PLAN NOTES

When to add the WebSocket add-on, and when to wait

The raw WebSocket add-on has no term discount, so buying it early saves nothing. Add it when your recovery plan exists, not at checkout.

The raw WebSocket add-on attaches to the three REST plans at $99 per 30 days, and unlike the base plans it has no term discount, as the pricing page lists it, checked 2026-10-07. That single fact settles most of the timing question: buying the add-on early saves nothing. Add it when your application is ready to consume raw frames, and not before.

What the add-on actually adds

The add-on relays raw frames: a first snapshot of your subscriptions, then every update, for you to reconstruct market state yourself. It is a building material, not a feature. The WebSocket page is blunt about the boundary: forwarding raw frames is not reconstructing a market, and a reconnect invalidates old assumptions, checked 2026-10-07.

Until you have the consumer for that material, the add-on is a monthly fee for frames you will not read.

The readiness test

The page's own standard is the right gate: subscribe, establish a baseline, then consume, and test the failures you intend to survive. If you have not run your consumer through an incomplete snapshot, a duplicate chunk and a dropped connection, you are not ready, and the add-on will simply bill you while you learn.

The honest sequence is build first, buy second. Your consumer can be written and rehearsed against recorded frames before a single 30-day block starts.

Why waiting costs nothing

Because the add-on has no term discount, there is no price to lock in. The 30-day block at $99, the 90-day at $297 and the 180-day at $594 are the same per-day price, per the pricing page. Starting the add-on the day your consumer passes its failure tests costs exactly what starting it at checkout would have cost, minus every day you did not pay for an unused transport.

If the raw frames turn out not to fit, the money you did not spend is the win. REST snapshots or processed SSE alerts cover most applications, and the page says so directly.

The exception: raw-dependent products

If your product is defined by the frames, a low-latency state engine or a market reconstruction service, the add-on is not an add-on at all; it is the point, and it belongs in the cart with the base plan from day one. The plan guide notes the first frame is a snapshot and REST becomes the recovery path rather than the main one, which is the shape such a product takes.

Build the consumer, test the failures, then start the clock. The add-on rewards readiness, not eagerness.