Why Franchise Training Completion Rates Don't Predict Competence

Franchise training completion rates tell you an assignment was closed on time. They do not tell you whether the person can run a shift. Competence surfaces weeks later, in remakes, refunds, a close that runs long, and two locations answering the same customer question differently.

Every operations review I sit in on has a training slide, and it is almost always green. Completions in the nineties, a red cell or two, an action item about chasing three locations.

Then someone from field support mentions the new cohort of assistant managers is struggling, and nobody connects the two facts, because on paper they contradict each other.

What the completion number is actually measuring

Completion is a real metric with a legitimate job, worth being precise about before dismissing it.

What a completion genuinely proves:

• The content was delivered and was technically accessible.
• Somebody was assigned it, so somebody thought about who needed it.
• You can produce a record on request — for an insurer, a food-safety programme, a franchisee dispute.
• Where completion is low, something is broken. A location at a third of its assignments is telling you something true.

What it cannot prove is that anything changed on the floor. It is an input metric wearing an outcome's clothing, read as an outcome because it is the only training number most brands can produce.

There is a family resemblance to another franchise measurement problem. FranConnect's 2021 operations index found franchisee self-reported scores pulling 33% further from actual audit scores during 2020. Same structural flaw: the party being assessed supplies the assessment. A trainee clicks their own completion; a location scores its own compliance. Neither number is a lie. Both answer a question about paperwork.

Why franchise training completion rates run high and tell you little

Look at how the number is produced and the ninety-odd per cent stops being reassuring.

The completion event is a click, not a demonstration. A module finished twenty minutes before the deadline, on a phone, with the audio muted, completes identically to one worked through on a slow Tuesday with a manager nearby.

Shared logins are normal. In a location with one back-office computer and a shift lead who wants the red cell gone, somebody clicks through somebody else's assignments. Everybody in operations knows this; it appears in no report.

Recall is not performance. Multiple choice at the end of a module tests whether the answer is still in short-term memory, not whether a person can hold the standard at 12:15 with a queue out the door.

The metric moves when you change the denominator. Split a long module into three short ones and completion improves. Retire a module nobody finishes and it improves again. No skill has changed.

Once it is scored, it stops carrying information. Put completion on a field coach's scorecard and you have not created training; you have created a reason to close assignments.

None of this argues for abandoning the metric, only for demoting it: franchise training completion rates are a hygiene check worth reading when they are low, not evidence worth citing when they are high.

What real competence looks like in the weeks afterwards

Competence is observable, just not in the LMS. The signals worth a habit are downstream of the training and cheap to collect.

1. Time to first unsupervised shift, and whether it slips. A manager who keeps delaying it has made an assessment. Ask why, and you learn more than a completion report contains.
2. Remakes, voids and refunds in a new person's first weeks. Not to police anyone — to see whether the curve bends. A flat curve at week five is a training gap, not an attitude problem.
3. The shape of the questions a new hire asks. Volume that drops to zero in week two is usually not mastery. It is someone who has decided asking is expensive.
4. Answer variance across locations. Ask the same operational question at three sites and compare the replies. It takes an afternoon and it is the most honest training audit in franchising, because the divergence names the paragraph of the manual that reads two ways.
5. Whether the person is still there in ninety days. BLS JOLTS puts total separations in accommodation and food services at 5.5% a month across 2025, against 7.1% in 2021. Cooler than the peak, still the largest recurring training bill in the network — and every departure resets a location's competence to zero regardless of what its completion rate said.

The observation that costs nothing

The strongest competence check I know takes ninety seconds and no software. Watch one person do one thing.

Make the drink. Run the refund. Do the pre-close walk. Handle the phone order while two guests wait. One task, watched by a peer or shift lead, marked fine, coach-again, or re-teach.

Three rules keep it useful. It has to be the real task in real conditions, not a demonstration staged for a visitor. It has to be quick enough to happen weekly, because an assessment needing a scheduled hour will happen twice. And it must stay out of the compliance file — the moment an observation can be used against a location it becomes a formality, and you have manufactured a second completion rate with extra steps.

The output is a short note about what needed re-teaching, never a score — the most valuable training-design input you will get, because it names the step that did not survive contact with a Saturday.

What to do with the number you already have

Keep collecting it. Then pair franchise training completion rates with one outcome from the list above, and never show either of them alone.

The obstacle is structural rather than analytical. Completion lives in the LMS, remakes in the POS, refunds in accounting, the new hire's start date in payroll — four systems that do not meet, which is franchise software tool sprawl doing its quiet work on a question you actually needed answered. At most brands the LMS is also the only training instrumentation in the franchise technology stack, so the metric it happens to emit becomes the metric the business manages.

If your brand is planning an AI trial, this is a better target than most of what gets proposed. The question log, the observation notes and the answer-variance check are text, already being generated, and reading them is hard for a person and not for a model — more defensible than most of what passes for AI adoption in franchising. It also meets people where they already are, because the trainee who cannot get an answer asks a chatbot anyway — the real question behind whether operators should use a public chatbot.

A green training slide next to a struggling cohort is the metric working exactly as designed, answering a question about assignments in a meeting about skill.

Underneath the green slide sits what most brands are actually running.