA, inspection results for hospitality, craft and retail businesses
Frequently asked questions
Read next
Operational drift: cost averages out across sites, consequence does notWhy variance on regulated procedures does not average out across sites, with Belgian and Dutch enforcement data, and what that means for multi-site operators.Blog · 25 Aug 2026
What is knowledge retention?Glossary · 2 min read
Training records an auditor can checkGuide · 7 min read
Keeping brand standards aligned across locationsGuide · 7 min read
