Starter
Measure the basics. Prove the channel is working.
$399/month
- 50 AI decisions / mo
- 2 seats
- 90 days of history
- Core attribution: first, last, linear, time decay
Pricing
The Causal AI Decision Engine
for Commerce
Triangulate multi-touch attribution, incrementality testing, and Veridict — our multi-model deliberation engine. The same methodology family as Dreamdata and Google DDA, plus Phase 10 causal calibration.
Save 17 %
Most Popular
Growth
Attribute causally. Unlock Veridict-powered decisions.
$999/month
7-day free trial · no credit card
- 300 AI decisions / mo
- 8 seats
- 395 days of history
- Markov attribution with causal calibration
- Phase 10 geo-experiments (1/quarter)
- Veridict — multi-model deliberation engine
Scale
Run the full causal stack. Everything Veridict sees, you see.
$2,499/month
- 1,000 AI decisions / mo
- Unlimited seats
- Unlimited history
- All attribution models including Shapley / DDA
- Unlimited Phase 10 experiments
- Audit log, priority support
Enterprise
SSO, SOC 2, data residency. On your procurement calendar.
Custom
- 10K+ fair-use decisions
- SSO + custom RBAC
- SOC 2 Type II, GDPR DPA
- Data residency (US / EU)
- Dedicated solutions engineer
Compare plans, feature by feature
Six groups, everything you'll actually care about. Every row maps to a backend capability — nothing here is aspirational.
| Feature | Starter | Growth | Scale | Enterprise |
|---|---|---|---|---|
| AI Decisions & Veridict | ||||
| Monthly decision cap | 50 | 300 | 1,000 | 10K+ fair use |
| Monthly AI budget | $10 | $50 | $200 | Custom |
| Overage (per 100 decisions) | $12 | $10 | $8 | Negotiated |
| Veridict multi-model deliberation | Not included | Included | Included | Included |
| Signal lifecycle (48h half-life) | Included | Included | Included | Included |
| Attribution | ||||
| First-touch | Included | Included | Included | Included |
| Last-touch | Included | Included | Included | Included |
| Linear | Included | Included | Included | Included |
| Time-decay (7-day half-life) | Included | Included | Included | Included |
| Markov (Anderl 2016, IJRM) | Not included | Included | Included | Included |
| Markov + causal calibration (Gordon 2023) | Not included | Not included | Included | Included |
| Connectors | ||||
| Shopify | Included | Included | Included | Included |
| Meta Ads | Included | Included | Included | Included |
| Google Ads | Included | Included | Included | Included |
| Google Analytics 4 | Not included | Included | Included | Included |
| Google Search Console | Not included | Included | Included | Included |
| Causal & Incrementality | ||||
| Geo-experiments per quarter | 0 | 1 | Unlimited | Unlimited |
| DiD + CUPED | Not included | Included | Included | Included |
| Synthetic DiD (Arkhangelsky 2021) | Not included | Not included | Included | Included |
| Wild Cluster Bootstrap | Not included | Not included | Included | Included |
| Trimmed Match | Not included | Not included | Included | Included |
| Parallel-trends diagnostic (PASS/WARN/FAIL) | Not included | Included | Included | Included |
| Team & History | ||||
| Seats included | 2 | 8 | Unlimited | Unlimited |
| History retained | 90 days | 395 days | Unlimited | Unlimited |
| Audit log | Not included | Not included | Included | Included |
| Compliance & Security | ||||
| SOC 2 Type II | Not included | Not included | Not included | Included |
| SSO (SAML + OIDC) | Not included | Not included | Not included | Included |
| Data residency (US / EU) | Not included | Not included | Not included | Included |
Methodology
Evidence-first. Paper-cited. Causally calibrated.
Every claim the product makes is anchored to an experiment or a published paper. If we couldn't cite it, we wouldn't ship it.
Anderl et al. 2016
Mapping the customer journey
International Journal of Research in Marketing 33(3)
Markov chain attribution with removal-effect crediting — the foundation of our multi-touch model.
Gordon, Moakler & Zettelmeyer 2023
Close enough? A large-scale exploration of non-experimental approaches
Marketing Science
MTA + incrementality triangulation — the motivation for our causal-calibrated attribution layer.
Arkhangelsky et al. 2021
Synthetic Difference-in-Differences
American Economic Review 111(12)
SDiD estimator used in Phase 10 geo-experiments when parallel-trends assumption is at risk.
AttributionOps 2026
Deliberation engine (internal)
Two independent analysts → cross-critique → neutral judge → grounding validation. Named Veridict.
Enterprise
Procurement-ready. SOC 2 Type II. Data residency.
Custom volume tiers, SSO, dedicated solutions engineer, and a live MSA + DPA within your procurement cycle. Spend bands from $10M ad spend upward.
Frequently asked
Ready to know which channel is actually working?
Start a 7-day free trial on Growth. No credit card. Veridict and Phase 10 causal calibration included.