Benchmark · Intent Data

Intent Data Benchmark

Intent products are the hardest to compare because every vendor defines signals differently. This suite grades every provider against the same panel of disclosed purchases, then catalogs what each actually exposes.

Not measured yet. The first run of this suite will follow the method below.

Methodology

01
Planned test set
200 accounts are drawn from a panel of companies that disclosed a purchase decision in one of 8 software categories during the window, plus 200 matched controls that did not. Each provider is queried for intent on all 400 accounts across the 8 topics, on the same days.
02
Planned grading
Signal accuracy is the share of flagged accounts that actually bought; false-positive rate is the share of controls flagged. Predictive lift is the buy rate among flagged accounts divided by the control buy rate. Freshness is the lag between the first observed signal and the disclosed decision date. Granularity, resolution level, frequency, source diversity, customization and historical depth are catalogued deterministically from the API responses captured during the run.

What we measure

Every metric is defined once and applied the same way to every provider, so it compares across providers. The core metrics (coverage, accuracy, latency, pricing, and freshness and API quality where they apply) recur across suites; the rest are specific to this data type.

Common core

Accounts with signal

Share of buying accounts for which any signal was returned.

Signal accuracy

Flagged accounts that disclosed a purchase in the window.

Freshness

Median lag between first signal and the disclosed decision.

Latency p50

Median time to an answer per request, from the same place for every provider (for asynchronous APIs, until the job completes).

Price / signal

List price per useful result, on the central pack of the self-serve price ladder.

API quality

Rubric score from public material: docs, spec, auth, rate limits, errors, SDKs, status page, sandbox, changelog.

Specific to Intent Data

Predictive lift

Buy rate among flagged accounts vs. control group.

False-positive rate

Share of control accounts flagged with intent.

Topic granularity

Distinct topics available per account.

Resolution level

Account-level, person-level or both.

Signal frequency

How often signals refresh.

Source diversity

Distinct source families disclosed (web, content, hiring, tech…).

Geographic coverage

Countries with accounts carrying signals.

Custom topics

Buyer can define its own topics or keywords.

Historical depth

Oldest signal retrievable.

Questions about the Intent Data benchmark

What does the Intent Data benchmark measure?

It grades providers on 15 metrics: accounts with signal, signal accuracy, freshness, latency p50, price per signal, API quality, predictive lift, false-positive rate, topic granularity, resolution level, signal frequency, source diversity, geographic coverage, custom topics and historical depth. It has not been run yet, so no provider is ranked on it.

How will the Intent Data benchmark be measured?

Every provider will answer the same frozen, verified test set (200 buying accounts + 200 controls · 8 topics · set v2026-09-02.1) in the same window, graded by declared-purchase panel (objective). Each value will get a letter grade against the rest of the field, A for the top fifth down to D for the bottom fifth.