Google's own setup keeps them apart
Its guidelines put the street address in Address Line 1 and the suite in Address Line 2. They also say a suite number "may" be included.
Google's Business Profile guidelinesA suite number helps customers find your door. It doesn't decide where you rank.
Google places your business by its street address, so "101 Baltimore St" and "101 Baltimore Street, Ste 4090" point to the same building. That makes the suite number a decision about your customers, not a NAP-consistency risk. We include it when visitors need it. In a few situations we leave it off your Google Business Profile and still use it on your website and other listings.
5 of 5 listings point to this building
Every listing points to the same building. Google ties your business to the street address, treats "Suite," "Ste" and "#" as the same thing, and can place you even when the suite is missing.
VisitorInside the building, the suite number is what gets someone to your door. A missing suite leaves them looking, and a wrong one sends them to another office.
Illustrative example built on the address used in this guide. The other tenants and suite numbers are made up.
The suite number is a detail for visitors.
Its guidelines put the street address in Address Line 1 and the suite in Address Line 2. They also say a suite number "may" be included.
Google's Business Profile guidelinesGoogle and Bing understand abbreviations and small variations, such as "Suite" vs. "Ste," so these don't break NAP consistency.
WhitesparkJoy Hawkins of Sterling Sky, quoting listings expert Nyagoslav Zhekov, says Google can work out the address without the suite if the rest of the listing is accurate. Her advice: unless you're in a huge building where it matters, don't stress about it.
Local Search ForumMoz's Miriam Ellis notes they're not considered a big deal for SEO, though they can confuse customers.
Moz Q&AWhere accuracy still matters is customers. A directory that shows the wrong suite sends people to the wrong door, so keep it right on the listings people actually use.
That's the deciding question. Find your situation below.
| Your situation | On your Google Business Profile | On your website, schema markup and other listings |
|---|---|---|
| Customers walk in and need the suite to find you (shared retail, medical or professional building) | Include itin Address Line 2 | Include it |
| You have the whole building, or the suite tells visitors nothing | Optionala judgement call | Include it if it's part of your mailing address |
| Your own staffed office inside a coworking or executive-suite center | Often left offVisitors check in at reception, and the center's suite number is often shared with its mail-only clients | Include it wherever mail and visitors need it |
| Service-area business with the address hidden | Not shown publiclyEnter the full address so Google can verify it | Include it if you publish the address |
| Mail-only or virtual office address | Not eligiblefor a profile at all. Suite numbers can help Google identify ineligible virtual office locations. | Use it as a mailing address only |
Under Google Business Profile Help, automated spam and verification systems actively scrutinize and restrict addresses tied to virtual offices, private mailboxes (such as The UPS Store), and co-working spaces.
A classic sign of a virtual office is multiple businesses sharing the same suite number, and such cases trigger automatic scrutiny by Google's spam and verification systems.
A suite number won't separate you from a competitor, can't be invented, and won't make an ineligible address eligible.
When two businesses in the same category share an address, Google often shows only the stronger one for a search. A suite number doesn't make your location a separate address (BrightLocal). Practitioners have also found that separate suites don't help same-category businesses in one building escape that filter (Moz Q&A).
Don't create a "Suite 100-A" and "Suite 100-B" to split businesses at one address. Google's guidelines require your actual, real-world location, with nothing in the address lines that isn't part of it (Google).
A rented mailing address you don't operate from can't have a Business Profile. An office in a coworking space qualifies only if it has clear signage, receives customers during business hours, and is staffed by your own employees during those hours (Google).
Not for rankings. Google places you by the building, so a missing suite isn't a ranking problem. Fix any listing that shows the wrong suite, because that sends customers to the wrong door.
No. Google and Bing treat common abbreviations as the same address.
Google's address help page says to include building, suite and floor details, while its main guidelines say they may be included. We include the suite whenever it helps customers. When we leave it off, the street address on your profile is still complete and accurate.
Your website and schema markup should show your full mailing address, including the suite. Google can work out your address with or without the suite, so the difference doesn't create a conflict.
Keep permanent signage with your business name. Google's guidelines expect it at any address you show on your profile. Keep your lease or a utility bill handy too, in case Google asks you to verify the location.
Tell us how your location is set up. We'll tell you what belongs on your Google Business Profile, your website and your other listings before anything changes.
Ask About Your AddressGoogle's internal map database gives a business's position, building and floor their own structures, and none of them is a suite number.
In Geostore, a business is placed by its geometry, a point or storefront shape, and by links to the floor, building and entrances around it. Its address is a separate postal attribute built from parsed parts like the street number and street, and the documented address model has no dedicated suite field.
None of the 72 ranking signals recovered from a 2026 Geostore binary is tied to an address. The leak doesn't reveal how much any field weighs. What it does show is consistent with the guide's main point: a suite number helps people find the door, but it isn't how Google locates or ranks the business.
Simplified from Geostore classes in the 2024 Content Warehouse leak
TYPE_ESTABLISHMENT_POI
point + storefront shape = map position
geometry_precision_meters (accuracy)
The business has an address
street number, checked against ranges
street name, linked to one street feature
city, state and postal code
unparsed, most specific text
where a suite would sit (our reading)
A suite number lands here as textkept out of the address entirely
The business is linked to
business links to it via RELATION_ON_LEVEL
its own address is always empty
lists its levels and floor count
routing endpoints, e.g. the road segment for the 100–200 block of Main Street
If Google keeps a suite number, it sits in the address text. What puts a business on the map is its own geometry and its links to the floor, building and entrances.
A suite number, if Google keeps one, is unparsed text in addressLines. The documented model gives it no link to a map feature, a floor or a building.
The map position comes from the business's own point and storefront geometry. Its floor, building and entrances are separate features joined to it by relationships.
Two bodies of material describe Geostore, Google's map database. Both show how it is structured, not how much any field counts.
Internal API documentation for Google's systems, now mirrored on HexDocs. Its Geostore* classes describe how map features, addresses and the relationships between features are stored. Most references below link to these pages.
"Google Map Dissected" cross-checks that leak against a more recent binary with a non-public Geostore scope, Maps protocols and network traffic. It catalogs 1,623 protobuf descriptors, 4,013 messages and 10,936 Geostore declarations.
What a schema can and can't prove. A schema shows which fields exist and how they connect. It doesn't show how much each one weighs in ranking: Resoneo notes the coefficients are missing. Some fields were also stripped from the binary, including 25 slots of the main feature record and 4 fields of its trust data.
The structure is written in the documentation.
SuggestsOur own reading of it. These notes sit in dashed boxes below.
An address is a postal attribute attached to a business: parsed parts plus a block of free text, never including the business's own name.
TYPE_STREET_NUMBER
Checked against the street's known number ranges.
Street component
Linked to one street feature, so "St" and "Street" land on the same street.
Suggests
Unparsed text in addressLines, with no link to a map feature.
GeostoreAddressProto is "attached to features to identify their address(es)." Addresses never include the feature's name; the docs' example is a pizza shop whose address is just "123 Main Street." On the main feature record, addresses follow the postal hierarchy, not the political one.
The component list holds parsed parts such as the street and city, ordered from the street number and street outward. Index 0 is the most specific component (GeostoreAddressComponentProto).
addressLines holds the unparsed portion of the address. It is separate from the parsed parts and must be the most specific portion of the address.
SuggestsA suite that isn't parsed as its own component sits here, as text. The docs never name suites, so treat this as our reading.
The component type field names TYPE_FEATURE and TYPE_STREET_NUMBER. Most components map to map-feature types like country, locality or route, and a few have none, such as PO boxes. The full list lives in source code that wasn't leaked. Neither the address pages nor the main feature record define a suite, unit or sub-premise type.
A street component can be linked to the actual street feature through feature_id, and formatting follows that link to pick the name. Google also keeps an AreAddressComponentsEquivalent() check for comparing components. That's why "101 Baltimore St" and "101 Baltimore Street" land on the same street.
Road segments carry address-number ranges (GeostoreAddressRangeProto). Resoneo found that autocomplete grades a street number as IN_RANGE, IN_GAP, OUTSIDE_MINMAX or OUTSIDE_SEARCH_RANGE. The street number is the part of the address Google validates against the map; nothing comparable is documented for suites.
A business's position is stored as geometry and as links to other map features, separately from its address.
On the main feature record, point "defines the geometry of the feature," and center is the conceptual center used for routing. Only business features (TYPE_ESTABLISHMENT_POI) get a storefront geometry (GeostoreFeatureProto). A geometry_precision_meters value says the true location lies within a circle of that radius (v0.4.0).
TYPE_COMPOUND_BUILDING features carry building data such as the number of floors and the list of their levels (GeostoreBuildingProto).
A level feature represents a logical level such as a floor. Anything on that floor points to it with RELATION_ON_LEVEL, and the level records which building it belongs to. Its number gives height in floors: 0 is ground, 1 is the first floor, -3 is three levels down. A level's own address field "should always be empty" (GeostoreLevelProto). For indoor features, Google prefers RELATION_ON_LEVEL to other ways of ordering things vertically (GeostoreVerticalOrderingProto).
A relation records a geographic or logical relationship between one feature and another (GeostoreRelationProto). Google's public Vector Tile API gives the example of a building occupied by a given business (Relation).
An access point links a business or building to another feature. The docs' example: the access point for a building at 123 Main Street would likely be the road segment for the 100–200 block of Main Street. Indoor access points record the level they're on (GeostoreAccessPointProto). Entrances are features too, referenced by the places they serve (GeostoreEntranceReferenceProto).
SuggestsWhen Google knows which floor or entrance a business uses, it models that with these links, not by reading "Ste 4090." For most small businesses in buildings without indoor mapping, the point and the building are what Google has.
Every value has a recorded source and trust level, and a new value has to clear a bar before it replaces the old one. All figures here come from Resoneo's analysis unless linked otherwise.
Each field records its field_path, provider and dataset, and the provider catalog holds 793 sources. One business can draw on several sources at once, with a different origin for each field. Addresses and relationships carry field-level metadata too (GeostoreAddressProto, GeostoreRelationProto).
Directory data sits in the middle of the scale:
SUPER_TRUSTED64Highest tierTRUSTED48A reliable sourceYP_FEEDS40Yellow-pages feeds: data bought or ingested from business directoriesNOT_TRUSTED32BLOCKED16UNKNOWN0Lowest tierGeostore exposes three merge methods: CONFLATION_PICK_FIRST_VALUE (the default), CONFLATION_UNION_CSV and CONFLATION_SUM. The default takes the first admissible value in priority order rather than letting sources vote. Resoneo cautions that this is the machinery Geostore holds, not proof that every Maps attribute is merged this way.
Each attribute also has an edit barrier: the level a future edit must reach to replace it. The formula wasn't recovered. In practice, changing a value in Business Profile doesn't necessarily replace the old value right away.
Existence states include closed, removed, moved, rebranded, duplicate, replaced and spam, each confirmation stored with its origin and timestamp.
SuggestsA directory feed that formats a suite differently doesn't outvote a more trusted value by default. And if you remove a suite from a profile, Google may keep showing the old one for a while.
None of the ranking signals recovered from the binary is tied to an address, and their weights aren't known. All points here are from Resoneo's analysis.
Ranking signals recovered from the binary
Described as address-specific
Signal weights recovered
None of the 72 RankSignalProto.Signal values is described as address-specific. The closest are APPROACHABILITY, TOTAL_ROAD_SEGMENT_USAGE and LANDMARK, which are named but whose formulas aren't given.
The coefficients are missing, so the weight of any signal is unknown. That cuts both ways: the leak can't prove a suite number carries zero weight, and nothing in it gives a suite one.
It adapts to how dense the area is and to the query.
The article doesn't use the words NAP or citations. Its advice that name, address, phone, category, URL and brand should "tell the same story" is the author's interpretation, not a field in the data. Its sharpest practical line: "An identity error is worse than a poorly optimised field."
Open questions. The article doesn't name the fields Google uses to detect duplicates, and it doesn't cover how businesses sharing a building are filtered.
The internal documentation backs the guide's position: a suite number is a detail for visitors, not how Google places or ranks a business.
A business is placed by its point and storefront geometry and by its links to floors, buildings and entrances. The address is a separate postal attribute.
The documented parts are the street number, street, city and other postal parts. The most specific free text goes in addressLines, kept as plain text with no link to a map feature.
Street names link to one street feature, and Google keeps an equivalence check for address parts. "St" vs. "Street" or "Suite" vs. "Ste" is not a meaningful difference.
None of the 72 recovered signals is address-specific. The weights are missing, so this is strong consistency with practitioners' findings, not proof of zero effect.
Moves, duplicates, rebrands and spam are tracked states. A wrong street number, an old address or a duplicate listing is a real problem. A missing or reformatted suite isn't.
Trust levels and the edit barrier mean directory feeds rank below trusted sources, and a profile edit may take time to replace the stored value.
GeostoreAddressProtoAddress structure, addressLinesGeostoreAddressComponentProtoComponent types, feature linksGeostoreAddressRangeProtoStreet-number rangesGeostoreFeatureProtoGeometry, storefront, related featuresGeostoreLevelProtoFloors, RELATION_ON_LEVELGeostoreBuildingProtoBuilding featuresGeostoreRelationProtoRelationships between featuresGeostoreAccessPointProtoRouting endpointsGeostoreEntranceReferenceProtoEntrancesGeostoreVerticalOrderingProtoIndoor levels|
⚖️ How a law firm pulled ahead of 4 competitors in AI search, across Google, ChatGPT, and Gemini → |
After you complete this form you’ll be prompted to schedule a Complimentary Zoom Strategy Session with Chaz