Service areas map

Rover · 2024–2025 · Search & Supply

Service Areas
& Travel Distance

+4% bookings · +7% match quality

From Hackathon to Product

Sitters historically set a circular radius — blind to real geography. The feedback was consistent: "I'm getting requests within my radius that are practically impossible." This started at a Maker Days hackathon and became a global feature.

As lead product designer, I was responsible for the full travel-distance service area experience — from exploratory research through experiment design, global rollout, and search integration.

Facebook post: sitter reports distance mismatch between stated and actual travel
+7%
match quality score
+4%
bookings
Research prototype — travel-distance polygon

Understanding What "Distance" Means

In moderated sessions, sitters reacted to their areas being redrawn, adjusted distance/time and travel mode, and talked through how far they'd go for different services.

Three design anchors: sitters think in miles and minutes; irregular road-based shapes feel more intuitive than circles; different services need different areas.

The unified model
Distance (1–100 km/miles) or time (1–60 min), walking/cycling/driving. House sitting, drop-ins, walks each get their own area. Existing radius backfilled automatically.

A Model Sitters Could Trust

Sitters can now define service areas by distance (1–100 km/mi) or estimated travel time (1–60 min), with walking, cycling, or driving modes. House sitting, drop-ins, and walks each get their own area — matching how sitters think about effort vs. reward.

Mapbox isochrones generate realistic polygons that respect roads, water, and barriers. Existing radii were backfilled automatically so sitters could refine rather than restart.

Search integration
Search now uses the new polygons instead of simple circles, stops surfacing sitters on the wrong side of major barriers, and weights travel-distance preferences in ranking alongside booking probability.
A/B test grid: service area states across two test waves

Validated & Shipped Globally

Two geographic A/B tests. Both waves showed better matching, better conversion, and strong sitter adoption. We committed to global rollout instead of extending the split.

Search-to-request +2–3%. Search-to-book +1.6–2.8%. Match quality +7%. Revenue per searcher +5–6% in US markets.

The initial conversion trade-off
Small initial dip (–1–2%) in new-owner conversion from stricter "outside service area" behaviour. The sitters who did show up were closer and more likely to accept — a good trade for better long-run quality.
+5–6%
revenue per searcher US
4–5mi
avg distance drop page 1 results
Next: Graduated Fees →