Everyone's quoting the "starting at $2/month" line from OpenClaw's pricing page. That's for the hobby tier, which caps at 10k events. It's a trap. When you scale, their model flips to something far less charming: per-seat for the UI, plus per-million-events for ingestion, plus a mysterious "analytics units" fee for queries.
I just helped a team migrate off it after their bill hit $1,200 for a month they projected would cost $150. Their actual volume? 1.2 million events. The devil is in the config, specifically how you define an "event" and how their query engine scans data.
Here's the kicker: their 1M events/month plan isn't a simple line item. It's a bundle. You get, say, 1M events and 100 "analytics units". If you run a heavy dashboard that re-queries historical data every 5 minutes, you burn through those units in a week. Then you're into overage territory, which is billed at a premium.
A rough, realistic breakdown for 1M events/month with a small team:
* **Platform Fee (Basic):** $25
* **Events Overage (you will likely exceed 1M some months):** $50 per additional 0.5M block. Assume $50.
* **Seats (3 engineers + 2 analysts):** 5 seats × $15 = $75
* **Analytics Units Overage (almost guaranteed):** ~200 units × $0.25/unit = $50
You're looking at **~$200/month**, not $2, and that's *before* any committed use discounts. And forget about high-cardinality data or wide events—they'll silently throttle your queries and suggest you upgrade to a "premium" tier.
The real cost isn't the bill. It's the engineering time spent shaving events and writing custom filters to stay within their arbitrary limits. You start building your own analytics system to monitor your analytics vendor.