Prospecting
Prospecting is how you ask a question that watches the answer. You compose the question by filling four dimensions, save it with alerts, and allocate the hereditaments that matter to a portfolio. Where Search finds what's there now, Prospecting tells you when the answer changes.
What prospecting is
The wizard surfaces four dimensions: Where, What stayed the same, What changed, and Over what period. Fill them in any order; the Explorer composes the right kind of search from what you enter. Press ⌘↵ (Mac) or Ctrl+Enter (Windows / Linux) to run the search when the question is well-formed.
First-touch walkthrough
First-time researchers see an optional walk-through that opens with a worked example — Camden shops whose rateable values stand out from the peer median on the 2023 rating list. Dismissable; bring it back from Settings → Reset help.
Where
The geographic scope. Empty Where means national. Pick at the level that matches your question — Region, Combined authority, or Billing authority.
What stayed the same
The fields shared by every hereditament in the set. Pin a category, a scheme, a ratepayer name, a floor area. This dimension also defines the peer set when What changed is a deviation: the median or mean is taken across the group you have pinned here.
What changed
The field whose change, deviation, absence, or threshold-crossing is the question. Each row carries the field, the variation pattern, the value, and (for a numeric pattern) the direction, all on one line. + Add field below the dimension creates another row.
Any relief asks about relief of any type: whether a hereditament (or every holding of an account) holds none now, or gained / lost relief between two dates — held none, then some, or the reverse. Relief type asks about specific reliefs: pick the types to watch, then whether a hereditament (or every holding of an account) lacks them now, lacks everything except them, or gained / lost one between two dates. Neither applies a rateable-value bracket; the small-business question is composed as Relief type is absent with rateable value held at most £15,000, and its frame grades the eligibility tier. One honesty note on gained and lost: charges carry no end date, so the transition reads what each authority published at the two moments — an authority that stops publishing charge rows will read as widespread loss. Treat a cluster of losses from one authority as a publication change first.
Changed more than its peers did answers the question surveyors ask most: "who got a larger increase than their peer group would predict?" Pick the field (rateable value, say), the two moments in Over what period, and how much beyond the group's own change a hereditament must move — in percentage points (their change plus 25 points) or as a multiple (1.25× their change; this reading needs a rising group). The peer group is exactly what you define: the authorities in Where and the constraints in What stayed the same. Hold "category is Office" and RV "stayed at or above £500,000", and the group's change is computed over precisely those offices — never a wider population trimmed afterwards. The multiple reading against a falling group returns nothing by design; switch to percentage points to ask about falls. Behind advanced… sit the statistic (the group's average change, or a chosen percentile of it) and the minimum group size a comparison needs.
Over what period
The window the question reads across. Five shapes: As of a date and Around a date read one moment; Between two dates, Between two lists, and One list, two dates read two. The picker offers only the shapes every row can run under — a two-moment pattern, a hold that reads one side of the window, or a hold that is unchanged across it all need a paired shape — and says which row narrowed the choice. A window across two lists reads whole list states, each at its own date; the line under the picker says which lists it resolved to. When What changed compares each hereditament against its peers, the peers are read at this window too.
Picking a geography
Type to find a region, a combined authority, or a billing authority; picking a region or combined authority picks every billing authority beneath it. Each pick shows what its authorities publish and how recently. Save a set of picks as a named list to reuse it; a saved list applies with one pick and can be deleted from the same control.
Data coverage
Each billing authority shows a coverage hint inline — "Camden — ratepayer data through Sep 2025; account names available; quarterly updates", or "City of London — sparse; last ratepayer data Aug 2019". A thin-coverage authority returns fewer ratepayer matches.
Variation patterns
The pattern picker is grouped by how many moments the question reads. How it compares right now (one moment): sits far from its peers' typical value, its peers' average, or the extreme tail of its peers; for a relief, is absent, is absent except, or is absent account-wide. How it changed over the period (two moments): changed more than its peers did, rose or fell by an amount or by a percentage, crossed a boundary, changed, changed from one value to another, and for a relief was applied / became absent (a named type) or was gained / was lost (any relief). Pick the pattern first — its controls follow — and the period picker offers only the windows the pattern can run under.
Peer-group size
Behind advanced… on a peer-comparison pattern sits peer group ≥ N, resting at 5. A comparison against a handful of peers says little, so a group thinner than N is annotated in the results, never hidden. The peer set is exactly what Where and What stayed the same define.
Schemes are within a rating list
A scheme reference is per-revaluation — the 2023 list has its own schemes; the 2026 list has its own. When Over what period pairs two lists, the scheme in What stayed the same anchors the cohort on whichever list carries it, and the run says so if neither does. Across lists the only stable thread is the hereditament itself (its UARN); to keep like with like, hold Total floor area · is unchanged as well.
Charges have no expiry date
Charge records carry no end date, so a relief is read by its absence at a moment, never by its expiry: is absent, is absent except, and is absent account-wide read now; was applied / became absent (a named type) and was gained / was lost (any relief) read what each authority published at the two moments.
The three "absent" readings differ in what they let through. Is absent keeps ratepayers holding none of the types you name — they may hold others. Is absent except keeps those holding one of the types you name and nothing else, which is the question which of my retail-relief recipients claim nothing further? For ratepayers holding no relief at all, ask Any relief · is absent. An authority that stops publishing charge rows reads as widespread loss; treat a cluster of losses from one authority as a publication change first.
Cross-list invariance
Comparing across two rating lists is honest only while the hereditament is still the same thing. Hold Total floor area · throughout · is unchanged in What stayed the same, with a tolerance in square metres if you want one, to keep like with like; the same operator holds any valuation-side field, or the occupier, steady across the window. Without it, every hereditament valued on both lists counts — extensions, splits, and mergers included — and a change in value may be a change in the building. The submit card's Query defaults line says which reading applies.
First-time loads
A first run over an area nobody has asked about recently reads from cold storage and takes longer; asked again, the same question is much quicker. National questions run in the background whatever their state.
Adjusting from the question row
The results table's Question row shows four colour-coded chips — one per dimension. Click a chip to edit one pick inline; the query re-runs and the table reloads in place. For a larger rebuild, click Edit composition to return to the wizard with the current state seeded.
Saving with alerts
Save (⌘S / Ctrl+S) names the question and configures alerts in one dialog. Re-opening a saved question runs it against current data; the dialog reopens seeded with the saved values, the primary action becomes Update bookmark, and Save as new keeps your edits as a second bookmark under its own name, leaving the original untouched. A question saved with alerts re-runs on its cadence and reaches your inbox when the count of matches changes.
Query defaults
The submit card's last line names the readings the question applies that you did not set — which charges count as reliefs (published or rule-confirmed; surcharges never), how the occupier at a date is chosen (by the account's own start date), which occupation states are included, what a peer group is, whether a one-moment question in a paired window reads the later moment, and the result cap. A default drops from the line when a row of yours overrides it, so what you see before you run is what runs.
Trigger types
Three triggers. Added — a hereditament newly matches the question. Updated — a hereditament in the set has any tracked value changed since the last alert (generic, not per-field). Removed — a hereditament no longer matches. Three cadences — Weekly digest, Monthly digest, Quarterly digest. The cadences are aligned to the VOA update rhythm: the VOA publishes change updates twice weekly (Wednesdays) and full epoch refreshes every two months — sub-weekly digests would either repeat or be empty most days. Two channels — In-app and Email. Digests arrive shortly after VOA updates land, on your chosen cadence.
Allocate to a portfolio
Tick the rows you want and click Allocate selected in the results-table toolbar. The side panel opens with a source summary and a target picker — an existing portfolio or a new one. The action label carries the named target — for example "Allocate 2 to Camden retail watch".
Prospecting versus portfolio
A prospecting search is a question whose answer changes — the Explorer keeps running it and alerts you on the changes. A portfolio is a fixed list of hereditaments you chose, with its own alerts on the chosen members. Allocating from a prospecting search doesn't stop the question running; the prospecting search and the portfolio are independent.