The symptom
Consider a franchise network where site visits are scheduled and announced well in advance, and have been for years. Results are consistently favourable — sites pass, standards look well maintained, and the network’s own confidence in its consistency is, in part, built on this steady record.
What the record doesn’t distinguish is the difference between a site’s ordinary operating state and its prepared state. A visit announced in advance gives a site time to clean, restock, brief staff, and correct anything visibly out of place before the visitor arrives — all of which is reasonable, expected behaviour, and none of which tells the network much about what the site looks like on an ordinary Tuesday with no visit scheduled.
Years of consistently strong results can, in this light, mean two quite different things: either the site genuinely operates at this standard continuously, or the site has become reliably good at reaching this standard specifically when it knows someone is coming to check. From the visit record alone, these two possibilities are indistinguishable — both produce exactly the same pattern of favourable outcomes.
What the evidence suggests
The phenomenon of behaviour changing specifically because it’s being observed is well documented under the name the Hawthorne effect, and one of the more precisely quantified demonstrations of it comes from healthcare hygiene research. In a 2014 study in BMJ Quality & Safety, Srigley and colleagues used an electronic monitoring system on two units of an academic acute-care hospital over eight months to record how often staff used hand-hygiene dispensers when an auditor was visible and when none was. Use was roughly three times higher while the auditor was in view: the same staff, the same protocol, a genuinely large difference driven entirely by the presence of observation itself.
Nobody in a setting like this needs to be deceiving anyone. Staff simply, and largely unconsciously, behave differently while being watched, which is what research on observation effects would predict.
Applied to an announced site visit, the mechanism is the same, simply given more advance notice to act on. Where an unannounced observer captures behaviour changing in the moment, an announced visit gives a site days or weeks to prepare — which doesn’t create a different effect so much as a considerably larger and more thorough version of the same one. The site being observed is not lying to the visitor. It is showing the visitor a genuine, achievable state — just not the one that’s actually running most of the time.
It is tempting to assume that only struggling sites have a reason to prepare heavily for a visit. In practice, preparation is not confined to weak sites: a site that is already performing well has every reason to show itself at its best when a recognised visitor is coming.
Why this doesn’t require anyone to be deceiving the visitor
It’s worth being direct about what this pattern is not. Site staff preparing for an announced visit are not typically engaged in a deliberate cover-up — they’re doing exactly what any reasonable person would do before a scheduled assessment: making sure things look their best. The issue isn’t individual dishonesty. It’s a structural one: a visit that’s always announced can only ever measure a site’s best achievable state, never its typical one, and a network relying entirely on that measure has less visibility into ordinary operations than the consistently favourable results suggest.
This also means the fix isn’t about catching anyone out. It’s about recognising that announced and unannounced observation are measuring two genuinely different things, and that a network wanting a complete picture needs some exposure to both — not because the announced version is dishonest, but because it was never structurally capable of showing anything other than the prepared state in the first place.
Where this shows up
The pattern described here is most recognisable in multi-site businesses, part of the wider Multi-Site and Franchise grouping. The same organisational questions appear, under different names, in other sectors.
Questions worth considering
- Across your network’s recent site visit history, has a visit ever been unannounced — or has every one given the site advance notice to prepare?
- If results have been consistently strong for years, would that consistency itself be different if a portion of visits were unannounced?
- Is there a way to observe ordinary operating conditions — through data, informal check-ins, or unscheduled visits — that doesn’t rely on the same advance-notice pattern as the formal inspection?
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.