There is a long-replicated finding in the estimation literature: simple statistical models tend to beat expert judgement at estimating uncertain outcomes. Not because the models are clever — most are crude — but because they are consistent. The expert varies against themselves; the model does not. A model built from an expert’s own past judgements often outperforms that same expert.
Before you buy more people, technology, or services — do you know how effectively your existing resources are being converted into outcomes?
PFV shows where an investment would actually improve capacity, flow, throughput or exposure — and where it would not.
What that means here. You supply the facts, because only you can: which steps exist, how long each waits, whether the approval is real. The model does the combining, because that is the part where judgement quietly underperforms. And unlike an opinion, it commits to a number in advance and can be shown to be wrong — which is the whole point of building it.
Describe the wait, get a commitment
The step
Name it so the ledger below means something to you later.
How long is the wait, and what is it waiting on?
Business days, median. These two are required; everything below sharpens the prediction and narrows the band.
What share of all waiting in the process is this?
Percent. Dominant queues reward fixing further than their size suggests.
Arrivals and completions, per period
The single most valuable pair of numbers here. Any period works — day, week, month — because the model uses the ratio; just use the same one for both. If arrivals exceed completions the queue is capacity-bound and the model cuts its own prediction back accordingly, because flow work does not fix a growing queue.
Does it wait for a cadence?
A weekly review, a Friday, a monthly batch.
Defect rate at or downstream of this step
Percent of items reopened, rejected or redone. A wait that causes rework pays twice when removed.
The ledger
Every prediction, dated. Enter the actual once the change has settled, and the scoring below updates. Three numbers decide whether this model deserves to be used: whether roughly 90% of actuals land inside the 90% band, whether its average miss is small, and whether it beats the dumbest possible rule — halve the wait. If it cannot beat that, it has not earned its complexity and should be thrown away.
This browser is blocking local storage, so predictions will not survive a reload.
What this model is, honestly
Its weights are borrowed, not earned. They start from what each situation returned across 42 modelled sample processes. That is a starting policy, not evidence about your organisation, and it should be replaced by weights fitted to real outcomes as soon as this ledger has enough entries to fit them. Until then the model’s honest claim is consistency, not accuracy.
It predicts one thing only: business days of lead time removed by fixing that wait. It says nothing about whether the fix is politically possible, whether the team will sustain it, whether the process should exist at all, or what it will cost. Those are not estimation problems and no model of this kind touches them — they stay with the expert, which is exactly where the literature says they belong.
It can be beaten by a coin flip, and reports when it is. The scoring compares every prediction against a rule so simple it is almost a joke. Publishing that comparison is the only reason to trust the rest of it.
Use at your own risk
Process Flow Visibility is a method and a set of estimating tools, not professional advice.
Every figure it produces is derived from numbers you enter, most of which are estimates. It does not
measure your process, audit your data, or know your business.
Nothing here is legal, financial, security, safety, medical or engineering advice. Do not use
these outputs as the sole basis for any decision that carries operational, financial, regulatory
or safety consequences. Validate anything that matters against your own records and your own
professional judgement.
Provided “as is”, without warranty of any kind, express or implied, including
merchantability, fitness for a particular purpose and non-infringement. To the fullest extent
permitted by law, Don Woodward accepts no liability for any loss or damage — direct, indirect,
incidental, consequential or otherwise — arising from the use of this site, these tools, or any
figure they produce. You use them entirely at your own risk, and you agree to hold the author
harmless for any outcome that follows.
The worked examples are modelled from industry data, not audited case studies.
Named methods and standards are the property of their respective owners and are referenced for
description only; no endorsement or affiliation is implied.
