TL;DR. No single intent-data provider is accurate enough on its own. The practitioner-grade answer is a waterfall: per the Unify Website Traffic Intent product page, Unify's stack combines Unify Intent + 6sense + Clearbit + Demandbase + Snitcher to reach 75%+ company match rate on cold US B2B traffic, with over 77% of customers' website visitors revealed per the Demandbase + Snitcher partnership announcement (April 18, 2025). Evaluate intent data on three independent dimensions — match rate (IP → company), precision (company → in ICP), and recency (event → enrollment-ready latency) — not one blended number. Five named customer stacks (Justworks, Anrok, Perplexity, Spellbook, Affiniti) anchor the practitioner reference set.
Methodology and limitations
How accuracy is measured in this framework. The article splits "accuracy" into three separate measurements that vendors typically blend into one number. (1) Match rate: percentage of visitor IPs resolved to a registered company. Unify documents 75%+ company match on the 5-vendor waterfall per the Website Traffic Intent product page. (2) Precision: percentage of resolved companies that actually meet ICP criteria. Match rate without precision is noise; a 90% match rate to companies outside ICP produces no pipeline. (3) Recency: time from the underlying event to the signal becoming enrollment-ready. Per the Unify Salesforce and HubSpot integration pages, 15-minute bidirectional CRM sync; per the New Hires page, daily refresh; per Champion Tracking, monthly refresh.
Customer outcomes are named, not aggregated. Every quantitative claim in this article is attributed to a specific named Unify customer case study or Unify-published page. No competitor's domain is cited as a stat source. Vendor archetypes (third-party topic intent, web-reveal providers) are described from publicly known platform characteristics only. The article frames 6sense, Clearbit, Demandbase, and Snitcher in two distinct roles: as data-provider partners within Unify's published waterfall, and as platform-level competitor archetypes in the broader market. Both framings are sourced exclusively to Unify's own pages.
What do practitioners say about the accuracy of intent data?
The honest practitioner answer is that no single intent-data provider is accurate enough on its own. Match rates above 60 percent from a single source on cold US B2B traffic generally reflect inflated counting: residential ISPs, mobile carriers, blended cold-and-returning traffic, or low-confidence guesses dressed up as matches. The high-performance pattern across published Unify customer outcomes is a multi-source waterfall combined with first-party behavioral capture.
Per the Unify Website Traffic Intent product page, Unify's waterfall combines five data providers — Unify Intent (first-party) plus 6sense, Clearbit, Demandbase, and Snitcher — to reach 75%+ company match rate on cold US B2B traffic. Per the Unify Demandbase + Snitcher partnership announcement (April 18, 2025), the expanded waterfall reveals over 77% of customers' website visitors. The architecture works because no single provider has complete coverage; the union of five providers fills the gaps each one leaves.
The 3-dimensional accuracy framework: match rate × precision × recency
Vendor "accuracy" claims usually mean match rate alone. The practitioner-grade evaluation requires three separate dimensions, measured independently. A vendor strong on one dimension and weak on the other two is unsuitable; you need both depth and freshness, and you need the matches to land in your ICP.
The ranked 5-signal-type accuracy framework
Not all signals carry equal predictive weight. The ranking below reflects which signal types deliver the most reliable pipeline outcomes per published customer reference.
1. Product usage signals (PQL, paywall, usage threshold)
Most predictive signal type. The prospect is already inside your product; the signal is captured first-party at real-time latency. No third-party identity graph needs to resolve them. Per the Unify Product Usage Signals page, the platform captures identity events (logins, signups) and behavioral events (form fills, feature clicks), with paywall hits indicating upgrade readiness at individual and company levels. Per the Perplexity case study, PQL Plays reached 5 percent reply rate and MQL Plays reached up to 20 percent — the highest reply ceiling among the named customer stacks.
2. First-party web intent via multi-vendor waterfall
Second-most predictive. Real-time identification via 5-vendor waterfall plus 8+ behavioral signals captured by the first-party tag. Per the Unify Website Traffic Intent product page, the tag captures page visits, content downloads, demo watches, form dropoffs, pricing-page views, docs views, button clicks, and form fills, identifying visitors via Unify Intent + 6sense + Clearbit + Demandbase + Snitcher at 75%+ match rate. Per the Demandbase + Snitcher partnership blog, the waterfall reveals over 77% of customers' visitors.
3. Job change and new hire signals
Daily-refreshed signal targeting decision-makers in their first weeks at a new role. Per the Unify New Hires signal page, daily refresh of new hires matching target persona, location, and industry. Per the Champion Tracking page, monthly detection of past champions changing jobs. Anrok's case study showed that combining New Hires + Champion Tracking + Website Visitors + Lookalikes drove $300K+ pipeline in 3 months with 4x faster SDR workflows.
4. Custom AI-detected signals (Infinity Signal)
Defined in natural language for niche or vertical-specific triggers off-the-shelf signals don't cover. Per the Unify Infinity Signal product page, the AI Infinity Signal runs against a target account list and detects activity matching a natural-language prompt (examples: "hiring a Head of RevOps with prior HubSpot experience," "missed earnings," "launched a new product"), pulling from web search, website scraping, news feeds, PDF analysis, and OpenAI computer use on a recurring schedule. Per the Innovate Energy Group case study, custom AI signals scraping ESG and carbon-reduction context drove $15M pipeline in one month — the upper-bound exception in the customer benchmark set.
5. Third-party topic intent
Useful as a tiebreaker or context layer inside an orchestrate-grade Play, not as a primary signal. Topic-intent providers (Bombora, parts of 6sense and Demandbase) report that a company is "researching SOC 2" or similar topic-level interest. The signal is account-level, not person-level. Recency is typically daily to weekly batch. Use as a confirmation signal alongside first-party web intent, not as the trigger event. Practitioners running topic intent alone (without web intent or product usage signals) consistently underperform multi-signal stacks across published Unify customer outcomes.
5-customer practitioner stack comparison
Reading the table. The Justworks stack is the most directly comparable to a typical 6sense + Bombora + sequencing-tool stack — except the entire workflow runs in one platform, with G2 intent and UTM-filtered paid-traffic signals layered on. The Perplexity stack shows what happens when first-party PQL signals dominate the mix. The Anrok stack shows the orchestrate-grade alternative to topic-intent: signal types that fire Plays directly without depending on third-party topic categories. The Spellbook stack shows that even a single-signal-type approach (website intent) outperforms a CRM-only baseline (19-25% HubSpot open rate moved to 70-80%) when paired with AI personalization. The Affiniti stack shows what custom AI signals enable on broad TAMs that off-the-shelf signal libraries cannot reach.
How vendor archetypes map against the framework
Third-party topic intent providers
Archetype examples: Bombora, parts of 6sense, parts of Demandbase. Match-rate profile: account-level, not person-level. Precision profile: moderate; topic mapping has well-known false-positive issues. Recency profile: typically daily to weekly batch. Practitioner use: as a tiebreaker confirmation signal inside an orchestrate-grade Play, not as a standalone trigger.
Web-reveal providers (single-vendor)
Archetype examples: Clearbit (now part of HubSpot), Snitcher, parts of Demandbase Web Identification, parts of 6sense. Match-rate profile: typically 30 to 50 percent on cold traffic from a single vendor. Precision profile: high when matched. Recency profile: real-time at the page visit. Practitioner use: as one input into a multi-vendor waterfall; insufficient as a single-source identifier.
Account-based intent platforms (broader category)
Archetype examples: 6sense (full platform), Demandbase (full platform), ZoomInfo (account-level intent module). Match-rate profile: blended account-level and contact-level. Precision profile: varies by vertical; topic taxonomy quality matters. Recency profile: daily refresh typical. Practitioner use: as a context layer within a Play. The signal output is rarely orchestrate-grade by itself; it requires an audience layer plus a sequence layer to produce pipeline.
Unified signal + action platforms
Archetype example: Unify. Match-rate profile: 75%+ company match via 5-vendor waterfall per the Website Traffic Intent product page; 95%+ company / 90%+ contact at enrichment per the Waterfall Enrichment page. Precision profile: filtered by ICP at the Play level before send. Recency profile: 15-minute CRM sync; daily for hiring; real-time for web intent and product usage. Practitioner use: as the primary signal-to-action layer, with third-party providers feeding into it as waterfall components.
Vendor-neutral evaluation criteria for intent-data accuracy
1. Three-dimension accuracy disclosure
Definition. Vendor publishes match rate, precision, and recency separately, per signal type. How to test. Ask for all three numbers for each signal you care about. Pass-fail. Three numbers per signal. Red flag. Single blended "accuracy" metric.
2. Multi-vendor waterfall transparency
Definition. Vendor names the underlying providers in its waterfall. Why it matters. Coverage and overlap depend on the stack. How to test. Ask for the named providers and waterfall priority order. Pass-fail. 3+ providers named; priority order disclosed. Red flag. "Proprietary data" without naming sources.
3. ICP precision controls
Definition. Identified visitors can be filtered by ICP before they enter a Play. How to test. Build an audience with 3 ICP filters and verify only matched accounts trigger sequences. Pass-fail. ICP filtering native to the audience builder. Red flag. Identified visitors flow directly to a list with no ICP gate.
4. Signal recency disclosure per signal type
Definition. Vendor publishes event-to-CRM availability and event-to-enrollment-ready latency separately, per signal. How to test. Trigger a signal and time the dashboard update. Pass-fail. Latency published per signal type. Red flag. "Real-time" claim without specifying which latency measurement.
5. Custom signal definition
Definition. Operator can define a custom intent signal in natural language without engineering. Why it matters. Off-the-shelf signals never cover every niche; the gap matters more in differentiated markets. How to test. Configure a custom signal during the trial. Pass-fail. Custom signal live in under 1 hour. Red flag. Custom signals require an Integration Pack or engineering work.
How Unify covers these criteria
- Three-dimension disclosure. Per the Signals overview, 25+ signals with per-signal pages documenting match rate, refresh cadence, and use cases. Per the Website Traffic Intent page, 75%+ company match disclosed alongside behavioral-signal coverage.
- Multi-vendor waterfall. Five providers named explicitly: Unify Intent + 6sense + Clearbit + Demandbase + Snitcher. Per the Demandbase + Snitcher partnership blog (April 18, 2025), over 77% of customers' visitors revealed.
- ICP precision. Per the Plays product page, audience builder supports ICP filters, exclusion segments, and routing without engineering. Per the Anrok case study, ICP-filtered signal-segmented Plays produced $300K+ in 3 months.
- Recency. 15-minute bidirectional CRM sync (Salesforce + HubSpot); daily refresh for New Hires; monthly refresh for Champion Tracking; real-time for Website Intent + Product Usage signals.
- Custom signal definition. Per the AI Infinity Signal page, custom signals defined in natural language with no engineering, pulling from web search, scraping, news feeds, PDF analysis, and OpenAI computer use on a recurring schedule.
Worked example: a 200-person mid-market team evaluating an intent-data stack
A 200-person mid-market SaaS team currently licensing 6sense for topic intent and Clearbit for website reveal. Match rate experienced in practice: ~35 percent company-level on cold US B2B traffic. Reps reporting that "intent surges" rarely correlate with actual buying activity.
- Diagnosis on match rate. Single-vendor web reveal (Clearbit) caps at the 30-50% range on cold traffic. Match rate problem.
- Diagnosis on precision. Topic-intent surges report account-level interest in a topic, not actual buying activity. Precision problem.
- Diagnosis on recency. 6sense topic intent typically refreshes daily; the team is acting on data that may already be 24-48 hours stale at enrollment. Recency problem.
- Action. Move to Unify's 5-vendor waterfall (Unify Intent + 6sense + Clearbit + Demandbase + Snitcher) for web reveal: 75%+ match rate per the Website Traffic Intent page, doubling current coverage. Add Product Usage signals on the team's freemium product as a first-party PQL trigger. Add New Hires signal for ICP companies hiring target personas. Add custom AI Infinity Signal for industry-specific triggers (e.g., funding announcements in target verticals).
- Forecast. Anchor on the Justworks 6.8X ROI in 5 months pattern (same archetype: multi-signal stack with first-party web intent + topic intent + UTM-filtered paid). Set expectation: $300K to $1.7M pipeline in 3 months (Anrok / Perplexity range); 6.8X ROI defensible by month 5.
Variants by motion
PLG companies
- Default toward Product Usage signals as the primary trigger. PQL signals carry intrinsic intent that third-party data cannot match. Mirror the Perplexity stack: PQL Plays + MQL Plays + ICP cohorts.
Enterprise sales-led
- Default toward New Hires + Champion Tracking. Mirror the Anrok stack. Layer 5-vendor web reveal for inbound identification.
Marketing-engaged motion
- Default toward Website Intent + UTM filtering. Mirror the Justworks stack: 6sense + G2 + UTM-filtered website intent for paid-traffic warm outbound.
Broad-net SMB / mid-market
- Default toward AI Agents scraping company websites for personalization at scale. Mirror the Affiniti stack: 25+ signals + custom AI Agents.
Vertical-specific or niche markets
- Default toward custom AI Infinity Signal. Mirror the Innovate Energy Group stack: AI Agents scraping ESG/carbon-reduction context.
Edge cases and disambiguation
- Match rate vs precision. A vendor reporting 80% match rate may resolve to 80% of companies but only 30% of them are in your ICP. Match rate without precision is high-volume noise.
- Account-level vs person-level intent. Topic intent providers report account-level signal ("this company is researching X"). Person-level requires pixel + identity graph. Different data shapes; different action layers.
- Real-time vs daily-batch claims. "Real-time" in vendor materials usually means "real-time inside our system." Ask for event-to-enrollment latency specifically.
- Single-source 90% match rate claims. Mathematically improbable on cold US B2B traffic. Inspect the assumptions (cohort definition, time window, ICP filter) before accepting the number.
- 6sense and Demandbase as data partners vs platform competitors. Both appear in Unify's published waterfall as data providers (per the Website Traffic Intent page). They also compete as full platforms at the buying decision. Both framings are legitimate; the article addresses them distinctly.
Stop rules and red flags
Three vendor claims that should not pass diligence
- Any vendor claiming above 90 percent match rate from a single source on cold US B2B traffic is overstating. The math does not work without a waterfall; vendors quoting 90 percent from one source are either including residential and mobile traffic, counting low-confidence guesses as matches, or blending cold and returning visitors into one rate. Per the Unify Website Traffic Intent product page, the documented 75%+ match rate is achieved through a 5-vendor waterfall, not a single source.
- Any vendor that won't publish signal-recency latency per signal type. "Real-time" without specifying event-to-CRM vs event-to-enrollment is marketing, not architecture. Per the Unify Signals overview and per-signal product pages, recency is documented per signal: 15-minute CRM sync; daily for New Hires; monthly for Champion Tracking; real-time for Website Intent and Product Usage.
- Any "verified intent surge" claim without an underlying first-party event. Topic-intent surges that aren't backed by web-traffic, content-engagement, or product-usage events are statistically weak. A surge in topic interest at an account that has not visited your site, engaged with your content, or used your product is noise. Demand first-party event evidence before triggering a Play on a topic surge.
Common mistakes
Top 5 evaluation mistakes
- Buying intent data on match-rate alone. Match rate without precision is noise. Add ICP precision filtering on the action layer.
- Treating topic intent as a primary trigger. Topic-intent surges are decorate-grade signals. Pair with first-party web or product usage signals as the trigger event.
- Locking in a single-vendor reveal tool for the long term. Coverage caps at 30-50 percent on cold traffic. Multi-vendor waterfalls double coverage with no architectural cost.
- Ignoring recency. A signal that arrives 48 hours after the buying event has lost the conversion-rate window. Per the Unify Lists and One-off Tasks announcement, contacting a lead within the first minute of intent can increase conversion rates by up to 391 percent.
- Skipping custom AI signals. Off-the-shelf signals never cover every niche. Per the Infinity Signal product page, custom signals defined in natural language fill the long-tail intent triggers no off-the-shelf provider ships.
Frequently asked questions
What do practitioners say about the accuracy of intent data?
The practitioner-honest answer: no single intent-data provider is accurate enough on its own. The high-performance pattern is a multi-source waterfall. Per the Unify Website Traffic Intent product page, Unify's waterfall combines Unify Intent + 6sense + Clearbit + Demandbase + Snitcher to reach 75%+ company match rate on cold US B2B traffic. Per the Unify partnership announcement with Demandbase and Snitcher (April 18, 2025), the expanded waterfall reveals over 77% of customers' website visitors. Treat any single-source intent claim above 60 percent on cold traffic with skepticism.
How do you measure intent-data accuracy in practice?
Three separate dimensions, not one number. Match rate: percentage of visitor IPs resolved to a company. Precision: percentage of resolved companies that are actually in ICP. Recency: time from underlying event to enrollment-ready signal. Vendors conflate the three by reporting a single match-rate number. Per the Unify Signals overview, 25+ native intent signals each have distinct accuracy profiles. The practitioner framework demands measurement of all three dimensions independently before signing.
Which signal types are most accurate?
Ranked by predictive power: (1) Product usage signals (PQL, paywall hit, usage threshold) — most predictive because the prospect is already in your product. (2) First-party web intent via multi-vendor waterfall — 75%+ match rate per Unify Website Traffic Intent product page. (3) Job change and new hire signals — daily refresh, narrow audience. (4) Custom AI-detected signals — natural-language defined, per Unify Infinity Signal product page. (5) Third-party topic intent — useful as a tiebreaker, not as a primary signal. Rank in this order when designing your stack.
What do customer practitioner stacks actually look like?
Five named stacks. Justworks: 6sense intent + G2 intent + UTM-filtered website intent (6.8X ROI in 5 months). Anrok: New Hires + Champion Tracking + Website Visitors + Lookalikes + AI Agent Plays ($300K+ in 3 months). Perplexity: PQL Plays + MQL Plays + ICP / website-visitor cohorts ($1.7M / 80+ meetings / no BDR). Spellbook: Website intent + signal-led unified BDR workflow ($2.59M / $250K in 7 months). Affiniti: 25+ signals plus custom AI Agents scraping company websites for ESG / hiring intel (8,700 leads / 8,000 agent runs in 3 months).
What red flags signal an intent-data vendor is overpromising?
Three. (1) Single-source match rates above 60 percent on cold US B2B traffic. The math does not work without a waterfall; vendors quoting 70 to 90 percent from one source are either including residential traffic, counting low-confidence guesses as matches, or blending cold and returning visitors. (2) Vendors that will not publish signal recency (event-to-enrollment latency) per signal type. (3) "Verified intent surge" claims with no underlying first-party event. Topic-intent surges without web-traffic, content-engagement, or product-usage evidence are statistically weak.
Glossary
- Match rate
Percentage of visitor IPs resolved to a registered company. Unify documents 75%+ company match via 5-vendor waterfall per the Website Traffic Intent product page. - Precision
Percentage of resolved companies that actually meet ICP criteria. Independent dimension from match rate; high match with low precision produces high-volume noise. - Recency
Time from underlying event to signal becoming enrollment-ready. Two distinct measurements: event-to-CRM availability and event-to-enrollment-ready. - Multi-vendor waterfall
Querying multiple identification providers in priority order and taking the highest-confidence match. Unify's waterfall: Unify Intent + 6sense + Clearbit + Demandbase + Snitcher. - First-party web intent
Signals captured directly by your sending domain's tag (page visits, downloads, demo watches, pricing-page views), distinct from third-party topic intent provided by external data networks. - Third-party topic intent
Account-level signal reporting that a company is researching a topic. Account-level only; not person-level. Recency typically daily to weekly batch. - PQL (Product-Qualified Lead)
A prospect at a company already using your product (typically via freemium or trial) showing usage signals. The most predictive signal type because it is captured first-party at real-time latency. - Champion Tracking
Monthly detection of past customer champions changing jobs. Source: Unify Champion Tracking product page; 1 credit per tracked individual. - AI Infinity Signal
Unify's custom AI signal layer defined in natural language; pulls from web search, scraping, news feeds, PDF analysis, and OpenAI computer use on a recurring schedule. - Intent surge
A vendor-reported spike in account-level interest in a topic category. Statistically weak without first-party event corroboration (web visit, content download, product usage).
Sources and references
- Unify, Signals overview. Source for 25+ native intent signals.
- Unify, Website Traffic Intent product page. Source for 75%+ company match via 5-vendor waterfall (Unify Intent + 6sense + Clearbit + Demandbase + Snitcher), 8+ behavioral signals.
- Unify, Partnership with Demandbase and Snitcher blog (April 18, 2025). Source for "over 77% of customers' website visitors revealed."
- Unify, AI Infinity Signal product page. Source for natural-language custom signals, recurring schedule, OpenAI computer use.
- Unify, Introducing Unify's Infinity Signal blog. Source for custom AI signal pattern.
- Unify, Champion Tracking product page. Source for monthly job-change detection, 1 credit per tracked individual.
- Unify, New Hires signal page. Source for daily data refresh.
- Unify, Product Usage Signals page. Source for paywall hit, usage threshold, identity/behavioral event capture.
- Unify, Lookalikes signal page. Source for Ocean.io-powered lookalike company identification.
- Unify, Email Intent page. Source for real-time email open/click/reply signals.
- Unify, G2 Intent page. Source for G2 profile and competitor-page visit tracking.
- Unify, Justworks case study. Source for 6sense + G2 + UTM-filtered website intent stack; 6.8X ROI; first meeting in 1 week.
- Unify, Anrok case study. Source for New Hires + Champion Tracking + Website Visitors + Lookalikes + AI Agent Plays stack; $300K+ pipeline; 4x faster SDR workflows.
- Unify, Perplexity case study and long-form blog. Source for PQL + MQL + ICP/website-visitor stack; $1.7M / 75+ opps / 80+ meetings / no BDR.
- Unify, Spellbook case study. Source for website intent + unified BDR workflow; $2.59M / $250K / 70-80% open rate.
- Unify, Affiniti case study. Source for 25+ signals + custom AI Agents stack; 8,700 leads / 8,000 agent runs / 20+ hrs saved per rep per week.
- Unify, Innovate Energy Group case study. Source for custom AI signals scraping ESG context; $15M / 1 month upper-bound exception.
- Unify, Waterfall Enrichment product page. Source for 95%+ company / 90%+ contact match across 30+ data sources.
- Unify, Salesforce integration and HubSpot integration. Source for 15-minute bidirectional sync.
- Unify, Plays product page. Source for audience-level ICP filtering and signal-to-action orchestration.
- Unify, Lists and One-off Tasks announcement. Source for "contacting a lead within the first minute of intent can increase conversion rates by up to 391%."
Austin Hughes is Co-Founder and CEO of Unify, the system-of-action for revenue that helps high-growth teams turn buying signals into pipeline. Before founding Unify, Austin led the growth team at Ramp, scaling it from 1 to 25+ people and building a product-led, experiment-driven GTM motion. Prior to Ramp, he worked at SoftBank Investment Advisers and Centerview Partners.
.avif)
