The symptom
Consider a franchise operation where the documented procedure for a routine task takes roughly three times as long as the method staff actually use day to day. Both produce the required result. The documented version was written for a system that was retired four years ago; the practiced version evolved, informally, as staff adapted to the replacement system without anyone updating the manual to match.
Nobody is hiding this. Ask any experienced site staff member how the task actually gets done, and they’ll describe the shorter, current method without hesitation. The documented procedure simply isn’t what anyone consults anymore — new hires learn the real method from a colleague within their first week, and the manual becomes something referenced only during an audit, if then.
What leaders usually conclude
The common response is to treat this as a documentation failure — the manual is out of date, and the fix is a refresh project to bring it back in line with current practice. This is a reasonable enough starting point, but it treats the gap as a one-time lapse rather than what the evidence suggests it actually is: a structural, recurring feature of how procedures and practice relate to each other over time. A refresh project, run once, fixes the gap as it currently stands — and does nothing to prevent the same gap reopening gradually over the following few years, for exactly the same reason it opened the first time.
What the evidence suggests happens to routines
Martha Feldman and Brian Pentland’s research on organisational routines, published in Administrative Science Quarterly in 2003, offers a precise account of why this gap opens and keeps reopening. Their central distinction is between what they call the ostensive aspect of a routine — the abstract, documented version, the one written down and referred to as “how this is done” — and the performative aspect: the actual, specific actions taken by specific people, at specific times, that constitute the routine as it really happens.
Feldman and Pentland’s key finding is that these two aspects are not simply the same thing described twice. The performative aspect doesn’t just execute the ostensive one; it actively modifies it over time, as people adapt the practiced version to circumstances the written version never anticipated — a system change, a local constraint, a shortcut that turns out to work better. Each adaptation is individually small and rarely written back into the documented version, because updating the manual was never anyone’s specific job in the moment the adaptation happened. The practiced version keeps evolving. The documented version, absent a deliberate act of maintenance, does not.
This reframes the retired-system example directly. The four-year gap isn’t evidence that nobody cares about documentation — it’s the predictable result of a system change that triggered dozens of small, sensible adaptations in the performative routine, none of which were individually significant enough to justify stopping and rewriting the manual, and which simply accumulated into a substantial gap over time.
There is a further implication in Feldman and Pentland’s framework worth naming directly: the drift isn’t a sign that the routine is failing. In their account, this same ongoing gap between the ostensive and performative aspects is precisely what allows a routine to adapt and improve at all, rather than staying rigidly fixed to conditions that no longer hold. A franchise network with zero drift between documented and practiced procedures across the board wouldn’t necessarily be a well-governed one — it might simply be one where nobody has adapted anything in response to changing conditions, which carries its own, different risk.
Why wide divergence is the normal state, not an anomaly
It’s worth being explicit about what this means practically: a documented procedure that has been in place for several years, covering a task that has continued to evolve, should be expected to have drifted from actual practice to some degree. The absence of any drift would be the more surprising finding, not its presence.
The genuine risk isn’t that drift exists. It’s when the two versions diverge enough that the documented procedure actively misleads — a new hire following it produces a worse result, or an auditor checking against it flags a compliant, well-reasoned practice as a deviation, because the standard being checked against is the outdated one rather than the one that’s actually functioning. The practical distinction worth drawing is between drift that’s simply unrecorded and drift that’s actively harmful: unrecorded drift is a maintenance backlog; harmful drift is a live operational risk, and the two call for different degrees of urgency.
Where this shows up
People working in multi-site businesses will probably recognise this pattern first. The wider Multi-Site and Franchise grouping and the other sectors page show how it presents elsewhere.
Questions worth considering
- For your most-used documented procedures, when was each one last checked directly against what staff are actually doing — not assumed, checked?
- If a new hire followed the written procedure exactly, would they produce a worse result than someone trained informally by a colleague?
- Is there a system, process, or tool referenced in a current procedure that has since been replaced, without the procedure itself being revisited?
Your team is facing similar challenges?
Subscribe to The Riddle Ledger
Perspectives are sent as they are published, with no promotional email, and you can unsubscribe at any time.