> ## Documentation Index
> Fetch the complete documentation index at: https://v3.captaindata.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Inputs Processing: Single vs. Batch

> This guide outlines both methods, when to use them, and examples to help you determine the most suitable approach for processing inputs.

When working with Captain Data’s API, a common question arises: **Should you process inputs one by one or in batches?** The answer depends on your specific needs, such as **speed** and **responsiveness**, which can be critical for your product or service.

## TL;DR

| **Single Input Processing**              | **Batch Input Processing**                  |
| ---------------------------------------- | ------------------------------------------- |
| ⚡ **Real-time** results for fast actions | 🏋️‍♂️ **Efficient** for large datasets     |
| ⏱️ Process **one input** at a time       | 🚀 Send **multiple inputs** in one go       |
| 🚫 **Cold start** may occur              | ✅ Reduces network overhead                  |
| 💡 Best for **time-sensitive** tasks     | ⚡ Great for **bulk updates**                |
| ❌ More **API calls** for large data      | 🛠️ **Fewer API calls**, better for scaling |
| 🔄 Needs **faster feedback**             | 📈 Best for **large-scale enrichment**      |

## 🚀 **Single Input Processing**

### What is it?

Single input processing involves sending **one input at a time** to the API endpoint using [POST Launch a Workflow](/v3/api-reference/workflows/launch-workflow).
Each input is handled individually, ensuring that each one is processed separately.

### When to Use It?

Use single input processing when you need to **act quickly** on individual pieces of data, especially when **timing** is crucial.

**Examples:**

1. **Enriching a new user**:
   * A user just signed up on your platform, and you need their data enriched immediately for a personalized experience.
   * Sending one input at a time ensures **real-time updates**, keeping the user experience smooth.
2. **Updating specific database entries**:
   * When updating **100 profiles** in your HubSpot database, sending inputs one by one ensures you only process what's necessary, avoiding the overhead of handling entire batches.

**Why?**

* Single input processing is ideal for **small to medium-sized datasets**.
* It delivers faster, more **targeted results** for time-sensitive needs.

<aside>
  💡 Note that processing single inputs one by one may result in a cold start.
</aside>

## 📦 **Batch Input Processing**

### What is it?

Batch input processing groups multiple inputs into a single API request. Instead of sending inputs one by one, you send **a large set of inputs in one go**.

### When to Use It?

Batch processing is more efficient for **large datasets** or when performing **bulk actions**.

**Examples:**

1. **Updating your HubSpot database**:
   * If you're updating **10,000+ profiles**, batch processing is much more efficient than handling each profile individually.
   * This method reduces unnecessary API calls and minimizes network overhead.
2. **Bulk data enrichment**:
   * When enriching a list of leads, processing them in bulk reduces both **processing time** and **API costs**.

**Why?**

* Batch input processing excels at handling **high volumes of data efficiently**.
* It minimizes API calls, which is crucial for large-scale updates.

## ⚖️ **Single vs. Batch: How to Decide**

A few examples:

| **Use Case**               | **Recommendation**   | **Reason**                               |
| -------------------------- | -------------------- | ---------------------------------------- |
| Enriching a user at signup | **Single Input**     | Faster response for real-time needs.     |
| Updating large databases   | **Batch Processing** | More efficient for high volumes of data. |

## **Rate Limiting on Parallel Inputs**

Captain Data applies rate limits depending on how many inputs you launch in parallel.

If you exceed your rate limit, you will receive a `429 error`. Runs won't be queued automatically, so you’ll need to handle this error on your side.

### Examples Based on a Rate Limit of 100 Inputs in Parallel

* Launching 100 workflows with 1 input each is acceptable. The 101st workflow will be rate-limited and queued.
* If you launch a single workflow with 5000 inputs, no rate limit applies due to batch processing.
* Launching 20 workflows, each with 5 inputs, will hit the parallel input limit.

### **Rate Limiting by Plan**

* **AppSumo, Pro, Lunar, Mars plans**: Limited to 100 inputs launched in parallel.
* **Scale & Growth plan**: Limited to 1000 inputs launched in parallel.
* **Enterprise plan**: No rate limits apply.