27 Intent Signal APIs, Grouped by How You Actually Connect Them

27 Intent Signal APIs, Grouped by How You Actually Connect Them

Stop reaching out to companies that are not buying.

That sentence is easy to agree with and hard to operationalise, because almost every list of intent data providers is organised by vendor category. Categories are useless when you are the one building. What you need to know first is how the thing connects, because that determines how long the integration takes, how fresh the data is, and whether you can act on it the same day.

So here are 27 providers grouped by connection method, which is how I actually think about them when wiring a client's stack.

MCP servers: native tools an agent can call

These expose a Model Context Protocol server, so Claude Code and similar agents call them like built-in tools with no wrapper code.

1. PredictLeads. Hiring surges, funding rounds, tech installs, news.
2. LeadMagic. Website visitor identification and B2B enrichment, pay-per-use.
3. Apify. Scrapes any website into structured data. LinkedIn, job boards, review sites.
4. Firecrawl. Crawls and extracts clean markdown from any page, built for model pipelines.
5. Smartlead. Cold email campaign management from the terminal.

What these are good for: exploratory work where you do not know in advance which call you need, because an agent can compose them mid-task. Ask it to find companies hiring for a role, enrich them, and check their tech stack, and it chains three of these without you writing an integration. The caveat: MCP server availability changes often, so check before designing around one.

REST APIs: one curl command away

These take a single HTTP call to connect and slot into any scheduled job.

6. BuiltWith. Detects what tech stack a company runs.
7. Similarweb. Traffic and competitor analytics. Owler covers competitor alerts.
8. Crunchbase. Funding rounds, investors, leadership changes.
9. Bombora. Topic-level content consumption signals across 5,000+ B2B publishers, per the vendor.
10. ZoomInfo. Over a billion buying signals monthly across 500M contacts, per the vendor.
11. 6sense. Predictive buying stage detection.
12. G2. Review site buying behaviour and category research.
13. Demandbase. Account-level intent plus ABM orchestration.
14. Cognism. Phone-verified contacts with GDPR compliance.
15. Apollo.io. 275M+ contacts with email and phone data, per the vendor.
16. Hunter.io. Email finder and verification API.

What these are good for: batch scoring. You run them on a schedule over a defined account list, write the result to a column, and sort. This is the backbone of most account prioritisation, and it is where the majority of intent budget goes.

The latency cost, which nobody mentions: a nightly pull is by definition up to a day stale. For funding rounds and tech installs that is irrelevant, because the signal stays valid for weeks. For a category-research spike on G2 or a topic surge on Bombora, a day is a meaningful share of the window in which the signal means anything. Match pull frequency to how fast the signal decays, not to your cron schedule.

Coverage differs sharply by geography and company-size band, and the marketing does not tell you where. We break down the biggest three in Clay vs Apollo vs ZoomInfo, and the wider field in data enrichment tools for SDR teams.

Webhooks and triggers: the signal comes to you

These push an event as it happens rather than waiting for you to ask.

17. RB2B. Person-level website visitor identification.
18. Dealfront. European-focused visitor identification and prospecting.
19. Common Room. Unified community, product, and social signals.
20. Crossbeam. Partner overlap and co-selling data.
21. Trigify.io. Real-time LinkedIn engagement monitoring.
22. PhantomBuster. LinkedIn data extraction and engagement automation.
23. Expandi. LinkedIn outreach with engagement tracking.

What these are good for: anything where the value collapses with time. Someone reading your pricing page is worth contacting today and close to worthless in a fortnight. This is the group that justifies building real-time routing, and it is the only group where the latency argument genuinely earns the extra engineering.

The trade is volume and noise. A webhook fires on every match, including your own team, your existing customers, and bots. Suppression lists and a same-day routing rule are not optional here, they are what makes the group usable at all. Background on the mechanics is in website visitor identification and our RB2B review and alternatives.

Orchestration: stacking multiple signals

24. Clay. 100+ enrichment providers in one workflow.
25. HubSpot, Salesforce, Attio. CRM sync for signal-to-pipeline tracking.
26. Instantly.ai, lemlist. Signal-triggered outbound execution.
27. Perplexity. Research to enrich any signal with context.

What these are good for: turning several weak signals into one decision. A single signal is almost never enough to justify contact. Hiring for a role plus a competitor tech install plus a pricing page visit is a different proposition than any one alone, and orchestration is where that combination becomes a score you can sort on.

The cost trap

Intent data is the easiest line item in a GTM budget to overspend on, and the mechanism is always the same: the data is bought before the play exists.

The sequence that wastes money: buy an intent product, get a dashboard of surging accounts, look at it weekly, feel informed, book nothing. The sequence that works: define one play with a written trigger and a written first message, run it manually against a handful of accounts, then buy the API that automates the trigger you have already proven.

A signal you cannot act on within its useful window is a subscription to interesting information. Before adding a provider I want written answers to three questions: what event does this detect, what do we send within how many hours of it, and who owns the reply. If any is blank, the provider waits a quarter.

Intent narrows a list you have already defined and cannot rescue one defined badly, which is why the ICP work comes before any of this. For category background, see the B2B intent data guide.

How to choose between the four groups

The rule I use is decay speed:

• Signal stays valid for weeks (funding, hiring, tech install): a REST pull on a schedule is fine and cheaper.
• Signal decays in days (category research, topic surge): pull more often than nightly, or accept you are late.
• Signal decays in hours (site visit, engagement): webhook, or do not bother collecting it.
• You do not know which signal matters yet: MCP tools and manual exploration until you do.

Ben's parallel piece on which APIs survive automation is the companion to this one: the 18 APIs that actually help you sign clients, plus the fuller 22-API outbound stack.

Frequently Asked Questions

What is the difference between an intent data API and a webhook provider?

An API is pulled on your schedule, so the data is as fresh as your last request, and a nightly job is by definition up to a day stale. A webhook pushes the event as it happens. Use pulls for slow-decaying signals like funding and hiring, and webhooks for signals that lose value within hours, like a pricing page visit.

Which intent signals are worth paying for first?

The ones that precede a purchase in your specific market and that you can act on the same week. Hiring surges and funding rounds are unambiguous events with clear timing. Category research on a review site is high quality because the person is already comparing vendors. Start with one, prove a play, then expand.

How much intent data does a small team actually need?

Usually one source, applied to an account list that is already well defined. The common failure is buying several providers before a single play exists that acts on a signal within its useful window. If nobody owns the reply and there is no written trigger, more data will not produce more meetings.

Can Claude Code call these intent APIs directly?

The five MCP-native tools listed here connect as native agent tools with no wrapper code. The REST group takes roughly one HTTP call each to integrate. MCP server availability changes frequently across vendors, so verify current support rather than assuming it from a list.

What does person-level visitor identification add over account-level intent?

Account-level intent tells you a company is researching a topic. Person-level identification names the individual on your site, which turns a prioritisation signal into a contactable one. It also carries more noise, since it fires on your own team, existing customers, and bots, so suppression lists matter more here than anywhere else.

RevenueFlow builds AI-native pipeline systems and you pay per qualified meeting, not a retainer. No paying for activity. You only pay when we book you a qualified sales meeting. See if you qualify.