B2B Intent Data: What the Evidence Actually Supports

Most published writing about B2B intent data repeats numbers that no named source actually reports. We know, because an earlier version of this page did it too. We rebuilt it around claims that survive checking, and the result is shorter than what we started with. That is the finding.

What follows is what we can stand behind, with the source and the date attached to each figure, followed by an explicit list of the claims we removed.

The cookieless future did not arrive, and a lot of intent strategy is still built on it

This is the correction that matters most, because it is load-bearing for budget decisions being made right now.

Google reversed its plan to deprecate third-party cookies in Chrome in April 2025. It did not accelerate, delay, or soften the timeline. It abandoned the prompt-based phase-out and kept existing cookie controls in place. In October 2025 Google went further and deprecated the remaining Privacy Sandbox APIs, including Topics and Protected Audience, the replacements it had spent roughly six years building.

Third-party cookies remain enabled by default in Chrome today.

What is true, and what a lot of vendor material blurs into the same story, is that Safari and Firefox have blocked third-party cookies by default for years and have not changed course. So the constraint is real, it is just older and narrower than the "cookieless future" framing suggests, and it was never going to be resolved by a Chrome deadline.

What this means for an intent program. If your provider justified a price increase, an architecture migration, or a move to declared-data partnerships on the basis of imminent Chrome deprecation, that premise expired in April 2025. Ask what changed in their roadmap after the reversal. Server-side collection and first-party signal capture are still defensible on their own merits, including consent enforcement and measurement durability. They are not defensible on a deadline that no longer exists.

Mature account-based programs convert at roughly 1.6 times the rate of less mature ones

Demandbase's State of ABM 2026 report, drawn from 1,452 companies, reports a median marketing-qualified-account to pipeline conversion rate of 22.33 percent among mature ABM programs against 14.19 percent among less mature ones.

The same report attaches numbers to three other practices:

How to read this. These are correlations across a large sample, not a controlled test, and Demandbase sells the category it is measuring. The buying-group findings are the most useful because they describe a change in unit of analysis rather than a change in spend. Moving from lead-level to group-level tracking is an operational decision you can make without buying anything.

One documented case for first-party review intent

How to read this. This is a vendor-published case study about a single customer, and that customer is itself an intent data provider. It is evidence that first-party review-site signal can outperform a baseline in at least one documented instance. It is not a category benchmark, and anyone presenting it as one is overreaching.

The broader claim often made here, that review-site intent converts at some multiple of third-party topic intent, is not something we found support for. G2 has announced expanded coverage delivering "up to 2x more signals," which is a statement about signal volume, not conversion rate. Those two claims get conflated constantly, including in the earlier version of this page.

What we removed, and why

The previous version of this brief listed fifteen trends. Each carried an "Evidence" line attributing a statistic to a named research firm. We checked them. Most did not survive.

We removed claims attributed to Demandbase about procurement leading intent renewal reviews, identity match-rate variance across providers, written marketing-to-sales service level agreements in top versus bottom quartile programs, demand-state routing adoption rates, and the share of top-quartile programs publishing signal-to-pipeline benchmarks. Demandbase's published ABM benchmark reporting contains none of these.

We removed a claim attributed to Cognism quantifying the share of sales development capacity wasted on unrouted signal. We could not find that benchmark in Cognism's published research.

We removed a claim that review-site activity converts to opportunity at roughly twice the rate of third-party topic surges, attributed to G2. G2's published conversion data is the ZoomInfo case study above.

We removed a claim about category relevance decay windows attributed to Bombora methodology documentation, and a claim about signal overlap exceeding half of common categories attributed to Bombora and ZoomInfo jointly.

Several of these appeared with the number missing entirely, rendered on the live page as a placeholder marker. That is how we found them.

What to actually do

Ask every provider for the primary source behind their headline number. Not a blog post that cites a blog post. The report, the sample size, the date. A provider that cannot produce it within a day is quoting the same secondhand material everyone else is.

Benchmark against your own pipeline, not the category. The most reliable number about your intent program is the conversion rate from signal to qualified opportunity in your own data over the last four quarters. No published benchmark is worth more than that, and most published benchmarks are worth considerably less.

Separate the volume claim from the conversion claim. More signals and better signals are different products. Vendors price them the same way and describe them interchangeably.

Re-examine anything justified by cookie deprecation. The Chrome timeline is gone. The privacy constraints that predate it are not, and they are the honest reason to invest in first-party capture.

A note on method

We are a B2B marketing agency writing about a category we sell into, so treat this as an interested party reporting on itself. That is precisely why every figure above carries its source, its date, and its sample where one exists, and why the removals are listed rather than quietly deleted. If a claim here cannot be traced to a named primary source, it should not be here, and we would like to know about it.