Course Completion vs PVIFR: Four Tests for Mobility KPIs

An employer academy can post an 80% completion rate across its course catalog and still record zero proficiency-verified internal fills in the same quarter. That gap is the defining measurement failure of 2026 mobility programs: completion behaves like a friction gauge, climbing when modules get shorter and navigation gets looser, while revealing nothing about whether anyone became capable enough to move into an open role.

As the Pedowitz Group frames it in its August 2026 guidance, completion answers 'Did they finish?' — a leading indicator of adoption and exposure, not skill. Its best-practice model separates three layers: completion as participation, proficiency as capability, and performance as business impact, validated with cohorts and controls. High completion can still produce low impact when content isn't relevant, coached, or embedded into daily workflows.

That is why completion belongs in content-QA telemetry, not the headline. The replacement is PVIFR — the share of internal fills backed by verified proficiency — which any academy can baseline in a single quarter using scenario-based assessments, certification rubrics, and CRM behavior signals. One quarter of disciplined instrumentation converts a vanity metric into a genuine mobility forecast.

Seat Time Isn't Signal

A SCORM 1.2 package hands your LMS one portable bit about a learner: cmi.core.lesson_status, set to completed or not. SCORM 2004 splits that into cmi.completion_status and cmi.success_status and adds an optional raw score — and still offers no field for applied behavior. There is nowhere in the record to store whether the employee ever performed the skill on live work. That structural emptiness is why Gloat, Eightfold AI, and Fuel50 match people to internal openings on verified-skills profiles and ignore LMS completion flags entirely: your academy's finish rate is invisible to the engine that decides who gets surfaced for a role.

Here lies the myth worth killing outright: the belief that if they finish the curriculum, they're ready to move — that a 90 percent finish rate means the academy is feeding the pipeline. According to Christensen et al.'s 2013 analysis of MIT/Harvard edX data, average completion across 69 MOOCs was 3.13 percent, and institutions moved that number dramatically by changing course length and navigation friction while underlying competence stayed untouched. Completion responds to module design and auto-advance settings, not to promotions, transfers, or internal fills. It is a friction dial wearing a capability dial's clothes.

The development side agrees. According to Lombardo and Eichinger's Center for Creative Leadership model (1996), roughly 70 percent of capability forms through challenging experiences and another 20 percent through other people, leaving formal coursework — the only input completion measures — at about 10 percent. A headline KPI built on the 10 percent slice cannot forecast movement driven by the other 90.

The replacement record already has a specification. An xAPI statement shaped "actor scored 0.86 on the [role] rubric, witnessed by assessor, timestamped" carries actor-verb-object-score provenance that joins cleanly to HRIS requisition data. The full anatomy:

Every statement falling inside the 90-day window before a requisition closes becomes candidate evidence for that fill — that join is what turns assessment logs into mobility data. Pin both KPIs down operationally now, so every later calculation matches:

PVIFR wins the headline slot; completion keeps a narrower job verifying that modules load, track, and play. The reason sits in the causal chain a mobility KPI must capture: learning event → assessed proficiency → eligibility flag in the marketplace/ATS → application → internal fill. Completion data falls out at the third link. Eligibility flagging needs evidence a matching engine trusts — a scored rubric plus attestation — and a boolean carries neither, so the high finisher is never flagged, never surfaces, and the requisition routes external. That is the mechanical answer to why high-completion academies still buy external talent.

The stakes compound as skills reprice: according to a July 30, 2026 Summit Events post, some skills are becoming increasingly valuable while others decline in relevance — a market that pays for verified evidence, not seat-time proxies. The Pedowitz Group's Completion-to-Effectiveness Scorecard reaches the same verdict, defining completion as the "Participation & exposure" layer (LMS completion, time-in-module, quiz pass) in a chain where completion drives proficiency and proficiency drives behavior — and noting that high completion still yields low impact when content isn't coached or embedded in workflows. Proficiency, in the working definition attributed to Michael Hyatt ("Skill + Contribution = Proficiency"), begins when other people are impacted by your work — precisely what a witnessed rubric score captures and a completion flag cannot. Your move before the Q1 2026 baseline opens: pull twelve months of completion extracts and closed requisitions, attempt the join on employee ID plus requisition close date, and count how many internal fills the completion file can verify. In most cases the answer is none — and that null result is your business case for the switch.

The Tenure Spread

Open with the spread LinkedIn Economic Graph data documents: employees at companies with high internal mobility stay 5.4 years on average, against 2.9 years where movement is rare — nearly double. Neither column mentions course consumption. People stay where the organization demonstrably moves them; movement itself, not module consumption, is the retention-relevant variable. Any headline KPI that cannot distinguish those two employers is measuring the wrong thing, however green its charts look.

The mechanism keeping completion rate alive is easier to explain than defend. Completion is manufactured by instructional-design choices — module length, sequencing, auto-advance settings — not by any promotion, transfer, or internal fill. An academy can post a flawless completion record and still log zero internal moves; the metric cannot tell a talent pipeline from a playlist. Kill the belief outright: finishing the curriculum is not readiness to move, and a high finish rate is not evidence the academy feeds anyone's pipeline.

Why do capable teams ship the wrong dashboard anyway? Instrumentation cost. According to ATD's "Evaluating Learning" research (2019), about 83% of organizations measure reaction (Level 1) and roughly 50% measure knowledge (Level 2), but only about 25% measure behavior (Level 3) and around 12% measure results (Level 4). Completion crowds the top tier because it is nearly free to collect; every mobility outcome — fills, ramp curves, retention of movers — lives in the barely-instrumented bottom. Your peer group is largely measuring what is easy, not what moves headcount.

Now price the alternative. According to CEB (now Gartner) ramp research, a typical external hire needs about 8 months to reach full productivity, so every avoidable external fill imports a multi-month proficiency deficit. Mind the edge case: the penalty applies even to excellent external hires, because proficiency arrives late regardless of candidate quality — an internally gated candidate cleared the bar before day one. That asymmetry is why proficiency-verified internal fill rate leads external hiring spend: each fill routed internally deletes the ramp drag from the plan.

Gallup sizes what the vanity view conceals: according to Gallup, losing an employee costs one-half to two times annual salary. There is no defensible conversion from modules finished to dollars protected — which is precisely why completion survives review cycles. It flatters without pricing anything.

Close on direction of causality. According to Josh Bersin Company research on internal talent marketplaces (2022), adopting organizations report faster internal fills and stronger retention among movers — directional confirmation that measuring and enabling movement precedes movement, not the reverse. And the evidence base contains no comparable dataset linking completion counts to fill rates. The verdict follows: proficiency-verified internal fill rate takes the headline slot because it alone instruments Level 3–4 territory tied to discrete fill events; completion drops to designer-owned content QA. Before your next quarterly review, build the ledger the old dashboard hid — recent external fills weighted by the ramp curve above, plus Gallup's cost band on every regretted exit.

Four Tests, One Winner

Put course-completion rate and PVIFR through the same gauntlet — predictive validity for internal fills, gaming resistance, data availability from your existing LMS-plus-HRIS stack, and auditability for external reporting, each weighted up to five points, twenty on the table — and three of the four tests aren't close. Data availability goes to completion rate, and it wins that one for the least flattering reason possible: it requires nothing, because it verifies nothing.

The gaming gap is mechanical, not moral. Completion lives in platform settings: shorten the modules, flip on auto-advance, unlock the navigation, and the finish rate climbs while not one learner gains capability. A proficiency gate runs on the opposite substrate — the record that counts is a person, an ID, a timestamp, a score against a pre-published rubric — and manufacturing that at scale means recruiting or corrupting witnesses, which gets expensive fast and leaves a trail. The gate also has to satisfy the reliability property EduSynch specifies: according to its June 30, 2026 guidance, results must be stable regardless of when or where they're administered, so two similarly proficient employees score consistently. Settings toggles answer to no such discipline.

Credit where due: data availability is completion rate's lone 5, and it's real — any LMS reports it this afternoon. PVIFR asks for exactly one integration, assessment scores joined to requisition-close records in the HRIS, and that build costs roughly one sprint of engineering time. Price it honestly instead of pretending it's free; one sprint is still the cheapest credibility purchase in this comparison, because it turns a vanity counter into a number finance can trace back to requisitions.

Auditability is where the standard itself settles the argument. ISO 30414-1:2018, the human capital reporting guideline, includes internal-mobility ratios in its disclosure set, and a proficiency-gated fill rate maps onto that structure cleanly: a defined denominator of closed requisitions, a defined numerator of candidates who cleared a published rubric with manager attestation, both reconstructable from system records. Completion rate appears nowhere in the standard — no disclosure it feeds, no reconciliation it supports, no external stakeholder who can check it against anything.

None of this makes completion worthless; it makes it narrow. Compliance tracking is the one job where "finished" is literally the legal requirement, and there it stays. The FMCSA's Training Provider Registry FAQ (cited by DriverTrack) shows the pattern: entry-level driver-theory training carries no minimum-hours rule, and completion requires an overall assessment score of at least 80% — even the regulator pairs "done" with a score floor. DriverTrack's library states the logic flatly: if your state sets a minimum, that's your floor, full stop, and the question worth asking is where to set your target above it. Above compliance tracking, completion holds no position in the mobility KPI hierarchy. None.

So the verdict, stated once and without hedging: PVIFR wins all four tests — projected roughly 17–18 of 20 points versus 9–10 for completion rate, with the spread coming from three tests completion loses outright and one it wins by default. Four tests, one winner, no tie.

What the Data Doesn't Tell You

Content for What the Data Doesn't Tell You is being prepared.

What a 13-Week Baseline Can't See

Goodhart's law does not spare the winning metric. The day PVIFR becomes the headline KPI, some manager optimizes the gate itself: instructors teach to the rubric, coaches drill the sample items, and pass rates climb while capability stays flat. According to EduSynch's June 30, 2026 guidance on assessment validity, the diagnostic question is whether an instrument measures actual practical abilities or memorization and test-taking strategy — a rubric optimized under KPI pressure drifts toward the second. The guide's countermeasure is structural: refresh every rubric annually and recalibrate assessors against live work samples, so last year's coached answers stop scoring. Be honest about the ceiling: no baseline fully defeats Goodhart. An annual refresh bounds the decay; it does not eliminate it.

The small-n problem is arithmetic, not opinion. A single quarter often yields fewer than 30 internal openings, and below that threshold each hiring decision moves the rate by whole points — at 29 openings, one fill is 3.4 points. A confidence interval that wide cannot distinguish real improvement from noise, so the Q1 2026 baseline is a starting position, not a stable estimate. Seasonality compounds it: post-appraisal transfers cluster in January and February, inflating first-quarter fills relative to any mid-year run-rate.

Cut scores hide judgment. A uniform 80% threshold treats a welder's practical assessment and a financial analyst's case exercise as equivalent signals, and they are not — difficulty, reliability, and rater variance all differ by role family. Per-family calibration is therefore mandatory, and it injects human judgment the headline number conceals. Regulators already accept this openly: according to the FMCSA Training Provider Registry FAQ (via DriverTrack), the federal behind-the-wheel completion trigger is a professional judgment by the instructor, made element by element, because 49 CFR Part 380 sets completion standards, not clocks. The schools that survived FMCSA's audit sweeps were the ones that could show why their judgments were defensible — publish your calibration rationale the same way.

Now the strongest counter-evidence, stated squarely: according to Guild Education's published analyses of frontline programs — the Walmart Live Better U population — credential completion is associated with materially higher odds of advancement. In hourly, credentialed pathways, completion can track mobility. The guide's claim is therefore scoped to professional and salaried academies, not universal. What this does not resurrect is the dead myth that finishing a curriculum means readiness to move: Guild's association runs through credentialed frontline pathways, while salaried-academy completion still responds to module length and auto-advance settings. Anyone quoting Live Better U to defend a high finish rate in a salaried academy has imported a finding from a different labor market.

Last, the blind spot no gate fixes: PVIFR snapshots state, not trajectory. Proficiency is present-tense — as Axel Neree has written, you can't be proficient at something you're not doing — so the gate records today's capability and misses learning velocity. An employee at 74% and climbing may outperform a stagnant 83% passer within the quarter, and PVIFR as defined cannot see her. According to Bob Sornson's work on competency-based progression, closing that gap takes ongoing observational formative assessment — adjusting instruction until learners reach complete proficiency — a practice the century-and-a-half-old coverage model abandoned. Pair the gate with velocity notes, or accept that your best risers are invisible.

Action for the Q1 2026 close: publish the denominator beside the rate. Report PVIFR as X/Y with Y printed, split January–February fills from March, attach the role-family calibration memo and assessor roster, and label the quarter "pilot estimate." None of these caveats reverses the KPI switch — they define how honestly the first baseline gets read.

A Worked Quarter

Everything in this section computes on one page. The case is a composite — a 4,000-employee industrial employer academy running the gate this guide defined earlier — and the parameters below are frozen so any reader can rebuild the quarter from the formulas already given. That is the standard a baseline has to meet: if a figure cannot be traced to a locked input and simple arithmetic, it is decoration, not evidence.

Read side-by-side, the same quarter tells two completely different stories. Completion landed at 81%; PVIFR landed at 19%. Under the old belief that finishing the curriculum means readiness to move, that 81% would have been celebrated as a healthy pipeline feeding talent upward. The second reading is the correction: twenty-three verified moves, and ninety-five openings the business still had to backfill from outside.

None of the readings hold up if the numerator has to be rebuilt by hand. Because the gate's scored statements wrote to HRIS weekly and each attestation carried a timestamp, the April 3 readout was a saved query rather than a reconciliation project. That plumbing detail decides whether the baseline survives contact with Q3: a number anyone can regenerate gets defended; a number someone has to reassemble gets negotiated down.

The 2026 switch fails in governance, not arithmetic. An academy can compute proficiency-verified internal fill rate flawlessly and still lose the metric — the gate drafted after the first scores existed, the rubric changed quietly mid-year, a broken HRIS join hidden inside a monthly batch. According to the Pedowitz Group's August 24, 2026 guidance, the recurring failure mode in measurement programs is launching without success metrics defined first. Five rules keep the switch auditable.

Rule 1 — Publish the gate before the window opens. The cut-score, rubric items, assessor roster, and the 90-day recency definition carry formal sign-off by December 2025, before Q1 scoring begins. A gate retrofitted after seeing Q1 results invalidates the baseline outright — you fitted the yardstick to the measurements — and invites gaming from day one, since everyone who saw the draft scores knows which rubric lines ran soft. If the sign-off already slipped, log the real go-live date as the baseline start; never backdate.

Rule 2 — Version-control the instrument. Any mid-year change to the cut-score, rubric items, or assessor calibration resets the baseline clock: a February recalibration makes January and March different tests wearing one name. Log every change with date and approver so the 2026 trend line compares like with like. Calibration drift is the sneaky variant — two assessors scoring the same demonstration differently leaves no mark in the score column, so drift masquerades as a shift in candidate quality. According to EduSynch's June 30, 2026 analysis, accurate baselines enable consistent progress tracking while inaccurate ones breed frustration and slower progress; an unstable instrument does the same damage to the trend line.

Five Rules for the 2026 KPI Switch

Rule 3 — Sync weekly, not monthly. Run the scored-assessment-to-HRIS feed every Friday. Monthly batches hide the classic breakage — a reassigned employee ID, a renamed requisition code — until the quarter is unrecoverable. Weekly runs surface a broken join mid-quarter, while the affected weeks can still be repaired and re-pulled. This is the operational cousin of the Pedowitz Group's named failure mode of tracking only LMS reports: the LMS knows who passed the gate; only the HRIS join knows whose pass filled an actual opening.

Rule 4 — Run a quarterly adverse-impact parity check. Compare gate pass rates across demographic groups and investigate any divergence wider than 10 percentage points before the KPI reaches executive reporting. Treat the threshold as a tripwire, not a verdict: gaps vary with rubric wording and assessor mix, so trace whether the divergence sits in the rubric, the roster, or the applicant pool. Pair the check with cohort controls — holdouts by region or team, per the Pedowitz Group's validation guidance — to separate gate effects from population effects. A mobility KPI that screens unevenly is a legal and ethical liability, not a win.

Rule 5 — Report one number upward: PVIFR with its n and confidence interval. Completion, hours, and enrollment counts move to a designer-owned appendix dashboard. The mechanism is blunt: shown two headline numbers, leadership picks completion, because it is older, larger, and reliably up-and-to-the-right — it responds to module length and auto-advance settings, not to a single promotion, transfer, or internal fill. A quarter can post record completion alongside no verified fills at all. Per the Pedowitz Group, completion signals adoption and exposure, not skill; its five signals — enrollment, start/completion, time-in-content, quiz pass rate, recertification cadence — belong in the designer's content-QA view, exactly where the decision rule demotes them.

This week's move: assemble the four artifacts — cut-score memo, rubric, assessor roster, recency definition — into one version-controlled document with named approvers, and schedule the Friday sync before the next ad-hoc report request lands. If December 2025 slipped, apply Rule 2 to the calendar: stamp the true go-live date, restart the clock, and let the first clean quarter argue for itself.

Rule 4 — Run a quarterly adverse-impact parity check. Compare gate pass rates across demographic groups and investigate any divergence wider than 10 percentage points before the KPI reaches executive reporting. Treat the threshold as a tripwire, not a verdict: gaps vary with rubric wording and assessor mix, so trace whether the divergence sits in the rubric, the roster, or the applicant pool. Pair the check with cohort controls — holdouts by region or team, per the Pedowitz Group's validation guidance — to separate gate effects from population effects. A mobility KPI that screens unevenly is a legal and ethical liability, not a win.

Rule 5 — Report one number upward: PVIFR with its n and confidence interval. Completion, hours, and enrollment counts move to a designer-owned appendix dashboard. The mechanism is blunt: shown two headline numbers, leadership picks completion, because it is older, larger, and reliably up-and-to-the-right — it responds to module length and auto-advance settings, not to a single promotion, transfer, or internal fill. A quarter can post record completion alongside no verified fills at all. Per the Pedowitz Group, completion signals adoption and exposure, not skill; its five signals — enrollment, start/completion, time-in-content, quiz pass rate, recertification cadence — belong in the designer's content-QA view, exactly where the decision rule demotes them.

This week's move: assemble the four artifacts — cut-score memo, rubric, assessor roster, recency definition — into one version-controlled document with named approvers, and schedule the Friday sync before the next ad-hoc report request lands. If December 2025 slipped, apply Rule 2 to the calendar: stamp the true go-live date, restart the clock, and let the first clean quarter argue for itself.

What to do next

Frequently Asked Questions

What minimum score has to appear on an xAPI statement for an internal fill to count toward PVIFR?

A passing record carries result.score.scaled of 0.86 against a pre-published, version-stamped role rubric whose cut-score is set at 80% or higher, with manager attestation attached in the context field.

How old can a proficiency assessment be and still make someone eligible for an internal opening?

Only statements timestamped inside the prior 90 days before a requisition closes become candidate evidence for that fill.

What is the exact PVIFR calculation?

PVIFR equals openings filled by candidates who cleared a pre-published gate — a rubric cut-score of at least 80% plus manager attestation — within the prior 90 days, divided by total internal openings.

Does MOOC research support the idea that finishing a curriculum means you're ready to move?

Christensen et al.'s 2013 analysis of MIT/Harvard edX data found average completion across 69 MOOCs was just 3.13 percent, and institutions moved that number dramatically by changing course length and navigation friction while underlying competence stayed untouched.

If formal coursework is such a small slice of development, where does capability actually come from?

Lombardo and Eichinger's 1996 Center for Creative Leadership model attributes roughly 70 percent of capability to challenging experiences and another 20 percent to other people, leaving formal coursework — the only input completion measures — at about 10 percent.

How large is the retention gap between employers with high versus low internal mobility?

LinkedIn Economic Graph data shows employees at companies with high internal mobility stay 5.4 years on average, against 2.9 years where movement is rare — nearly double.

Quick answers