Weighting anchor vs. inline lots in a shopping center occupancy index

When you're building a parking-based occupancy index for a shopping center, the first modeling decision that matters is how you weight the anchor lot against the inline lot. Get it wrong and your index just tracks the grocery anchor's baseline traffic, with some noise from the rest of the property riding along underneath it.

A typical grocery-anchored center puts 60-70% of its total stall count in front of or adjacent to the anchor box. Anchors negotiate generous parking ratios into their leases relative to their square footage, while inline shops, the nail salon, the phone repair place, the fast-casual spot, share a thinner band of stalls near the street-facing storefronts. Average fill rate across the whole lot with no weighting and you've built an anchor traffic index wearing a shopping-center costume. The inline signal gets drowned out by sheer stall count.

Why equal weighting by stall count misrepresents the center

Raw stall-count weighting feels neutral, but it isn't. Anchors like grocery or a discount box tend to run flatter, more predictable daily curves: steady weekday fill, a Saturday bump, done. Inline tenants, especially restaurants and services, run peakier patterns tied to lunch, after-work, and weekend windows. Weight your index by stall count alone and those inline peaks get averaged down to almost nothing against the anchor's steady baseline, even in a center where the inline mix is where your occupancy-cost-ratio exposure sits.

The fix most analysts land on is weighting by economic exposure rather than physical space. If you're trying to read through to NOI risk or sanity-check same-store sales guidance, weight each lot segment by its share of base rent or by GLA-adjusted sales productivity, not by how many stalls happen to sit in front of it. A center where inline tenants carry 40% of base rent but only 25% of the parking footprint should have that inline lot segment weighted up, not left at its raw stall share.

Building a two-series index instead of one blended number

The cleaner approach, and the one that holds up better when you're trying to call a quarter before the tenant reports it, is to stop forcing anchor and inline into a single blended score and instead track them as two parallel series against the same site. Zone the lot by proximity to anchor entrances versus inline storefronts, pull daily vehicle counts for each zone separately, and let the two series sit side by side on the same chart. You lose nothing by keeping them separate, and you gain the ability to see when inline fill is softening even while the anchor keeps the aggregate number looking fine. Anchor box traffic rarely moves enough quarter to quarter to tell you much about the smaller shop mix driving occupancy-cost-ratio stress, so that divergence is usually the more useful signal.

If you need a single blended figure for a screen or a quick comparable across a portfolio, weight each zone's daily fill rate by its trailing GLA share of leased space rather than by stall count, and disclose which weighting you used. Analysts comparing your index across properties will ask, and "stall count" versus "GLA share" versus "base rent share" produce meaningfully different answers on the same raw vehicle data.

One more wrinkle worth flagging: anchor parking fields are often shared with outparcels, bank branches, or a standalone pharmacy that technically sits on a different parcel but draws from the same lot. Skip that check and you'll misattribute traffic to the anchor that belongs to a pad tenant with its own lease and its own comp-sales trajectory instead. Worth checking the site plan before you draw your zones, not after you've already built three quarters of history on the wrong boundary.

A per-site occupancy series built from repeat high-resolution satellite and aerial passes delivers daily vehicle counts at a resolution fine enough to separate anchor-adjacent stalls from inline-adjacent ones. Once zone-level counts are flowing daily, the weighting question above becomes a modeling choice you control instead of a limitation baked into the data.

If you're building this kind of index across a watchlist of properties, it's worth seeing what a daily, zone-level parking read looks like before your next earnings call.