Case Study
How CoffeeSpace used Crustdata to power their matching engine without stitching together a stack of data vendors
Company
Coffee Space
use case
Candidate and company data for talent matching
company overview
CoffeeSpace is a talent-matching platform for high-growth startups. Match quality is the whole product, a candidate who clears the bar has roughly a 90% chance of reaching the first interview. That takes more data than candidates can supply, so after Proxycurl and Bright Data fell short they made Crustdata their sole data partner, reaching 99 to 100% coverage.
99–100% | <24 hrs |
Candidate data coverage | Most data requests turned around same day |
Toward job matches that don't go unanswered
CoffeeSpace is a talent-matching platform for high-growth startups. They connect engineers and operators who have startup or former-founder experience with Series A to C companies where one good hire can make a big impact. Carin Gan and her team built the platform around a single promise, that a candidate who gets matched has a real shot at the job.
CoffeeSpace deliberately limits how many candidates they let in and how many roles they show each one. The tradeoff is volume for relevance. A candidate who clears the bar has roughly a 70% chance of reaching the first interview, because CoffeeSpace has already read the rejection signals a company sends, the strict tech-stack requirements and the missing startup experience, and filtered against them before anyone applies. No one is left applying into the void, with no response from the company they applied to.
Matching a candidate and a job that fits them relies heavily on the data, far more of it than a candidate can be asked to supply. Early on, people filled out forms on the platform or pasted in their LinkedIn, which was slow, and nobody wanted to type out long paragraphs. So, the CoffeeSpace team decided to pull the public signal themselves - GitHub activity, socials, full LinkedIn histories, and keep the candidate's own input down to the few things only they know such as visa status and compensation expectations.
Before Crustdata
Pulling that data cleanly turned out to be the hard part. CoffeeSpace cycled through a series of providers and each had its own distinct drawbacks. They started on Proxycurl, which later shut down. They tested Aviato for proxy data. They tried BrightData, and while the volume was there, the quality was not.
What CoffeeSpace set out to build
The pattern taught CoffeeSpace what they actually needed, and it was not another proxy service. Most vendors hand back data already run through their own matching algorithm, which meant CoffeeSpace would be building on top of someone else's judgment about who fits. What they wanted was the raw material, clean, structured, accurate data they could control and shape into a model of their own. The matching logic was the thing CoffeeSpace was in business to own. The data underneath it just had to be good enough to trust, and complete enough that the team was not stuck cleaning it before they could use it.
What changed using Crustdata
The moment Carin and her team decided to choose Crustdata was when Daniel, our AE at Crustdata, sent over sample data, and for the first time it came back clean, useful, and structured the way CoffeeSpace needed.
A one-on-one session with Jai, our FDE, introduced Carin to the wide range of data points Crustdata offered. Crustdata carried company data too, fundraising milestones, traction, employee counts, the exact signals CoffeeSpace uses to tell a genuine high-growth startup from a name. That lets the team build their company categorization and proxy startup-caliber experience directly, without cutting separate API deals with Crunchbase or PitchBook.
It works because of three things. Crustdata supplies clean, structured data on both the person and the company. CoffeeSpace's model reasons over it, categorizing companies and scoring fit. And CoffeeSpace decides what a strong match means for a given role. Take the clean data away and the model goes back to guessing on half-empty fields the team would have to repair by hand.
Working with the team
Instead of filing tickets and waiting on email, CoffeeSpace has a dedicated Slack channel with our team, including the co-founders, where a problem gets raised and resolved inside the same conversation. Carin Gan flags bugs directly and they get fixed with a quick turnaround.
The bottom line
The numbers CoffeeSpace sees now are ones they could never get from candidate forms.
Crustdata delivers near-total candidate coverage, 99 to 100% on the profiles they need. About 95% of a data request lands within 7 days and most of it inside 24 hours, and when an internal bug does slip through it gets flagged and cleared fast. Downstream, the platform averages one offer and one placement every week.
CoffeeSpace set out to own their matching model and needed a data layer clean enough to build on, and they ended up with one partner sitting underneath the entire product, people data and company data in one place, close enough to fix a bug over Slack.
What CoffeeSpace built is a matching engine that keeps their promise, a curation layer that gives a candidate a real shot instead of another application into the void, running on data clean enough to trust. The judgment was always theirs. Now they have something solid to stand on.