How NearbyIndex Calculates Your Convenience Score

Most walkability tools give you a number. We give you the recipe.

At NearbyIndex, we believe a score you can't explain is a score you can't trust. That's why we're publishing exactly how our convenience score works — the categories, the weights, the math, and the trade-offs. Whether you're choosing a neighborhood, evaluating a rental, or just curious about your current address, understanding the methodology helps you interpret what the number actually means.

The 8 Categories

Your convenience score isn't a single measurement. It's a weighted average of 8 distinct categories, each measuring a different dimension of daily life:

The weights reflect a hierarchy of daily needs. Groceries and transit get 1.5× because you interact with them almost every day. Entertainment gets 0.6× because you visit a cinema a few times a month at most. This isn't arbitrary — it models how most people actually use their neighborhood.

How Each Category Is Scored

Every category score is built from three components, each capturing a different aspect of what makes a location convenient.

Count Score (0–60 points)

The count score measures how many relevant places exist within the search radius. But it uses a logarithmic curve, not a linear one. Why? Because the difference between 0 and 3 grocery stores matters far more than the difference between 15 and 18.

The formula:

countScore = 60 × log(1 + count × K) / log(1 + maxCount × K)

The K parameter (saturation constant) controls how quickly the score levels off:
• Restaurants (K = 0.3): Variety keeps adding value. The 15th restaurant still meaningfully expands your dining options.
• Groceries (K = 0.5): Balanced. Having 5–6 grocery stores nearby is great; beyond that, returns diminish.
• Parks (K = 1.0): Fast saturation. One or two good parks covers most people's needs. Ten parks doesn't make a neighborhood ten times better.

Each category also has a maxCount — the number of places that represents a "fully served" area. For restaurants, that's 25. For transit stations, it's 4 (since the data captures major hubs, not individual bus stops). Exceeding maxCount doesn't hurt your score — it triggers a density bonus instead.

Distance Score (0–25 points)

Having 50 restaurants in a 600m radius means little if the closest one is 590 meters away. The distance score rewards having at least one place that's genuinely close — within about 400 meters, or 40% of the search radius (whichever is smaller).

distanceScore = 25 × max(0, 1 − nearestDistance / closeThreshold)

If the nearest grocery store is 100m away, you get almost full distance points. If it's 350m away, you get a few. If it's beyond 400m, this component is zero (though you might still score well on count).

Density Bonus (0–15 points)

This is extra credit for areas that exceed the expected maximum. If a category's maxCount is 12 grocery stores and you have 20 within radius, the excess triggers a bonus:

densityBonus = 15 × min(1, log(1 + excess) / log(1 + maxCount × 2))

This rewards truly dense urban cores — the kind of neighborhoods where you can walk out your door and have a dozen options in any direction — without letting the bonus dominate the score.

The Compression Curve

Category scores are the sum of count + distance + density, which means the theoretical raw maximum is 100 (60 + 25 + 15). But in dense city centers, raw scores can exceed 100 because the density bonus stacks with strong count and distance scores.

To keep the final score on a meaningful 0–100 scale and maintain differentiation at the top, we apply an exponential compression above 60:

compressed = 60 + 40 × (1 − e^(−(rawScore − 60) / 50))

Below 60, scores are untouched. Above 60, it gets progressively harder to climb:

This matters for heatmap visualization. Without compression, Manhattan, central Paris, and downtown Tokyo would all cluster between 95–100, making the heatmap a uniform blob. With compression, a score of 90 feels meaningfully different from 85, and getting above 95 requires extraordinary density across every category.

Healthcare: The Special Case

Healthcare is unique because having three pharmacies is very different from having a pharmacy, a clinic, and a dentist. Our healthcare scoring uses sub-type diversity instead of treating all healthcare POIs equally.

We split healthcare into three sub-types:
• Pharmacy (pharmacies) — maxCount: 2, fast saturation
• Medical (hospitals, clinics, doctors) — maxCount: 4, slower saturation
• Dental (dentists) — maxCount: 2, moderate saturation

Each sub-type is scored independently and weighted by its share of the total maxCount. Then we add a diversity bonus: 5 points for each additional sub-type present beyond the first, up to 15 points.

So a location with 2 pharmacies scores lower than one with 1 pharmacy + 1 clinic + 1 dentist, even though the total count is higher in the first case. Real-world healthcare needs are diverse, and the scoring rewards diverse infrastructure.

The Minimum Count Rule

Categories have minimum thresholds. Restaurants need at least 3 nearby. Transit needs at least 1 station. Groceries need at least 1 store.

If a category falls below its minimum, the entire category score takes a 60% penalty (multiplied by 0.4). A single restaurant doesn't make a food scene. One pharmacy with nothing else doesn't constitute healthcare access.

This prevents areas with a single outlier POI from scoring misleadingly well. You need a baseline level of infrastructure before the full scoring kicks in.

The Overall Score

The final convenience score is a weighted average of all category scores:

overall = Σ(categoryScore × categoryWeight) / Σ(categoryWeight)

Since each category score is already compressed, no additional compression is applied at the overall level. The weights (1.5× for groceries/transit, down to 0.6× for entertainment) ensure the final number reflects what matters most for daily life.

Where the Data Comes From

NearbyIndex uses Overture Maps as its primary data source. Overture is a collaborative project backed by Amazon, Meta, Microsoft, and TomTom that curates and merges data from OpenStreetMap, Meta's mapping data, and Microsoft's building footprints.

This gives us:
• Global coverage — data for 32,000+ cities across every continent
• Verifiability — the underlying data is open and inspectable
• Regular updates — the dataset is refreshed as contributors add and correct data

For each category, we query specific tags (e.g., shop=supermarket, amenity=pharmacy, railway=station) within the defined search radius of any point.

What We Don't Score (Yet)

Transparency means being honest about limitations:
• Straight-line distance, not walking routes. We measure as-the-crow-flies distance. A river or highway between you and a grocery store isn't accounted for. Walking-route distance would be more accurate but is computationally expensive at global scale.
• No transit frequency. We know if a transit station exists nearby, but not whether buses come every 5 minutes or every 45.
• No safety or crime data. A well-served neighborhood with high crime rates still scores well. Safety is important but requires a separate data pipeline.
• Coverage varies by region. OpenStreetMap coverage is excellent in Europe and East Asia, good in North America, and patchier in parts of Africa and Central Asia. A low score might mean genuinely poor infrastructure — or it might mean the data hasn't been mapped yet.

We're working on improving several of these, particularly walking-route distance and transit frequency. For now, the score is most reliable in well-mapped urban areas.

Try It Yourself

Every score on NearbyIndex shows the full category breakdown — not just a single number. You can see exactly which categories are strong, which are weak, and what specific places are nearby.

Check any address at nearbyindex.com and see the methodology in action.