Almost every "Google Maps ranking factors" article published in the last decade skips the single geographic primitive that actually determines Map Pack rankings: the S2 cell.
S2 cells are Google's hierarchical grid of the Earth, 30 levels of nested quadrilaterals, spanning from country-sized down to building-sized. Google's own Maps infrastructure is built on them. Google's geographic queries route through them. And Google's Map Pack ranking pipeline uses S2 occupancy at Levels 10–14 as the primary signal for "which businesses are eligible to appear for this search."
This is the technical explainer. What S2 cells are, why they matter, how to map your business's actual cells, and how to engineer occupancy that moves you into the Map Pack.
Part of our Google Maps SEO in 2026 cluster.
Run the free GeoGrid scan, see exactly which S2 cells you currently dominate and which ones you're invisible in.
What S2 Cells Actually Are
S2 is Google's open-source spherical geometry library, publicly available at github.com/google/s2geometry. It's used inside Google Maps, Google Earth, Google Ads, and the Geolocation API. It's also the geometric substrate under Google's local search ranking stack.
The Hierarchy
S2 divides the Earth into a hierarchy of 30 levels. At each level, every cell is subdivided into four smaller cells, creating an exponential tree.
| Level | Approximate Cell Size | Practical Meaning |
|---|---|---|
| 0 | 85,000,000 sq km | Continent / large region |
| 5 | 330,000 sq km | Small country or U. S. state |
| 10 | 81 sq km (≈31 sq mi) | Metro area neighborhood cluster |
| 12 | 5 sq km (≈2 sq mi) | Neighborhood |
| 14 | 320,000 sq m (≈0.12 sq mi) | City block cluster |
| 15 | 80,000 sq m | City block |
| 18 | 1,250 sq m | Small building / lot |
| 20 | 78 sq m | Building interior |
| 30 | 1 sq cm | Precision of atomic clock |
Each cell has a unique 64-bit identifier (cell ID). Cells at the same level are the same size. Cells at different levels nest cleanly, a Level-14 cell is exactly 1/4 of the containing Level-13 cell, and contains 4 Level-15 cells.
Why Google Picked This Design
Two reasons make S2 the right primitive for Maps-scale problems.
Reason 1: Equal-area cells globally. Longitude-latitude grids distort badly near the poles. S2 projects the Earth onto a cube and subdivides each face recursively, giving cells that are roughly equal size anywhere on the globe. Critical for uniform geographic queries at scale.
Reason 2: Fast spatial indexing. Cell IDs are designed so that nearby cells have similar numeric IDs, which means "find all cells intersecting this radius" becomes a range query on a single indexed column. Maps-scale performance requires this kind of trick.
The math is elegant. The practical implication for local SEO is that Google's entire geographic ranking pipeline runs on these cell IDs. Your job is to engineer strong signals in the specific cells where your customers search.
Why Local SEO Operates at Levels 10–14
Different queries engage different S2 levels. A search for "best restaurants in Las Vegas" pulls from Level-10 or Level-11 candidate pools. A search for "roofer near me" from a specific address pulls from Level-13 or Level-14 cells around that address. A hyperlocal search ("coffee shop near me" while walking) can engage Level-15 cells.
The useful operating band for local service businesses is Levels 10–14.
| Level | When it matters |
|---|---|
| Level 10 (≈31 sq mi) | Metro-level searches, city-wide rankings |
| Level 11 (≈7.5 sq mi) | Major district or town-center searches |
| Level 12 (≈2 sq mi) | Neighborhood searches |
| Level 13 (≈0.5 sq mi) | Sub-neighborhood searches, "near me" from a specific point |
| Level 14 (≈0.12 sq mi) | Block-cluster searches, precise "near me" queries |
Level 14 is the level we target for neighborhood spoke page deployment. At that resolution, each cell maps roughly to a neighborhood quadrant, and occupancy signals are granular enough to move rankings without being too granular to manage.
How Google Uses S2 Cells in a Maps Query
The K-cluster candidate-pool process at the core of Google Map Pack selection runs on S2 cells. The full flow:
- Searcher location resolved: Google reads the searcher's lat/lon (GPS, IP-derived, or explicit) and maps it to S2 cell IDs at multiple levels.
- Query radius drawn: For local-intent queries, Google draws a rough 1–2 mile radius around the searcher (wider for low-density areas, tighter for urban).
- Intersecting cells pulled: Every Level-14 S2 cell that intersects the radius is added to the candidate cell set.
- Business index queried: Google queries its business index for every business whose indexed location falls in those cells AND whose primary category matches the query intent.
- K cluster formed: The returned businesses form the K-cluster candidate pool, typically 20–60 businesses for a competitive urban query.
- Stage-2 ranking: Within the K cluster, Google ranks by S2 Occupancy weight + RSVM + Entity Trust + Distance Decay.
- Top 3 surfaced: The Map Pack.
The critical insight: your business is either in the candidate cell set or it isn't. If your S2 signals don't put you in the cells near the searcher, you never enter stage 2, you don't even get ranked.
This is why most operators stuck at positions 4–20 are not losing stage 2. They're failing at stage 1, cell by cell. And this is why neighborhood spoke pages, one per target S2 cell, are the single highest-use tactic in local SEO.
The Four S2 Cell Inputs That Decide Ranking
For every cell you want to rank in, four inputs stack up into your S2 Occupancy Score for that cell. These are the levers you engineer.
Input 1: Primary Category Match
Your GBP primary category must match the category Google uses to build the K-cluster for the target query. A roofing company with primary category "Home goods store" will never appear in "Roofer" K-clusters, regardless of how strong the other signals are.
See the GBP SEO guide for category testing and selection. This is the non-negotiable floor.
Input 2: Business Index Presence in the Cell
Google indexes your business in a cell when it has enough evidence of your presence there. Evidence includes:
- Your GBP verified address (if the address falls in the cell)
- Your neighborhood spoke page referencing the cell's landmarks, zip codes, and microclimate
- Your citations mentioning addresses or service areas covered by the cell
- Direction requests and check-ins from within the cell (NavBoost signal)
- Backlinks from other sites that reference the cell geographically
A single GBP address puts you in one cell. A hub-and-spoke silo with 10–20 spoke pages puts you in 10–20 cells. The scale difference is multiplicative.
Input 3: RSVM (Rank Embed Spatial Vector Matching)
Within each cell, Google scores how strongly your content matches the cell's geographic features, landmarks, roads, schools, parks, microclimate. Covered in detail in the ranking factors article.
The short version: paraphrased spoke pages get discounted by the Information Gain patent (#11,366,956). Genuinely unique spoke content with real landmarks and real microclimate data earns RSVM credit.
Input 4: Competitive Density
S2 Occupancy is relative, not absolute. A cell with three competitors ranks you against them. A cell with thirty competitors is harder to break into even with strong signals. Competitive density is why some verticals in some cities are dramatically easier to rank in than others.
The practical implication: start with the cells where competitive density is lowest (often sub-neighborhoods that mainstream competitors ignore), build from there outward.
How to Map Your Business's Actual S2 Cells
You don't need to install the S2 library to get actionable cell data. Three approaches from simplest to most technical.
Approach 1: Run a GeoGrid Scan (Simplest)
Our free GeoGrid scan samples your Map Pack position across 49 or 169 grid points in your service area. Each grid point corresponds to a location that maps onto Level-13 or Level-14 cells. The heatmap shows you where you dominate (strong cells), where you're weak (contested cells), and where you're absent (cells you need to target with spoke pages).
Approach 2: Use an S2 Cell Viewer
Open-source S2 cell viewers (s2.sidewalklabs.com is the most accessible) let you click any point on a map and see the cell ID at each level. Paste your business address to get your home cell; click neighborhood centroids to get adjacent cells.
This is useful if you want to match your neighborhood spoke URLs to specific Level-14 cell IDs, some operators do this for analytical clarity, though it's not strictly necessary.
Approach 3: Use the S2 Geometry Library Directly
For analysts comfortable with Python or Go, the S2 geometry library on GitHub gives you programmatic access. A 50-line script can:
- Take a service area polygon (your city boundary, for example)
- Enumerate every Level-14 cell intersecting the polygon
- Output cell IDs, centroids, and GeoJSON for each
- Feed that list into your spoke page deployment plan
This is how agencies running multi-city clients scale the hub-and-spoke silo deployment. For single-operator use, Approach 1 or 2 is usually sufficient.
Engineering S2 Occupancy: The Spoke Page Pattern
The deliverable that moves S2 Occupancy is the neighborhood spoke page. Done correctly, it's a standalone URL that targets a specific cell with unique content referencing that cell's features.
Spoke Page Requirements
Every spoke page needs:
- A URL that reflects hierarchy:
/las-vegas/summerlin/rather than/las-vegas-summerlin-roofer/, the latter is a combo page and dilutes authority - 80%+ unique content: Real landmarks, real microclimate, real cost and permit data, real neighborhood FAQs. Not paraphrased.
- Real photos from the cell: On-location images with GPS EXIF intact. Stock photos and AI-generated imagery hurt rather than help.
- Hub-and-spoke linking pattern: 1 link UP to the city hub, 2–3 ACROSS to non-geo service pages, 2–3 ACROSS to geographically adjacent spokes
- Schema: LocalBusiness + FAQ + Review markup
See the Hub-and-Spoke Silo deep dive for the architecture and WP All Import bulk deployment workflow.
Why 10–20 Spokes
Most commercially meaningful local service areas have 10–20 neighborhoods that carry the majority of the revenue potential. More than that is usually diminishing returns, you end up with thin, underused spokes. Fewer than that fails to cover enough Level-14 cells to shift K-cluster selection.
Timeline to Effect
New spoke pages take 6–8 weeks to show in the S2 occupancy signal after deployment and indexing. This is the single biggest reason the 12-week protocol takes twelve weeks. Shorter timelines miss the spoke-indexing window.
S2 Cells FAQ
What is an S2 cell in Google Maps?
An S2 cell is a unit of Google's hierarchical geographic grid. S2 divides the Earth into a tree of cells at 30 levels, from country-sized at Level 0 down to square-centimeter-sized at Level 30. Local SEO operates at Levels 10–14. Every location on Earth has a unique S2 cell ID at every level.
How do I find my business's S2 cell?
Use s2.sidewalklabs.com and enter your address. It returns cell IDs at every level. For Level 14 (the most relevant for local SEO), you'll get a cell covering roughly 0.12 square miles around your address.
How many S2 cells should I target?
10–20 Level-14 cells for most local service businesses. Each cell gets one neighborhood spoke page. Covering more cells with thin content hurts more than it helps, the Information Gain patent discounts thin spokes.
Does Google actually confirm S2 cells are used in ranking?
Google does not publicly detail its Maps ranking pipeline, but the S2 library is public (github.com/google/s2geometry), the infrastructure is documented, and reverse-engineering across hundreds of live GeoGrid scans confirms the per-cell ranking behavior. The specific term "S2 Occupancy Score" is our framework language; the underlying mechanic is real.
Can I rank in cells outside my storefront's cell?
Yes. That's the entire point. With strong neighborhood spoke pages, Entity Trust Compression, and BERT-optimized service pages, you can rank in S2 cells 10–20 miles from your physical storefront. Distance decay is roughly 3% of the ranking equation in 2026, it's the tiebreaker, not the decider.
What's the difference between S2 cells and zip codes?
Zip codes are postal boundaries that don't nest cleanly, don't have equal area, and don't reflect Google's internal geographic indexing. S2 cells are hierarchical, equal-area (approximately), and are the substrate Google actually uses. Zip codes still have RSVM value when mentioned on spoke pages, but they aren't the ranking primitive.
Does my city's S2 cell count matter?
Yes indirectly. Dense urban cities have more Level-14 cells per square mile than rural areas because the population concentration creates more distinct search neighborhoods. This doesn't change the methodology; it changes how many spoke pages are worth building for full coverage.
How do I measure my S2 Occupancy without running a scan?
You can't directly, there's no public "S2 Occupancy Score" dashboard. The closest proxy is GeoGrid coverage: the percentage of grid points in your service area where you hit the top 3 of the Map Pack. Our free GeoGrid scan measures this in 30 seconds.
Next Step: Map Your Current S2 Coverage
You can't engineer S2 Occupancy without knowing which cells you currently dominate and which ones you're invisible in.
Thirty seconds to start. Heatmap by email in two minutes. See exactly which S2 cells need spoke pages and which ones you already own.
If your scan reveals you're invisible in most of your service area and your vertical plus territory is open, apply for the Maps Domination Program™. Full S2 Occupancy engineering via hub-and-spoke silo deployment, completed in 12 weeks, top-3 or you don't pay.
Methodology from The Google Maps Domination Playbook by Nick Thompson. For the broader framework, see Google Maps SEO in 2026 and Google Maps Ranking Factors 2026.
External references:
