> For the complete documentation index, see [llms.txt](https://docs.helodata.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.helodata.com/resources/fair-usage.md). # Fair-usage policy helodata is built on shared infrastructure — gateways, IP pools, support staff. Fair-usage protects everyone on the platform from outliers, and protects you from accidentally over-running resources. ## The short version * Run a normal business workload — you're fine. * Run a workload that disproportionately strains any shared resource — we'll reach out. * Run a workload that demonstrably damages the service for others — we'll throttle or suspend, and you'll get notified first. We've never silently suspended a customer for fair-usage. Communication first, action second. ## Specific guardrails ### Per-IP request rate (gateway products) We track requests per second per **exit IP** behind the scenes. A single sub-user firing 500 RPS through one sticky session is using one exit IP much harder than designed. We: * Surface this in your dashboard's **Statistics → Concurrency** panel * Email you when one IP exceeds 100 RPS sustained for >10 minutes * Throttle that IP at 200 RPS to prevent ISP-level retaliation against our pool Spread your traffic across more sessions to avoid this. ### Restricted target categories Detailed in [Restricted targets](/products/overview/restricted-targets.md). Trial accounts have an extra strict list; paid accounts can request exceptions for legitimate use cases. ### ISP IP replacement Free replacement up to 14 days from purchase, then 10% of batch per period. Replacing every IP every week defeats the "static" in ISP and forces us to over-provision the pool. Stay within the allowance or upgrade the batch quantity. ### Webhook delivery If we send a webhook to your endpoint and it fails: * 3 retries over 24 hours * If still failing, the webhook is disabled and we email you * Repeated re-enable + fail loops indicate your endpoint isn't really there — we'll pause until you confirm ### API polling Polling our API is fine, but polling **the same endpoint every second** when you could subscribe to a webhook or use `If-None-Match` is wasteful. We rate-limit and may further throttle endpoints with cheaper alternatives. ### Bulk operations Don't: * Create 10,000 sub-users in a loop (use one or a handful, with proper labels) * Place 100 separate ISP orders for 50 IPs each (place one for 5,000 IPs) * Replace IPs in a tight loop hoping for better ones (cool-off and ask support) Do: * Open a ticket if you have a legitimate need for unusual scale — we can usually accommodate with prior notice ## Service SLA (Pro+ plans) Our service SLA, where applicable: * **API**: 99.9% monthly uptime * **Gateway**: 99.9% monthly uptime * **Status page**: real-time at [status.helodata.com](https://status.helodata.com) SLA breaches result in service credits applied to the next invoice — see your plan's terms. ## What triggers a fair-usage review A combination of: * Per-IP RPS sustained > 100 for >10 minutes * Error rate from your traffic > 30% for >30 minutes (could indicate target rejection or abuse) * Abuse reports from target sites referencing your sub-user * ISP IP replacement requests > 20% of a batch in one week When triggered: 1. Email to the account owner with details 2. 3 business days to respond 3. Joint review: is this a legitimate workload pattern we can accommodate? 4. Outcome: adjustment to limits, plan upgrade recommendation, restriction, or all-clear ## What's not fair-usage Things that **aren't** governed by fair-usage: * High overall traffic, if it's within your plan and within per-IP norms * Many sub-users, if each is for a real purpose * Aggressive (but legal) scraping of allowed targets * Geographic diversity across many countries We don't punish scale — we punish patterns that damage the platform. ## How to stay clear * Read the [Choose a product](/getting-started/choose-a-product.md) guide and match the product to your pattern * Spread traffic across **multiple sessions** rather than hammering one IP * Use the [Statistics](/dashboard/statistics.md) dashboards to spot your own outliers before we do * Open a ticket when in doubt — we're happy to consult ## Contact Questions about fair-usage? **** (for abuse reports) or **** (for everything else).