Data extraction and processing is inherently messy. While there is no ‘perfect’ in this domain, quality is a cornerstone of our approach and we are committed to continual improvement. Our customers seem to agree — check out our G2 reviews.
Principles
Minimize false positives. Prefer accuracy over coverage.
Multi source. Leverage a variety of 1p techniques we are uniquely good at, and complement with carefully selected 3p providers.
Flexibility. Customers can create their own custom version of many signals and/or mix Keyplay data to create their own derivatives.
Service. Hands on team will review and iterate fast.
Resilience. Keyplay's unique strength is tying together multiple signals with AI lookalikes to prioritize your accounts. This approach allows any individual data issue to be minimized through a state of the art scoring strategy.
Data Sources
Our data superpower is extracting 1p insights from webpages and job posts. We map each domain into a series of locations (pricing, resources, job board, etc), and extract raw data from each location as well as all job posts. Many of our signals are based on this scraped data.
While the majority of our data is 1p, we do have a few 3p vendors. We are picky about selecting vendors with high quality that complement our own data, and proactively work to minimize our reliance on 3p vendors.
Data Quality
Our data quality begins with an atomic concept of an account. Every account in our system has a unique final domain, as well as a linkedin URL where we were able to verify a bi-directional link between the domain and linkedin page. This helps reduce duplicate accounts, and minimizes cases of mismatched firmographics.
We prefer correctness over coverage — when we see trade-offs between gathering more data or being right, we will bias in favor of having less coverage but higher accuracy.
Data Freshness
We update our first party scraped data on a rolling cadence. Data that changes frequently (e.g. job posts) is scraped on a weekly basis, while more static data (e.g. about page) is scraped every quarter.
Third party data is updated every 1-2 months, depending on how frequently we expect to see changes.