Note: This learning document is based on information found in the Maps Search Evaluation Guidelines. When in doubt, always refer to the official guidelines.

Rating for Distance FLOW:

.

 

 

.

.

FIRST, take a high-level look at the result in light of the queried term(s).

Before considering distance, make sure the result matches the user query intent in some meaningful way as described in Search Guidelines section 5.2.

If a result a suggestion is “non-relevant” for intent per the chart in 5.2, then rate the result “Bad” for intent. There is no need to consider distance.
If the result matches the primary intent do not demote for intent and proceed to evaluate the result for distance;
If the result does not match primary intent, but matches secondary intent, demote for intent—giving an initial rating of “Good” (-1) or “Acceptable” (-2) per the chart in 5.2 and/or examples, then proceed to evaluate distance.

NOTE: When the user’s query/intent is unclear, you may need search the area of locational intent first in order to determine what the user is actually looking for. This is often the case with single location POIs and unusual place names.

STEP 1. Determine locational intent (2.3 Location Intent/chart)

The locational intent will be one of the following:

the fresh viewport area
the user’s location,

o when the viewport is stale,

o when the user is inside the fresh viewport,

o or when the user explicitly queries “nearby” or “near me” etc.

near the named/specified location

o General: in/near a named neighborhood, locality, region etc. For general location intent, a single match within could be navigational; while multiple would each be rated excellent. zAny outside of the named locality would be demoted based on distance/density.

o Specific: on/near a named street. As long as at least one matching result is on the named street all other results would be rated bad. If there is no match on the street; then other matching results are rated according to distance/density.

2.Perform preliminary research based on location intent

Obtain the coordinates of the fresh viewport/user’s position OR other target locality etc. as described in Step 1, and enter this information in your preferred map application as the starting point for your real-world search for the queried business/POI, category or address.

TIP: Some map applications have more than one search function. For example, one popular map application has a “Directions” and a “Nearby” function that produce different types of results. “Direction” searches usually produce a set of matching street addresses or POI + street addresses in some cases. “Nearby” searches usually produce POI names + street addresses. It is a good idea to familiarize yourself with the various search functions of your chosen map application to understand what type of results can be produced from each.

3.Determine the intended result type

Depending on the intent type, the locational intent will be more or less important as described below.

Is intent a single location/non-chain business or point of interest (POI)?

o If there is only one clear business or POI match, a non-matching result would be demoted for “bad” for intent and distance is not critical.

Is intent a chain business with more than one location?

o If the intent is a chain business, distance from locational intent is critical.

Is intent a category/type of business or POI?

o If the intent is a category/type of business or POI, then distance from  locational intent is critical.

Is intent a specific single location address?

o If there is only one clear address match, a non-matching result would be demoted for “bad” for intent and distance is not critical.

.

 Is intent a partial address that has more than one match?

 

o If the intent is a partial address (for example a street name without a locality or a street number + common street name without a locality etc.), then distance from locational intent is critical.

Is intent transit-related? (a station, stop etc.)

o If the intent is a rail or bus station/stop, there are very specific rules with regard to demotions for distance. The major considerations for transit ratings per “5.16 Transit Intent” are:

Could the query intent be either the location/POI name or a transit (station/stop)? If the query is not clearly one or the other, the transit would be demoted for secondary intent (See the “Mall of America” example).
Could the query refer to a station name uses the locality name; OR could the query refer to a category “station” within the locality? If so, the station with the specific locality name would be rated “Navigational”; and other stations within the locality would be rated “Excellent.” Additionally, any other stations that fit the category but are outside of the named locality, would be demoted for distance according to the distance/density of matches. (see “Stockport Station” examples)
Is the transit query really a single, unique transit name for which only one matching result exists? If so, the matching result would be navigational and all others would be rated bad. Specific airports are a good example of this as are very unique station names. (See “12th Street Oakland Station” example)
4.Consider the distance/density of all possible results

Consider the density/number of possible results. “Density” most often comes into play for chain business and category queries; but it can also apply to partial address searches.

Many matches within area of locational intent FVP, locality etc. can mean harsher demotions for those outside of the target area. (5.6 Many Results/Starbucks example)
Few matches within area of locational intent can mean greater leniency for results outside of the target area. (5.7 Few Possible Results/Zara example, 5.8 Few Possible Results Greater Distance/Wartestrabe example, and 5.9 Rural Areas/American Legion example)

FINALLY, add distance demotions to any taken for intent to determine final relevance rating 

 

 

General Key points regarding distance:

Real-world matches: Always search for real-world query matches (not just those presented in TryRating) and consider the distance of any given result in light of all matches that exist in the intended location.
Understand how distance is viewed for viewport vs. user location intent:

o When the user’s position is the focus (either because the viewport is stale or because the user is inside the fresh viewport), distance is viewed as a radius from the user.

o When the viewport is the locational focus, distance is viewed as a fame that extends from the viewport shape

TIP: Take care not to be overly strict if one result is just slightly closer/further than another. Use common sense thinking about your experience as a user. For example, if a one result is just a bit further than another; but you as a user could imagine that not being a critical difference given the density of matching options, then rate both results similarly for distance.

User inside fresh viewport rule: When the user is within a fresh viewport, results are to be evaluated for distance from the user’s location. However, remember, that any matching results within the fresh viewport cannot be rated “bad” for distance alone. The lowest rating is acceptable in such cases.
Pay extra attention to transit rules/examples. In is important to reference the transit examples found in your guidelines when encountering a transit query. In addition to the rules listed above, the stops vs. station rule that says stations are to be considered stops as well as stations; but many stops are not stations.

.

.

.

.

.

..

Leave a Reply

Your email address will not be published. Required fields are marked *

Related Post