METHODOLOGY
Measured, not assumed
Flow bridges spatial geomarketing to read the territory, like-for-like retail analytics to track it, data science to qualify movement, media execution to act, and Geolift to measure the response.
Territory × Temporality = an optimised catchment area
Stable, privacy-clean territorial foundations
Building a reliable read of the ground without compromising privacy.
Anonymous Mobility Data
We process aggregated mobility streams. Zero personal data, zero individual tracking, and absolutely no advertising identifiers are ever used.
Neighbourhood Origins
Visitor origins are read at the neighbourhood level (census sections). It is fine enough to map a basin accurately, yet stable enough to compare over time. Postal codes serve as an export format, not an analysis grid.
Continuous Footfall
Unique visitors are measured monthly; footfall is tracked weekly for Geolift. Local market share and territorial penetration are assessed on an annual baseline.
A measured dynamic catchment area
We dismiss theoretical gravity models (like Huff). Primary and secondary trade areas are defined by their actual cumulative contribution to footfall (Pareto), cross-referenced with local market share. Segments are dynamic and recalculated every season.
Core
The neighbourhoods that carry the most weight in footfall: the structural foundation of the point of sale.
Consolidation
The rest of the primary trade area, representing an already captured base.
Traction
Secondary neighbourhoods showing structural acceleration.
Potential
Secondary neighbourhoods with high footfall volume but a local market share that remains low.
Churn
Primary neighbourhoods in durable decline compared to the previous year.
Isolating true movement from seasonality
Applying strict retail analytics principles to geographic data.
YoY Comparison
We compare the exact same calendar quarter from the previous year. A neighbourhood is always judged against its own site's baseline to neutralise global market effects.
Three Horizons
We analyse the annual structural baseline, seasonal shifts, and recent event-driven reactivity. Each type of highlight has its own reference window.
Materiality Threshold
We dismiss 4-5-4 calendars in favour of clean quarters. A quarter is flagged only when the decline is both material in absolute terms and abnormal relative to the site.
Turning raw variations into prioritised action
Mathematical filters to isolate what truly matters.
Z-Score & Double Criteria
A movement is judged abnormal based on the dispersion of neighbourhoods for the same site. An alert requires both absolute materiality and statistical significance (Z-score).
Scoring & Early Warning
Neighbourhoods are ranked and normalised by segment, turning rankings into exportable priority weights. Early warning lists identify shifts before structural decline sets in.
Spatial Autocorrelation
Next on our engine's roadmap: leveraging local Moran's I to distinguish isolated declines from concentrated hubs, alongside temporal Z-scores.
Territory × Temporality = Performance
The Strategist recommends, the planner decides. Flow plans, geoplans and measures. We do not broadcast campaigns.
Tactics & Distance
The response follows spatial friction: proximity requires short windows; distance requires longer conviction. We protect the core before conquering.
Temporal Formats
Different highlights dictate different windows: Urgency (very short), Promotional Peak (one to two weeks), or Destination (several weeks of consideration).
Bidding & Hold-out
Priority scores translate into bid modifiers. Coverage is set by the planner's budget. Control territories (hold-out) are purposefully never targeted.
Agnostic Export
Ready-to-use Geotargeting (radiuses, postal codes) and Geofencing (polygons) readable by any media buying platform.
Did the targeted territories react?
Scientific measurement of incrementality.
Difference in Differences
We compare the evolution of targeted territories against comparable control territories (hold-out) that were deliberately excluded from the geoplan.
Strict Comparability
A constant panel is required for measurement. Control groups are selected by rank comparability, not mere geographic isolation.
Territorial Measurement
We measure raw territorial response with a carry-over window, never inflated numbers. This is geographic incrementality, not channel attribution.
What we deliberately exclude
Our engine is defined as much by what it rejects as by what it measures.
No Simulated Zones
No isochrones, no theoretical distances, and no gravity models calibrated by guesswork.
No Individual Tracking
We evaluate aggregated neighbourhoods. We never track people or use advertising identifiers.
No Flawed Comparisons
No month-over-month comparisons that confuse seasonality with decline. No metrics displayed on incomplete data. No media attribution.
Strong season. Thanks to the center?
Your results look good, but where did the performance come from? Compare yourself with the center’s footfall. Flow isolates the shopping center’s performance so you know whether your growth comes from in-store execution or from the traffic the center brought in. And if the season fell short, the same question applies the other way around.
COMPARE MY STORE WITH THE CENTERFLOW METHODOLOGY · ILANEXPERIENCE.COM