See how ad traffic actually engages
Numbat scores every session on your landing pages by scroll depth, time on page and repeat visits. Those scores go back to your ad platforms as conversion signals, and every paid channel lands in one dashboard.
| Channel | Sessions | Engaged | Signals |
|---|---|---|---|
| Google Ads | 5,092 | 1,821 | 442 |
| Meta | 6,244 | 1,976 | 517 |
| 548 | 262 | 71 | |
| 1,190 | 271 | 68 |
Clicks say nothing about engagement
Depth is invisible
On long B2B cycles, a visit that bounces in four seconds and one that reads three pages look identical to your platforms. Both count as a click.
Signals arrive late and incomplete
Conversion events reach platforms days later and in fragments. Their optimization ends up learning from a thin, distorted sample of the visit.
Every channel reports in its own silo
Four dashboards, four attribution models, and no shared definition of a good visit across any of them.
How it works
Install a lightweight script
One first-party script on your landing pages. It collects engagement and device-context signals only, never form contents, keystrokes, or anything typed.
Every session is measured and scored
Scroll depth, time on page, and repeat visits are scored into a session verdict. The same scoring is what separates real visitors from invalid traffic.
Clean signals flow back
Verdicts return to your ad platforms as conversion signals their optimization can use, and every channel lands in one dashboard.
Every paid channel in one dashboard
| Channel | Campaign | Sessions | Engaged | Eng. rate | Median dwell | Signals sent |
|---|---|---|---|---|---|---|
| Google Ads | Search intent | 12,480 | 4,993 | 40.0% | 1:46 | 1,214 |
| Google Ads | Performance Max | 9,362 | 2,340 | 25.0% | 0:58 | 612 |
| Meta | Prospecting B2B | 18,905 | 5,294 | 28.0% | 1:12 | 1,371 |
| Meta | Retargeting 30d | 6,114 | 2,873 | 47.0% | 2:04 | 744 |
| ABM tier 1 | 2,238 | 1,052 | 47.0% | 2:31 | 296 | |
| Interest devtools | 4,870 | 1,120 | 23.0% | 0:49 | 292 |
Deep, active, skim, invalid. Every campaign broken down by what sessions did on the page rather than what they clicked.
See exactly how many enhanced conversion signals went back to each channel, and what they were based on.
The same scoring across Google, Meta, LinkedIn, and Reddit, so the channels can be compared.
Built for the platforms you already use
Common questions
What engagement signals does Numbat measure?
Per-session scroll depth, time on page, repeat visits, and device-context signals. Sessions are scored into one of four verdicts (deep, active, skim, invalid), which drive both the dashboard and the signals sent back to platforms.
How does signal emission work?
When a session crosses your engagement thresholds, Numbat returns a conversion signal to the platform that delivered the visit, through that platform's standard conversion APIs. Platforms optimize on engagement instead of clicks alone.
What does the script collect?
Engagement and device-context signals only. It does not collect form contents, keystrokes, or anything typed on the page.
How does Numbat approach GDPR?
The measurement model is built on engagement and device context rather than personal identifiers. What this marketing site itself collects is a separate question, and it is answered in the privacy policy, including how to reach us at privacy@numbat.io.
What does early access mean?
Numbat is pre-release. Early access customers are onboarded in small batches, work directly with us, and shape the roadmap. No SLA applies during early access; the terms cover the details.
Does Numbat detect invalid traffic?
Yes, as a by-product of the scoring. Invalid sessions separate out from real visitors, and they are excluded from the signals platforms receive.
Request early access
Tell us where to reach you and we'll be in touch as onboarding batches open.