# Churn prediction goals: it’s not about accuracy, it’s about impact

When building a churn prediction model, the goal isn’t to be “accurate”—because accuracy is often misleading. If only 1% of your customers churn and your model predicts “no churn” for everyone, you’ll still get **99% accuracy**—but provide **zero value**.

#### What You *Should* Optimize For

Instead, you should focus on two key metrics:

* 🔍 **Precision**: Of the customers flagged as *at risk*, how many actually churned?
* 🎯 **Recall**: Of *all customers who actually churned*, how many did your model correctly identify?

#### Why Relative Performance Matters

If your baseline churn rate is 1% and your model identifies churners with 10% precision, that’s already **10× better than random**.\
That leverage is where real business value is created.

But ideal targets depend on your business model and the **cost of acting** on churn signals.

#### Realistic Benchmarks by Industry

📦 **B2B SaaS (Low-touch, High-volume)**

* **Churn**: 5–15% monthly
* **Target**: Recall 70–85%, Precision 20–30%
* **Why**: Outreach is cheap and scalable—casting a wide net works.

**📰 Publishers / Media**

* **Churn**: 3–10% monthly
* **Target**: Recall 60–80%, Precision 20–35%
* **Why**: Low-cost interventions (email nudges, personalization) are effective at scale.

**🏋️‍♂️ Fitness & Wellness**

* **Churn**: 7–20% monthly
* **Target**: Recall 60–75%, Precision 30–40%
* **Why**: Actions often use limited resources (e.g. trainer time)—precision is critical.

**💳 NGOs / Recurring Donors**

* **Churn**: 20–35% annually
* **Target**: Recall 65–80%, Precision >40%
* **Why**: Donor value is high, and most interventions are worth the effort—even with some false positives.

**💡 Energy Providers**

* **Churn**: 10–20% annually
* **Target**: Recall 60–75%, Precision 35–50%
* **Why**: Churn often follows rate changes; success depends on delivering the right message to the right people.

**🎲 Lotteries & Games of Chance**

* **Churn**: 15–40% annually
* **Target**: Recall 70–85%, Precision 25–40%
* **Why**: High natural churn; reactivation is possible—timing and messaging matter most.

**🧩 Digital Platforms / Marketplaces**

* **Churn**: 5–20% monthly (often defined as inactivity)
* **Target**: Recall 60–80%, Precision 20–35%
* **Why**: Reactivation is typically low-cost and scalable; depends on monetization model.

**💼 B2B SaaS (Tech-touch)**

* **Churn**: 2–8% monthly
* **Target**: Recall 60–75%, Precision 25–40%
* **Why**: Mid-scale users benefit from lifecycle emails and nudging—good recall drives growth.

**🤝 B2B SaaS (High-touch)**

* **Churn**: 5–15% annually
* **Target**: Precision >50%, Recall 40–60%
* **Why**: CS time and incentives are costly—false positives are expensive.

#### 🎯 Final Takeaway

Don’t obsess over abstract model scores.\
Measure your model’s **uplift versus random**, and **balance precision and recall** based on the **cost and value of acting**.\
That’s how churn prediction drives real, measurable business impact.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.churned.io/churned-docs/getting-started-with-churned/churn-prediction-goals-its-not-about-accuracy-its-about-impact.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.