Find out what it could not see

Whose job: anyone with an account. You should see the unsurveyed length, stated rather than averaged away.

What you are doing

You are finding out how much of the network the numbers actually rest on. This is the step councils test us on, and it is the one to understand before you read any output at all.

It is nobody's job in particular, and everybody's to understand. The officer who presents a programme to a committee needs this figure more than anyone, because it is the first thing a sharp councillor asks.

What PaveKeep does with it

Nothing. The unsurveyed length is derived every time it is shown.

A road nobody has surveyed is reported as UNKNOWN. Not as zero, and not quietly averaged into the figure with everything else. That means a network with a quarter of it surveyed reports a condition for that quarter and says so, rather than reporting a confident average of a quarter of the evidence.

What it will not do

• The method — It cannot tell you whether the unsurveyed part is good or bad. Nobody can — that is what unsurveyed means. Assuming it matches the surveyed part is the specific mistake this disclosure exists to prevent, and it is a mistake that flatters a network, because the roads that get surveyed first are usually the ones somebody was already worried about.
• What you have not got yet — The figure only falls by surveying more. There is no setting that improves it, and a programme built on 8% of the network is a real programme for 8% of the network — useful, and not the same document as one built on 80%.
• The product — Coverage is measured by length, not by road count. Surveying twenty short streets moves it less than one long rural road, which is correct and can be surprising the first time.

How you know it worked

Before: a network with some of it surveyed.

Do: open the dashboard.

After: the surveyed share of the network, stated as a length and a percentage, beside every figure derived from it.

If it is lower than you expected, that is the screen working. The usual cause is a survey that covered fewer roads than the person filing it believed — the Surveys tab says exactly what each run touched.

Hand to: whoever will quote a figure from this product in public. Send them this number with it, every time.

If it went in wrong: there is nothing to undo at this step. What to do instead is under what it will not do, above.

← Look at what you have · Set the rules, or accept the defaults → · All 13 steps