The reviewer asked for one thing. She named a worker, gave a six-month window, and asked to see the supervision records.
The supervisor knew, with complete confidence, that the supervision had happened. Every other week, sometimes weekly when a case was difficult, in her office with the door shut. She could remember specific conversations — the placement that nearly disrupted in March, the documentation habit they had worked on through the spring, the week the worker had come in and said she was not sure she could keep doing this.
What she could produce was a calendar. Recurring appointment, every other Tuesday, some of them moved, two of them deleted. She found four handwritten pages in a notebook covering roughly the same period, none of them dated in a way that tied to a specific session.
The reviewer was not unkind about it. She wrote that supervision could not be evidenced for the period sampled, which was accurate, and which is a different finding from supervision not occurring — except that on paper, and in the plan of correction, and in the file that gets pulled again next year, the two are indistinguishable.
The gap is between doing and demonstrating
This is not a diligence problem. The supervisors who fail this sample are frequently the best ones, because good supervision is absorbing and the person doing it is concentrating on the worker in front of them rather than on the artifact the session should leave behind.
A supervision system is not a promise to meet. It is a set of records that exist whether or not anybody remembers the conversation. That is an uncomfortable reframing for people who came into this work because they care about practice, and it is worth being honest about why it matters: a supervision record is not for the reviewer. It is for the worker who transfers to a different supervisor in eighteen months, for the person investigating a complaint two years from now, and for the supervisor herself on the day she has to explain a decision she made when she was tired.
Three documents, and what each one settles
An evidenced supervision system needs remarkably little. Three things, and none of them is a form somebody invented to satisfy an auditor.
A session record with a standing agenda. Date, worker, duration, and a fixed set of headings that get covered every time — caseload review, cases of concern, documentation, development, worker wellbeing, actions agreed. The value of a standing agenda is not completeness. It is that the first heading is follow-through on the last session's actions, which is the thing that turns a series of conversations into supervision.
A documentation review rubric, scored per record. Supervision that never opens a case file is a conversation about how the worker feels the work is going. Pull a small number of records, score them against written criteria, and a pattern becomes visible that no amount of discussion would surface. A record scored below standard produces an action, or the review changed nothing.
An action log that carries between sessions. One line per action, with an owner and a date, and it opens the next session. This is the single cheapest change most supervision practices can make, and it is the one that most reliably stops the same thing being agreed and forgotten twice.
The sampling is the part people skip
Deciding to review documentation is easy. Deciding how much, on what schedule, and recording what was actually drawn is the part that separates a system from an intention.
Set the sample size and cadence once, in writing, against your own licensing standards and contract requirements — those vary, and they are yours to confirm rather than mine to assert. Then record what was sampled each month, including the months where the number fell short. A sampling log with an honest gap in it is more use to you than a policy with no log at all, because the gap is a fact you can act on.
What to do this month
Pick one worker and reconstruct the last three months of their supervision from your existing records alone — not from memory. Write down what you can evidence: dates, what was covered, what was agreed, whether the agreed thing happened.
Whatever you cannot reconstruct is what the next reviewer will also be unable to find. That exercise takes an hour, produces no paperwork, and tells you precisely which of the three documents above you are missing. Start the one that closes the biggest gap, and start it at the next session rather than at the start of a quarter.
The Supervisor's Field Kit is on the Open Shelf: the session record with the standing agenda already laid out, the documentation review rubric as a row per record with the six criteria as scored columns, the monthly sampling schedule, and the action log that carries between sessions. The cadence and sample sizes are marked for you to set against your own standards. No registration, no email address.