# Nexus: Find what is happening and why.

Nexus works across analytics, CRM, transactions, sales and media to explain what changed. It separates findings from hypotheses, shows the evidence behind each answer and says what the data cannot tell you.

## What Nexus can do.

- **Ask across the business.:** Analytics, CRM, transactions and media in the same question.
- **Trace the evidence.:** Sources, definitions and alternative explanations behind each answer.
- **Choose the right measurement.:** Modeled contribution, attribution credit and experimental effect, kept apart.

## Try Nexus in a simulation.

### Connect audience and business signals

The strongest relationship is repeat behavior, not a demographic label. Here is a connected view of the signals behind the opportunity.

### Show me the source evidence

These are the sources, their limits and the question each one can support.

### Where is the next opportunity?

Compare exploration with purchases. The pattern suggests a question to test; it does not prove the answer.

### Why did revenue change?

Start by separating baseline demand, seasonality and media. This constructed example decomposes orders rather than revenue, so you can inspect how the totals work before reading any change.

### Did the campaign cause additional sales?

Incrementality needs a credible comparison. Here is a randomized holdout example, with uncertainty kept visible.

### Compare attribution models.

Changing a credit rule changes the allocation of observed conversions. It does not change the number of conversions or establish causality.

### Analyze a business funnel by sector.

Keep each stage in the same cohort. The gap is an observation; investigate causes before reallocating spend.

## One example, five stages.

A retailer asks why repeat purchases fell in the North region. The same question, region and period travel through every stage. Illustrative workflow; the numbers are not client results.

- **Understand · Nexus:** Repeat purchases fell 11.8%, concentrated among customers whose last order was 35–60 days ago.
- **Define · Aura:** 42,380 customers match. Prioritize frequent buyers with no purchase in 45+ days.
- **Create · Prism:** Three retention concepts from your brand rules and product catalog.
- **Activate · Radian:** Meta customer audience, push and email. 80% test, 20% holdout. Ready for review.
- **Learn · shared system:** Illustrative result: 7.4% relative lift in repurchase rate versus holdout. Back to Understand.

---

Source: https://randomtruffle.com/agents/nexus