We started PredictLeads with the mission to find actionable business data in public documents.
HISTORY
Categorized news articles about companies was our first offering, with which our partners were able to display highly relevant news to their customers. But quickly we saw even bigger demand for more structured business data.
The main issue was that there was no way to extract such data with our existing classification approach. For example, there was no way to know which company was acquired and which one did acquiring. Or, who was hired to which position within a company.
Although we knew the solution to the problem, we wanted to build something for the long term. Today, after 6 months of building systems, gathering training data and QA processes, we’re proud to finally launch Business Signals API.
FUTURE
At its core is PredictLeads Entity and Relationship Extraction System (PERES) which performs 3 steps:
- It starts with scanning millions of public documents every day.
- Public document examples: news articles, blog posts, SEC filings, job openings, PR releases, industry news etc.
- System then extracts 20 different types of entities from the those documents.
- Entity examples: organizations, persons, locations, money, job titles, business divisions etc.
- At the last step, it tries to identify relevant relationships between these entities.
- Relationship examples: acquisitions, personnel changes, investments, expansions, office closures, recognitions etc.
Well, actually, there’s a fourth step, human supervision. We know that machines are not yet fully reliable, so most of the relationships are still human approved for highest quality.
To get a better picture what types of data are we talking about, here are two examples of data in JSON format:
{
"id":1656497,
"type":"signal",
"categories":[
"launches"
],
"title":"BMW launches ReachNow car-sharing service on Sep 19th 16'.",
"body":"BMW will launch its ReachNow car-sharing service in Portland on Sept. 19, the car-maker announced today.",
"domain":"bmw.com", "url":"http://www.geekwire.com/2016/bmw-unveils-details-reachnow-car-sharing-launch-portland-next-month/",
"found_at":"2016-08-15T02:00:00.000+02:00",
"removed_at":null,
"data":{
"location":"Portland",
"product":"ReachNow car-sharing service",
"date":"2016-09-19T02:00:00.000+02:00",
"organizations":[
"bmw.com"
]
}
}
Or
{
"id":1915491,
"type":"signal",
"categories":[
"invests_into_assets"
],
"title":"Domino Printing Sciences Plc invests into assets : factory on Aug 18th 16'.",
"body":"Cambridge-headquartered commercial inkjet printing business Domino Printing Sciences has opened its new £19m factory in China.",
"domain":"domino-printing.com",
"url":"http://www.insidermedia.com/insider/central-and-east/domino-printing-sciences-opens-chinese-factory",
"found_at":"2016-08-18T09:44:26.000+02:00",
"removed_at":null,
"data":{
"location":"China",
"amount":"25332000",
"assets":"factory",
"date":"2016-08-18T09:44:26.000+02:00",
"organizations":[
"domino-printing.com"
]
}
}
Above are 2 of 30+ possible types of signals. Full list is available here: https://predictleads.com/docs/#signal
APPLICATIONS
Types of applications are (almost) endless, but I will list a few we had in mind:
- Sales Enablement Solutions: empower customers to know everything about their prospects’ business without Googling around.
- Predictive Lead Scoring: high quality buying signals should be core data for predictive models
- Account Based Marketing or Sales: enable customers to generate automated highly personalized and relevant messages to their existing or potential customers.
Contact us for your API key or to find out more.
sales@predictleads.com
