How local rank tracking works, documented as an evidence-generating process.

Why the pipeline is an evidence process

Every scan is a record of what Google returned. That makes the pipeline an evidence-generating process, and evidence processes have rules: the design must be fixed before collection, the collection conditions must be consistent, and the archive must be preserved unchanged. Local rank tracking obeys or breaks those rules at each of its five steps.

Define: fixing the estimand

Binding the listing and keyword set fixes what is being measured. Changing keywords mid-campaign changes the estimand — records before and after describe different quantities and cannot be compared. Evidentiary rule one: the definition is part of the design.

Grid: fixing the sample

The geo-grid fixes where measurements are taken. Resizing the grid between scans changes the sample; the archive stops being a time series of one surface and becomes snapshots of several. Evidentiary rule two: geometry is frozen for the life of the comparison.

Sample and archive: collection and preservation

Per-point queries collect what Google returned; the archive preserves it with timestamps and geometry. Consistent scan windows reduce time-of-day variance. The archive is the evidentiary core — reports are just renderings of it.

Re-scan: the cadence that creates series

A fixed cadence — weekly or monthly — converts records into a comparable series, the only form in which movement can be evidenced. The consumers of that series are documented in the agency and multi-location entries; the tools that automate it are mapped on the comparison page, with LocalRank's implementation — from $29 per month — at localrank.so.

Return to the site home.

FAQ