Area Identification

Our area metrics support screening a whole market: score every area on explicit criteria and rank them into a shortlist. A scored screen covers the market on equal terms, including the areas that would not come up from familiarity alone. The same method (indicators, normalization, a weighted composite) serves expansion targeting, media planning, and development prioritization.

1. Define the candidate universe

Fix the area system and the market bounds before touching indicators: which areas compete for a place on the shortlist? A metro's neighborhoods, a region's grid cells, every area in a state. Two rules:

• Score within the universe. Percentiles and comparisons only mean something against the set a candidate actually competes with.
• Match the grain to the decision. Neighborhood grain for siting and local marketing; coarser grains for market entry. Finer grain means more candidates and more small-area noise (see step 5).

2. Choose scoring indicators

Indicators fall into seven families; a workable shortlist usually draws on one or two from each:

One trap in the audience family: shares rank areas by composition, which favors small areas with a lean mix over big areas with a big audience. If you care about absolute audience size, multiply the share by visitor volume before scoring. "Many of the right people" and "mostly the right people" are different indicators; pick one deliberately.

3. Normalize

Raw indicators live on incompatible scales. Convert each to a percentile rank within the candidate universe before combining. Percentiles are robust to the heavy skew of visitation data, where a mean-based z-score is dragged by a few giant areas.

Some criteria work better as gates than as scores: "at least N monthly visitors" or "destination share above X" filters the universe before scoring, instead of letting a strong score elsewhere compensate for a disqualifying weakness.

4. Composite score and weights

The composite is a weighted sum of normalized indicators:

score(area) = Σ  wᵢ · pctl(xᵢ)        with  Σ wᵢ = 1

The weights are the strategy, stated as numbers: a late-night concept weights timing, an expansion screen weights momentum and audience, a development agency weights mix and trend. What matters is that the weighting is explicit and defensible. Two disciplines:

• Sensitivity-check the shortlist. Move the weights (±10 points between the top factors) and re-rank. Areas that stay in the top set under any reasonable weighting are robust picks; areas that appear under one exact weighting are artifacts of it.
• Same window for every indicator. Compute all indicators over the same period (or an explicitly chosen one, like a momentum lookback) so the score compares areas, not periods.

5. Sanity-check the shortlist

Put the shortlist on a map: spatial clusters and lone outliers both carry information the table hides. Check the smallest areas for small-sample noise; low-activity areas swing more and get more privacy suppression, and a suppressed value is missing, not zero. Read the top areas' raw indicator values, not just the composite. A top score built from four mediocre percentiles is a different finding than one built from two exceptional ones.

From the shortlist, the next step is a store profile of what already operates in an area, or a void analysis of what it lacks.