The semantic layer that makes the rest real.
Truefold's semantic layer is the single source of truth for every metric, dimension, and definition your team uses. Connectors feed it, the auditor keeps it correct, AI chat queries it. Without this foundation, the rest is just chat with a database.
True metrics, not platform claims.
Every metric (true_revenue, true_roas, true_cac, true_ltv) is defined once, computed across every source you connect, refund-adjusted, and reusable everywhere. The numbers you can defend in any meeting.
- true_revenue: reconciled across Shopify, Meta Pixel, GA4 (net of refunds)
- true_roas: pixel over-attribution caught, real return per ad dollar
- true_cac: total ad spend across all paid channels divided by new Shopify customers
- true_ltv: first-order channel × every subsequent order for every customer
- Custom metrics: add your own, computed against the unified layer
Meta reports ROAS 5.2× on Spring Restock 2026. Truefold's reconciled answer: 3.4×. The 34% gap is the over-attribution you've been compounding into your spend.
| Metric | Platform claim | Truefold |
|---|---|---|
true_revenue 4 sources | Shopify $1.20M | $1.25M |
true_roas 3 sources | Meta 5.2× | 3.4× |
true_cac 4 sources | - | $39 |
true_ltv 3 sources | - | $312 |
new_customers 4 sources | Meta 142 | 118 |
See who actually drove the sale, not just who closed it.
Truefold runs four attribution models against the same set of orders. Compare them side by side and see how credit shifts across channels. The default Meta and Google reports are last-click; you've been over-spending on closers and under-spending on influencers for years.
- Last-click: what Meta and Google report by default
- First-touch: who introduced the customer to your brand
- Linear: equal credit across every touchpoint
- Markov: data-driven credit based on conversion probability shifts
- LTV-weighted attribution: credit by the lifetime value the channel actually delivered
On last-click, Meta gets 67% of credit for new-customer revenue. Under Markov, Meta drops to 34% and Email (Klaviyo) jumps from 8% to 22%. Reallocate your spend and you'll grow faster, without spending more.
| Channel | Last-click | First-touch | Linear | Markov |
|---|---|---|---|---|
| Meta Ads | 67% | 22% | 38% | 34% |
| Google Ads | 18% | 14% | 26% | 28% |
| Email (Klaviyo) | 8% | 12% | 18% | 22% |
| Direct / Other | 7% | 52% | 18% | 16% |
Last-click overstates Meta by 33 points. Email is doing 14 points more than it gets credit for.
Continuous audit. Human review. No surprises.
Platforms change. Schemas drift. Definitions decay. Truefold's auditor runs three times a day, catches what changed, and routes anything uncertain into a review queue with a confidence score. Your team approves before anything ships into a dashboard.
- Drift detection: new columns, dropped fields, type changes, broken metric SQL
- Alerts to Slack, email, or webhook within minutes of detection
- Review queue: every AI-suggested definition has a confidence score (0-100%)
- Anything below 80% defaults to needs-review; nothing publishes without sign-off
- Full audit trail: who approved which definition, when, and from what evidence
Shopify adds order.discount_code_id. Within 8 hours, an alert lands in your team's #data channel: 'New column detected. Affects true_revenue and 2 dimensions. Suggested mapping ready for review.' Two clicks to approve.
New column detected: order.discount_code_id
Affects 2 metrics, 1 dimension. Truefold drafted a mapping ready for your review.
- Approvedtrue_revenue96% confidence
- Pendingcustomer_segment82% confidence
- Pendingattribution_window71% confidence
- Pendingchannel_mapping_v264% confidence
Read-only on every source.
Truefold pulls; it never writes. No platform changes, no risk of clobbering your data.
You own everything.
Metric definitions, customer segments, queries: all exportable. Leave any time with what you built.
We don't train on your data.
Your business data never enters a model training set. Single-tenant by default.
Ready to see your real numbers?
Beta access is rolling out to ecommerce brands a few at a time. Tell us about your stack.