---
id: "concept-inverted-u-shape"
type: "concept"
source_timestamps: ["§ Distance has a U-shaped impact", "¶10", "¶12"]
tags: ["spatial-analytics", "targeting-optimization"]
related: ["concept-billboard-effect", "action-test-distance-bands", "contrarian-distance-decay", "quote-the-donut", "claim-stable-assortment-u-shape"]
definition: "A spatial targeting pattern where ad effectiveness peaks at moderate distances rather than very close or very far distances, producing a donut-shaped optimal zone."
sources: ["tail1"]
sourceVaultSlug: "hbr-seg-tail1"
originDay: 1
articleStem: "hbr-tail-115-location-based-advertising"
sourceUrl: "https://hbr.org/2026/03/a-better-strategy-for-location-based-advertising"
sourceTitle: "A Better Strategy for Location-Based Advertising"
---
# Inverted U-Shape Distance Impact (The Donut)

Conventional marketing assumes ad effectiveness **decays linearly with distance** — the assumption directly challenged by [[contrarian-distance-decay]]. The authors' research reveals an **inverted U-shape** for many retail categories.

## The three distance bands
- **Closest quartile (≈ within 4 miles):** *weaker* ad response due to the [[concept-billboard-effect]] — the store already reminds them.
- **Moderate band (≈ 4 to 14 miles):** *peak* response — close enough that travel costs aren't prohibitive, but far enough that customers need an advertising nudge to visit. This is the most responsive segment.
- **Far distances (> 14 miles):** *weaker* response due to high travel costs.

## The 'donut' takeaway
Because the innermost ring is redundant, the optimal targeting zone is **not a circle but a donut**, excluding the inner ring where ads add nothing and concentrating on the moderate-distance band where they change behavior (see [[quote-the-donut]]). You discover your brand's specific donut empirically via [[action-test-distance-bands]]. The pattern is asserted for stable-assortment categories in [[claim-stable-assortment-u-shape]].

## Enrichment context
Non-linear distance effects are supported conceptually and in related empirical domains (restaurant/OFD choice, proximity experiments). But the specific **three-band pattern, the 'donut' label, and the exact 4/14-mile thresholds are study-specific**, not a broadly accepted named model. A counter-perspective worth holding: optimal rings are **category- and density-dependent** — for micro-retail (coffee shops, small-format stores) in dense urban markets the relevant bands can be a walkable 0.5–1 mile, not 4–14.