The Match & Enrich API generates a list of enriched companies based on your input criteria. Each company result includes a Match score (match_score
) and a Match method (match_method
) attribute.
This section will guide you on how to use these attributes effectively to filter and refine the results according to your specific requirements.
Match score
The match_score
attribute is a numerical value ranging from 0 to 1, which indicates the confidence level of the match between the input data and the retrieved company information.
Score value meaning
A score closer to
1
indicates a higher level of confidence in the match, while a score closer to0
suggests lower confidence.
During the scoring process, two primary factors are considered:
- The presence and completeness of the information used for the matching.
- The overall relevance of the information in establishing a match.
Completeness
This factor refers to the level of specificity in a particular data point.
For example, if the input address is specified down to the street level, but the API matches at the city level, the match score will be slightly lower compared to a street-level match, especially if there are no other strong signals supporting the same match.
This is done to consider the discrepancies in input and output street information, which typically serves as a highly reliable indicator of a correct match.
Relevance
In this specific context, the term refers to the probability of a particular data point resulting in a successful match.
For example, a company's website is highly relevant as it is generally unique and offers the best chance of a precise match. In contrast, a company name by itself is less relevant due to its common occurrence, making it better to combine it with additional validation, such as an address or phone number, for a more accurate match.
The scoring mechanism also considers the relevance of each Veridion data source used in the matching process. In general, information obtained from a company's website is granted greater weight due to its higher level of reliability compared to data obtained from social media profiles or other online mentions.
Tip
By using the
match_score
, you can establish a threshold to filter out company results that do not meet your desired level of confidence.For example, setting a
min_match_score
parameter of0.6
will effectively exclude any API result with a lower score.
Match method
The match_method
attribute is a string that provides information about the criteria used to match the input data with the Veridion data.
Match methods have different confidence levels. For example, when the only available inputs are the company name and country, the match is performed using the exact_name_country
method. This method generally has a lower level of confidence compared to the fuzzy_name_city
method, for example. The lower confidence is attributed to the common occurrence of names and the extensive geographical area being searched.
Note
Keep in mind that a lower confidence level in the match method does not necessarily indicate an incorrect match.
In such cases, the Match score becomes crucial in determining a more robust match, thereby compensating for the lower confidence level of the match method used.
Matching waterfall
Here is the complete list of available match methods (i.e., the Matching waterfall), arranged in descending order based on the level of confidence:
Match method | Description |
---|---|
website | Exact match with website |
website_name | Partial match with website and exact match with company name |
exact_name_street_or_postcode | Exact match with company name and street or postcode |
exact_name_phone_city_or_region | Exact match with company name and phone number and city or region |
fuzzy_name_street_or_postcode | Fuzzy match with company name and street or postcode |
fuzzy_name_phone_city_or_region | Fuzzy match with company name and phone and city or region |
exact_name_city | Exact match with company name and city |
fuzzy_name_city | Fuzzy match with company name and city |
exact_name_region | Exact match with company name and region |
fuzzy_name_phone_country | Fuzzy match with company name and phone and country |
fuzzy_name_region_country | Fuzzy match with company name and region and country |
fuzzy_name_phone | Fuzzy match with company name and phone |
exact_name_country | Exact match with company name and country |
Tip
By analyzing the Match method, you can determine how the match was made and decide if the result is relevant to your desired output.
For instance, depending on the input provided, you could decide that
fuzzy_name_city
is not a strict enough match. Alternatively, you might considerfuzzy_name_city
as a suitable match when accompanied by amatch_score
of over0.7
.
Difference between match_score
and match_method
match_score
and match_method
In general, a higher Match score indicates a more accurate alignment between the input criteria and the company result.
The Match method that generates the result also influences the Match score. For instance, using the exact_name_city
match method tends to yield a higher match score than the fuzzy_name_city
method, as it implies a more precise match.
"Match score" to "Match method" relation
It is important to keep in mind that the relationship between the match score and match method is not strictly linear or deterministic.
The
match_score
provides an overall measure of result relevance, influenced by various factors, such as input criteria, data quality, and the specific matching algorithm employed.
In other words, the match_method
is a qualitative indicator of the matching technique, while the match_score
is a quantitative indicator of the strength of that match.
Example
A match result with an exact_name_city
match method and a high match score (e.g., 0.8
) indicates a strong match between the input data and the Veridion data, covering both the company name and the city.
On the other hand, a match result based on a fuzzy_name_city
match method and an even higher match score (e.g., 0.85
) suggests a good partial name match, potentially differing by just a few letters in the name, but backed by more reliable and / or multiple data sources. This situation is common for widely used names, such as "US Financial Service” and “US Financial Services."
Tip
To filter results effectively, you can combine the Match score and Match method to establish filtering criteria tailored to your specific requirements.
For instance, you might initially retrieve only results with a
match_score
over0.6
and then prioritize them based on combinations ofmatch_method
andmatch_score
.