# Come back next year

*Whose job: the surveyor. You should see the new survey becomes current; last year stays on the record.*

## What you are doing

You are doing it again, a year later. This is the step that decides whether PaveKeep is a system the council uses or a project it did once.

It is two actions, not the whole sequence again: file this year's condition, and run the programme. The network is already there, the configuration is already yours, and last year's survey stays exactly where it is.

## What PaveKeep does with it

A new survey run, dated, which becomes the current condition — and last year's run, untouched, which becomes the comparison.

That is the whole point of filing condition as dated RUNS rather than overwriting a number: the second year is when the product starts answering a question the first year could not, which is whether the network is getting better or worse.

## What it will not do

- **The method** — Two surveys make a difference, not a trend. Direction over two years is worth knowing and worth stating cautiously; a rate of deterioration needs more than two points, and the product will not pretend otherwise.
- **What you have not got yet** — A comparison is only as good as what both surveys covered. Surveying a different part of the network this year gives you a different sample, not a worse network — and reading it as deterioration is the commonest mistake in year two.
- **The product** — A road whose segments were re-cut between surveys loses its per-segment trend, because the thing being compared has changed shape. A re-import that leaves geometry alone does not re-cut, so this only applies when the network itself was corrected.
- **Your role** — Filing condition is the condition capability, as it was last year. Nothing about the refresh needs more permission than the first pass did.

## How you know it worked

**Before:** last year's network and survey, and this year's condition data.

**Do:** file the new survey, then run the programme.

**After:** the new run listed beside last year's, the map showing current condition, and a programme built on it. Two calls, not the whole sequence — and that count is held by a test rather than a promise.

**If it was wrong**, void the new run. Condition falls back to last year's, which is exactly where it was before you started.

**Hand to:** the same people as last year. The useful thing to send them this time is the comparison, not the snapshot.

---

**If it went in wrong:** this step can be undone, and the undoing stays on the record rather than erasing what happened.

[← Record work that actually happened](/how-it-works/ledger-record) · [Undo a survey you should not have filed →](/how-it-works/survey-void) · [All 13 steps](/how-it-works)

---

Source: https://pavekeep.com.au/how-it-works/annual-refresh · PaveKeep · derived content