# PredictLeads

> PredictLeads is a company signal data provider that crawls company websites, more than twenty million news and press sources, and public filings to produce structured, dated, source-linked records across job openings, technology detections, funding events, news events in thirty-seven categories, key customer relationships, similar companies, products, and firmographics for more than 120 million companies, delivered through a REST API, webhooks, flat files, and a Model Context Protocol endpoint, starting free at 100 API calls a month and $40 a month beyond that.

- Category: Buying Signals & Intent (https://saastracker.org/categories/intent-signals)
- Website: https://predictleads.com
- Starting price: $0 for the first 100 API calls per month, then a $40 monthly minimum
- Free plan: Up to 100 free API calls per month, self-serve, with no card required to start building against the API.
- Free trial: A free tier rather than a time-limited trial, available at sign-up
- Founded: 2015, HQ: Ljubljana, Slovenia, Ownership: Venture-backed, lightly funded
- Profile last reviewed: 2026-08-22
- Canonical profile: https://saastracker.org/products/predictleads

## Overview

PredictLeads is the raw material rather than the finished product. There is no visitor pixel, no Slack alert, no sequence, and no scoring model that decides which accounts you should call. What there is instead is a very large, carefully maintained body of company signal data with a primary source URL attached to every record, exposed through an API that a developer or a Clay table can query. If the rest of this category sells you a workflow, PredictLeads sells you the facts underneath one.

The company was founded in 2015 by Miha Stanovik, Roq Xever, and Matic Perovsek in Ljubljana, Slovenia, went through Y Combinator in the summer of 2019, and has stayed deliberately small and data-focused ever since. That is an unusual profile in a category dominated by heavily funded American application companies, and it shows in the product: PredictLeads is the vendor other vendors buy from, with partner integrations including Clay, Dealroom, and FactSet.

The coverage numbers are the argument. More than 120 million companies with firmographics, 9.8 million active job listings from 2.7 million companies plus more than 270 million historical job records going back to 2016, more than 54,000 technologies detected across 86 million companies with 1.4 billion total detections, more than 9 million news signals across 37 event categories, lookalike data on more than 18.5 million businesses, financing events, product listings, and key customer relationships extracted from case studies and testimonials by logo recognition.

Pricing follows the same philosophy: pay as you go, published, and self-serve. The first 100 API calls a month are free. Beyond that there is a $40 minimum monthly charge with credits at $0.04 each, falling to $0.02 above 5,000 calls, $0.01 above 10,000, $0.004 above 100,000, and $0.002 above 500,000. Enterprise adds full flat-file delivery via S3, Google Cloud Storage, or SFTP with unlimited company tracking and history back to 2015. A startup can genuinely start for nothing and scale the bill with usage.

## How it works

1. PredictLeads crawls continuously rather than licensing data from brokers. Company websites including career pages, product pages, and case study pages are checked on average daily, more than twenty million news outlets and press release sites are recrawled as often as every eight minutes, and public sources including regulatory filings are monitored around the clock.

2. Raw pages are parsed into structured records. A careers page becomes dated job opening records with title, location, and first-seen date. A homepage becomes technology detections. A press release becomes a news event classified into one of thirty-seven categories such as a product launch, a funding round, a partnership, or an acquisition. Every record keeps the primary source URL that produced it.

3. Quality control is a stated part of the operation rather than an afterthought. A team of more than ten specialists reviews thousands of records daily and automated anomaly detection runs continuously, which is what stops a site redesign from silently deleting a company's entire technology profile overnight.

4. You consume the data on your own terms. REST API endpoints answer queries per company, real-time webhooks push new signals as they are detected, flat files ship on daily, weekly, twice-monthly, or monthly schedules for warehouse loading, and a Model Context Protocol endpoint lets an AI agent query the signal set as a tool. Nothing about the delivery model assumes a human is looking at a screen.

## Best for

Technical go-to-market teams and developers building their own signal engine, Clay users who want dated source-linked signals inside their tables, and product teams embedding company intelligence into their own application, all of whom want data rather than another dashboard.

## Not the right fit for

- Anyone without engineering or Clay-level technical capability. There is no application to log into and no alert that arrives on its own, so a non-technical sales team will get nothing from this on its own.
- Teams that want website visitor identification. PredictLeads knows about the outside world and nothing about your own traffic; it cannot tell you who visited your pricing page.
- Buyers who want scoring and routing built in. The API returns facts; deciding that a Series B plus three engineering hires means an account is in market is entirely your logic to write.
- Companies that want person-level signals. This is company-level data. Job openings tell you a company is hiring, not that your champion just moved.
- Small teams needing a fast, non-technical result this week. The self-serve free tier is genuinely free, but turning API responses into someone's task list is a build, not a purchase.

## Features

### Signal datasets

Eight distinct datasets, each dated and traceable to a primary source.

- **Job openings**: 9.8 million active listings from 2.7 million companies plus more than 270 million historical records since 2016, sourced from company career pages rather than job boards, which is what makes the first-seen dates trustworthy.
- **Technology detections**: More than 54,000 technologies tracked across 86 million companies with 1.4 billion total detections, so you can trigger on a competitor's tag appearing or a prerequisite technology being installed.
- **News events**: More than 9 million signals classified into 37 event categories including product launches, partnerships, acquisitions, expansions, and leadership changes, drawn from more than 20 million news and press sources.
- **Financing events**: Funding rounds with amounts and dates, which is the classic budget-just-arrived trigger, delivered as structured records rather than a news headline.
- **Key customers**: Supply chain and vendor relationships extracted from case studies and testimonials using logo recognition, which is a genuinely unusual dataset and the basis for partner-ecosystem prospecting.
- **Similar companies**: Lookalike data on more than 18.5 million businesses, used to expand from a set of best customers into a target list without hand-building firmographic filters.
- **Products**: Product offerings extracted from company pages, which supports positioning research and competitive tracking at scale.
- **Company firmographics**: Descriptions, locations, corporate hierarchies, and revenue estimates for more than 120 million companies, so signals arrive attached to a usable company record.

### Data provenance and quality

The part that separates this from a resold database.

- **Primary source URLs on every record**: Every data point traces back to the page that produced it, so a rep can check the claim before acting on it and a disputed signal can be verified rather than argued about.
- **First-party crawling**: Data originates with PredictLeads rather than being licensed from brokers, which means gaps and errors are fixable at source instead of inherited.
- **Human review team**: More than ten specialists review thousands of records daily, backed by automated anomaly detection running continuously.
- **Dated records with history**: Signals carry dates and history goes back to 2016 for job data, with Enterprise access to history from 2015, which is what makes backtesting a signal definition possible.

### Freshness

Explicit crawl cadences, published rather than implied.

- **News recrawled as often as every eight minutes**: For event categories where speed is the whole value, such as funding announcements, the crawl interval is measured in minutes rather than days.
- **Job data refreshed at least every 36 hours**: A stated minimum rather than a best-effort claim, which matters because a stale job opening is worse than no signal at all.
- **Company websites checked daily on average**: Technology detections and product listings follow that cadence, so stack changes surface within days rather than quarters.
- **Continuous crawling**: Operations run around the clock rather than in a nightly batch, which is what allows real-time webhooks to be meaningful.

### Delivery

Four delivery models, none of which is a dashboard.

- **REST API**: The primary interface, queried per company or per dataset, metered by API call with published per-credit rates.
- **Real-time webhooks**: New signals push to your endpoint as they are detected, which is what makes triggered automation possible without polling the API on a schedule.
- **Flat file delivery**: Daily, weekly, twice-monthly, or monthly files via AWS S3, Google Cloud Storage, or SFTP on the Enterprise plan, for teams loading signals into a warehouse.
- **MCP endpoint**: A Model Context Protocol interface so an AI agent can query the signal set directly as a tool, which fits the way technical teams increasingly build go-to-market automation.
- **Partner integrations**: Available inside Clay, Dealroom, and FactSet, which means a Clay user can consume PredictLeads signals without writing a line of code.

### Commercial model

Pay as you go, published rates, no contract required.

- **Free tier of 100 API calls a month**: Enough to build and test an integration properly before any money changes hands, and self-serve throughout.
- **Volume-tiered credit pricing**: $0.04 per credit at the entry level falling to $0.002 above 500,000 calls, so the unit cost drops by a factor of twenty as usage scales.
- **Forty dollar monthly minimum**: The only floor on the paid path, which is a very low commitment for a data provider of this scale.
- **Enterprise flat file access**: Full dataset delivery with unlimited company tracking and history from 2015, quoted individually.

## Use cases

- **Clay user building a signal-driven table**: The team wants to prospect accounts that just started hiring for a role their product supports, but the job data available inside their existing enrichment is stale and undated. Outcome: PredictLeads is called from within Clay, returning dated job opening records with first-seen dates and source URLs, so the table filters to companies that posted the role in the last fortnight rather than at some point in the past year.
- **Technical founder building a custom signal engine**: Every packaged signal platform bundles a workflow the team does not want and charges thousands a month for scoring logic they would rather write themselves. Outcome: Webhooks push funding, hiring, and technology events into their own service, which applies its own scoring and writes tasks straight into the CRM. The bill is $40 a month plus usage rather than a five-figure annual contract.
- **Product team embedding company intelligence**: Their own application needs to show customers what is happening at the companies in their portfolio, and building a crawling operation is not a reasonable use of engineering time. Outcome: The REST API supplies news events, funding, and hiring data with source URLs users can click through to, and volume pricing at a fraction of a cent per call makes the unit economics work inside their own product.
- **Revenue operations lead backtesting a signal thesis**: The sales team believes hiring signals predict deals but nobody has evidence, and committing to a signal platform on a hunch is expensive. Outcome: Historical job records going back to 2016 let them test the thesis against closed-won accounts before buying anything, and the answer determines whether the motion is worth building at all.

## Pricing

Self-serve pay as you go metered by API call, with volume-tiered per-credit rates and a monthly minimum, plus a quoted Enterprise flat-file plan.

- **Free**: $0 per month. Up to 100 API calls per month; Full access to the API surface; Self-serve sign-up. Enough to build and validate an integration, not enough to run a production motion.
- **Pay as you go, 101 to 5,000 calls**: $40 minimum plus $0.04 per credit per month. Standard support; No contract; Published per-credit rate. The realistic starting point for a small team, and cheaper than any packaged signal platform.
- **Pay as you go, 5,001 to 100,000 calls**: $0.02 falling to $0.01 per credit per month. Priority support from 5,001 calls; 24/7 support from 10,001 calls; Unit cost halves at each threshold. The volume band where most production go-to-market automations land.
- **Pay as you go, above 100,000 calls**: $0.004 falling to $0.002 per credit per month. 24/7 support; Sub-cent unit economics; Suitable for embedding in your own product. At $0.002 a call the data becomes cheap enough to use inside a product you sell.
- **Enterprise**: Custom quoted. Full flat file access via AWS S3, Google Cloud Storage, or SFTP; Unlimited company tracking; Historic data from 2015; 24/7 support. The right shape if you are loading the whole dataset into a warehouse rather than querying per company.

Billing notes:

- Every rate is published on the pricing page, so the cost of a planned volume can be calculated before you sign up rather than negotiated.
- The $40 monthly minimum is the only floor on the paid path, with no annual contract and no seat licence.
- Unit cost falls by a factor of twenty between the entry band and the highest volume band, which rewards consolidating your data buying with one provider.
- Support level improves with volume rather than with a separate purchase: priority support above 5,000 calls and round-the-clock support above 10,000.
- Flat file delivery and full historical access are Enterprise entitlements, so warehouse-scale use is a quoted conversation rather than a self-serve one.

Value assessment: For a team that can write code or drive a Clay table, PredictLeads is the cheapest credible signal data on the market by a wide margin. A hundred free calls a month to prototype, a $40 floor to go live, and per-call rates that fall to fractions of a cent at scale, against a category where packaged platforms start at thousands of dollars a month for signals they source from providers like this one. The value collapses entirely if you cannot build: there is no interface, no alerting, and no scoring, so a non-technical buyer will spend $40 and get nothing. Judged as a data purchase rather than a software purchase, the coverage numbers, the published crawl cadences, and the source URL on every record make this unusually good value.

## Strengths

- Every record carries a primary source URL, which is the single most useful quality property in signal data and something most vendors cannot offer.
- Published crawl cadences rather than vague freshness claims: news as often as every eight minutes, job data at least every 36 hours, company sites daily on average.
- Coverage genuinely at scale, including 120 million companies, 270 million historical job records since 2016, and 1.4 billion technology detections.
- Data is crawled first-party rather than licensed from brokers, so errors are fixable at source and the provenance chain is short.
- Pay as you go pricing published in full, starting free and falling to fractions of a cent per call, with no contract and no seat licences.
- Four delivery models including webhooks, flat files, and an MCP endpoint, so the data fits whether you are building a service, a warehouse, or an AI agent.
- Available inside Clay, which lets a non-engineer consume the signals without writing code and is the most practical route for most small teams.
- Unusual datasets such as key customer relationships extracted by logo recognition and lookalike data on 18.5 million companies, which are hard to source elsewhere.

## Limitations

- There is no application. No dashboard, no alerts, no inbox, and nothing that tells a salesperson what to do this morning without you building it.
- Company-level only. Nothing about individual people, so champion job changes and person-level intent are outside the product entirely.
- No first-party signals at all: PredictLeads cannot see your website traffic, your product usage, or your email engagement.
- No scoring or prioritisation. Deciding that a funding round plus five engineering hires means an account is in market is logic you write and maintain.
- API-call metering means an inefficient integration can be expensive; polling broadly rather than using webhooks will burn credits for no additional signal.
- A small, long-standing company with a modest funding history, which is a stability argument in one direction and a support-bench argument in the other.

## Comparisons

- **PredictLeads vs Clay**: Clay is the orchestration layer that chains data providers into tables and workflows, and PredictLeads is one of the providers available inside it. They are complements rather than competitors. Buy Clay if you need a place to build the logic and blend many sources; call PredictLeads directly if you have your own engine and want the raw dated signals without paying for orchestration you will not use.
- **PredictLeads vs Trigify**: Trigify listens to public social and community activity and delivers person-level signals into Slack and your CRM for $40 a month flat. PredictLeads crawls company websites, news, and filings for company-level facts and delivers them through an API for $40 a month plus usage. The signal types barely overlap: Trigify tells you a person is talking, PredictLeads tells you a company is doing something. A technical team frequently runs both.
- **PredictLeads vs RB2B**: RB2B watches your own website and tells you which person showed up; PredictLeads watches the outside world and tells you what companies are doing. Neither can do the other's job. For a small United States focused team, RB2B is the faster path to a conversation, while PredictLeads is the better foundation if you are building triggered outbound against accounts that have never visited you.
- **PredictLeads vs Wappalyzer**: Both track technology installs, and PredictLeads covers more of them, over 54,000 technologies alongside seven other datasets, delivered through an API from $40 a month. Wappalyzer is a usable product rather than an endpoint, with a browser extension, a lead list builder, stack change alerts, and CRM connectors that a non-engineer can drive. Pick Wappalyzer if technographics are the whole motion and nobody on the team writes code; pick PredictLeads if you want several signal types from one API.
- **PredictLeads vs LeadMagic**: LeadMagic is a contact-first API that adds a job change detector and hiring and growth signals on top of email and mobile enrichment, from $49.99 a month with credits charged only on successful results. PredictLeads is signal-first with no contact data at all. The natural pairing is to use PredictLeads to decide which accounts matter and LeadMagic to find and verify the people to contact there, and teams building in Clay routinely call both.

## Implementation

- Setup time: An hour to a first API response, since the free tier requires no card and the endpoints are conventional REST. Getting to a production motion where signals create tasks is a genuine engineering project of one to three weeks, or an afternoon if you are consuming it inside Clay.
- Learning curve: Low for a developer and high for everyone else. The API is straightforward and the documentation covers the datasets clearly, but the hard part is not technical: deciding which combinations of signals actually mean something for your business is judgement work that the product deliberately leaves to you.
- Onboarding: Entirely self-serve from the free tier through every pay as you go band. Only the Enterprise flat-file plan requires a conversation. There is no implementation fee and no minimum term.
- Migration: Nothing to migrate in, because PredictLeads is a data source rather than a system of record. Historical access means you can backfill several years of job and news data at launch rather than starting from zero, which is a real advantage over first-party signal tools. On the way out, whatever you have written into your own database stays yours, since the data flowed into your systems rather than being held in a vendor application.

## Platform, API & security

- Platforms: REST API, Real-time webhooks, Flat files via AWS S3, Google Cloud Storage, and SFTP, MCP endpoint for AI agents
- API: The API is the product. Endpoints cover job openings, technology detections, news events, financing events, key customers, similar companies, products, and firmographics, metered per call with published volume-tiered rates and documented for direct integration.
- Compliance: Data is sourced from public company web pages, news, and filings rather than personal data brokers, European operating base subject to GDPR
- Data residency: The company is headquartered in Ljubljana, Slovenia. Flat file delivery targets are your own S3, Google Cloud Storage, or SFTP endpoints on the Enterprise plan.
- SSO: Not applicable in the usual sense, since access is by API key rather than an application login.
- Security notes: The dataset is company-level and built from publicly published pages, press coverage, and regulatory filings, which is a materially lower privacy risk profile than identity-graph tools that resolve named individuals. Access control is API key management on your side, so key rotation and scope discipline are your responsibility.

## Support

- Channels: Standard support on the entry band, Priority support above 5,000 calls, Round-the-clock support above 10,000 calls
- Documentation: API documentation covering every dataset, query patterns, webhook configuration, and flat file schemas.
- Community: No large user forum; the practical community is the technical go-to-market crowd building inside Clay and custom stacks.

## Company

- Founded: 2015
- Founders: Miha Stanovik, Roq Xever, Matic Perovsek
- Headquarters: Ljubljana, Slovenia
- Ownership: Venture-backed, lightly funded
- Employees: Small, including a data quality team of more than ten specialists
- Funding: Modest early funding including a Startupbootcamp investment in 2018 and a Y Combinator seed round in the summer 2019 batch; the company has grown on revenue rather than large rounds.

Funding history:

- Accelerator (2018): Small. Startupbootcamp early-stage investment.
- Seed (2019): Y Combinator standard investment. Participated in the Y Combinator summer 2019 batch.

Timeline:

- 2015: Founded in Ljubljana by Miha Stanovik, Roq Xever, and Matic Perovsek to turn public company web data into structured buying signals.
- 2016: Begins the continuous job opening and news crawling that produces the historical archive still available to customers today.
- 2019: Participates in the Y Combinator summer batch, formalising the API-first, data-provider positioning rather than building an application.
- 2024: Becomes a standard signal source inside Clay and other orchestration platforms, reaching go-to-market teams who never touch the API directly.
- 2026: Publishes coverage of more than 120 million companies across eight datasets and ships a Model Context Protocol endpoint so AI agents can query signals as a tool.

## Integrations

Clay, Dealroom, FactSet, REST API for custom integrations, Real-time webhooks, AWS S3, Google Cloud Storage, and SFTP flat file delivery, MCP for AI agents

## FAQ

### What is PredictLeads?

PredictLeads is a company signal data provider. It crawls company websites, more than twenty million news and press sources, and public filings to produce structured, dated records across job openings, technology detections, news events, funding, key customers, similar companies, products, and firmographics for more than 120 million companies, delivered by API, webhooks, flat files, and an MCP endpoint.

### How much does PredictLeads cost?

The first 100 API calls a month are free. Beyond that there is a $40 monthly minimum with credits at $0.04 each, falling to $0.02 above 5,000 calls, $0.01 above 10,000, $0.004 above 100,000, and $0.002 above 500,000. Enterprise, which adds flat file delivery via S3, Google Cloud Storage, or SFTP with unlimited company tracking and history from 2015, is quoted individually.

### Which signals does PredictLeads own?

Eight datasets, all company-level. Job openings from career pages, technology detections across more than 54,000 technologies, news events in 37 categories, financing events, key customer relationships extracted by logo recognition from case studies, similar-company lookalikes, product listings, and firmographics. It does not cover website visitors, product usage, or person-level events such as an individual champion changing jobs.

### How fresh is the data?

PredictLeads publishes its crawl cadences, which is rare. News sources are recrawled as often as every eight minutes, job data is refreshed at least every 36 hours, and company websites are checked on average daily, with crawling running continuously rather than in a nightly batch. Real-time webhooks push new signals as they are detected, so freshness reaches your systems rather than only the database.

### How does PredictLeads handle false positives?

Structurally rather than algorithmically. Every record carries the primary source URL that produced it, so any signal can be verified by clicking through to the page rather than trusted blindly. A quality team of more than ten specialists reviews thousands of records daily and automated anomaly detection runs continuously to catch failures such as a site redesign wiping a technology profile. What it does not do is judge whether a signal means intent, which remains your logic.

### Does PredictLeads score or route signals?

No. There is no scoring model, no routing engine, and no alerting. The API returns dated facts and webhooks push new ones as they appear. Deciding that a funding round plus a burst of engineering hires plus a competitor's technology being removed adds up to a buying window, and turning that conclusion into a task for a rep, is entirely your build. That is the trade you make for the price.

### Do I need a developer to use PredictLeads?

Either a developer or Clay. There is no application to log into, so the two realistic paths are calling the API from your own code or consuming PredictLeads inside Clay, Dealroom, or FactSet where the integration is already built. A non-technical sales team with neither will not get value from a direct subscription.

### What does it cost to act on a signal?

Fractions of a cent to a few cents, depending on volume. Because metering is per API call rather than per contact or per seat, the marginal cost of checking one more company is $0.04 at the entry band and $0.002 at the highest. The efficiency trap is polling: querying broadly on a schedule instead of subscribing to webhooks will multiply your call count without producing more signal.

### Can I backtest a signal before committing to it?

Yes, and this is one of the strongest reasons to use it. Job data goes back to 2016 and Enterprise access includes history from 2015, so you can test whether a proposed signal actually preceded your closed-won deals before building a motion around it. Most first-party signal tools cannot do this at all, because their data only starts when you install the tag.

### Who is behind PredictLeads?

It was founded in 2015 by Miha Stanovik, Roq Xever, and Matic Perovsek in Ljubljana, Slovenia, and went through Y Combinator in the summer of 2019. It has stayed small and lightly funded, growing as a data provider that other go-to-market vendors build on, with partner integrations including Clay, Dealroom, and FactSet.

## Editorial verdict

PredictLeads is what you buy when you have decided to build the signal engine yourself. The coverage is real, the crawl cadences are published rather than implied, and the primary source URL on every record is a quality property almost nobody else in this category offers. Pricing starts free, floors at $40 a month, and falls to fractions of a cent per call, which makes it dramatically cheaper than any packaged platform running on comparable data. The catch is unambiguous and you should not talk yourself past it: there is no dashboard, no alert, and no scoring, so the value only exists if you or your Clay table can turn API responses into something a person acts on. Technical go-to-market teams should treat this as a default component of their stack. Everyone else should buy a tool with an interface and let it worry about where the data came from.

---

Source: SaaSTracker (https://saastracker.org), an independent editorial project. This profile is compiled from public information, carries no peer reviews or paid placement, and was last reviewed 2026-08-22. Awards are judged on published criteria: https://saastracker.org/methodology