Optimizing Cost
Learn practical strategies to minimize scraping costs while maintaining accuracy and efficiency.
Running scrapers efficiently helps you get the most value out of MrScraper without unnecessary usage costs.
This guide explains how to optimize scraping costs through smart configuration, scraper mode selection, scheduling, and output management.
1. Choose the Right Scraper Mode
MrScraper offers two AI Scraper modes that affect both cost and performance:
| Mode | Description | When to Use | Cost Impact |
|---|---|---|---|
| Cheap Mode | Fast and lightweight extraction. Suitable for simple, structured pages. | Use for lists or data with consistent structure (e.g., product names, prices). | 💲 Lowest |
| Super Mode | More accurate and intelligent extraction. Ideal for complex layouts or nested content. | Use for unstructured data or multiple data types (e.g., job listings, reviews). | 💲💲 Higher |
Note
The cost difference between Cheap Mode and Super Mode depends on the target website. For websites that are easy to access and extract from, both modes may have similar costs. On more complex websites, Super Mode may perform additional attempts and processing to successfully extract data, which can increase token usage and overall cost.
2. Limit Extraction Scope
The broader your prompt, the more resources are used.
Narrowing your extraction target saves cost and improves precision.
| ❌ Expensive prompt | ✅ Optimized prompt |
|---|---|
| Extract all details from this e-commerce page. | Extract product name, price, and rating from this page. |
Best Practice
Focus on the fields you actually need. If you only need names and prices, don’t extract full descriptions or metadata.
3. Reuse and Recycle Results
Each time you run a scraper, you pay for the extraction process.
Avoid re-running scrapers unnecessarily by reusing existing results.
Example Workflow:
- Store your output in a Database Connection or export as CSV/JSON.
- Only re-run the scraper when the target site changes.
- Use Scheduling for incremental updates instead of full refreshes.
4. Use Scheduling Strategically
Scraping too frequently can increase costs without adding value.
Instead, match your schedule to how often the target data changes.
Tip
Avoid overlapping schedules. Running multiple scrapers at the same time can spike costs and reduce performance.
5. Optimize Proxy Usage
While proxies add flexibility, they can slightly increase scraping overhead.
- Use MrScraper’s built-in proxy instead of external providers to save cost and setup time.
- Only enable proxies for websites that require regional access or IP rotation.
- For internal or low-risk pages, disable proxy entirely.
Warning
Disabling proxies may change the content returned by websites that serve region-specific data. If you need accurate local pricing, product availability, or search results, use a proxy or Geo Targeting for the target region.
6. Monitor Usage
Keep an eye on your scraper usage and performance using the Analytics feature.
Regularly review:
- The number of pages or URLs processed
- Success rate and data completeness
- AI vs. manual scraper distribution
This helps you spot overuse early and plan your scraping strategy efficiently.
Pro Tip
Start small. Run initial tests on limited pages using Cheap Mode, verify output accuracy, and scale up gradually.
7. Leverage Reruns for Lower Costs
If you need to scrape a website repeatedly, use Rerun instead of creating a new scraper each time.
MrScraper uses caching to optimize repeated scraping tasks. When you rerun an existing scraper, the platform can reuse previously processed information, often reducing both token usage and execution time compared to a new scraper.
Reruns are ideal for:
- Refreshing previously collected data
- Monitoring pages for updates
- Scheduled scraping workflows
- Recurring extraction from the same website
8. Cap Retry Spending with Max Retries and Token Cap
A failed scrape that keeps retrying is one of the easiest ways to spend tokens without getting data back. In the Playground, the Basic Settings panel gives you two limits that bound the cost of a single run.
Turn on Retry first, then set either or both:
| Setting | Description |
|---|---|
| Max retries | Maximum number of retry attempts after the initial request. |
| Token cap | Maximum total tokens the initial request and all its retries can use. |
Retrying stops at whichever comes first: the scrape succeeds, it reaches Max retries, or the running token total reaches the Token cap. Each attempt's cost is rounded up and added to the running total.
How to use them to control cost:
- Set a token cap when testing an unfamiliar site. It puts a hard ceiling on a single run, so a page that fails repeatedly can't quietly consume your balance.
- Lower Max retries on sites that fail consistently. If a page is blocked, the third and fourth attempt usually fail the same way as the first. Fix the cause with Super mode or Geo targeting instead of paying for more attempts.
- Leave room for at least one retry. If the first attempt alone costs more than the cap, the scrape runs once and doesn't retry at all. On pages with variable load times or heavy bandwidth, set the cap to a few times your expected per-run cost.
Tip
Run a scrape once with no cap to see its typical token cost, then set the cap based on that number rather than guessing.
For worked examples of how the cap interacts with retries, see Playground Token Cap in Token Plan.