Jito bundles, explained without the jargon

What a bundle actually is

A bundle is a list of transactions submitted as one unit, with two guarantees attached. They execute in the order you specified, and they either all land in the same block or none of them land at all.

The second half of the mechanism is the part most people are actually buying: bundles are delivered to block producers through a dedicated path rather than broadcast into the general forwarding network. Your transactions are not sitting somewhere observable while they wait.

Those are two independent properties, and separating them is the whole point of this page, because most use cases want one and pay for both.

Atomicity: all or nothing

Atomic execution matters when transactions depend on each other. The classic case is a set of operations where a partial result is worse than no result: create a position and hedge it, or buy across several pools where landing half the plan leaves you exposed.

For a volume campaign, this property is largely irrelevant. Campaign swaps are independent of one another. If swap four hundred fails, swap four hundred and one is unaffected and the campaign carries on. There is nothing to make atomic.

This is worth knowing because atomicity is the property most often used to justify a premium, and it is the property a campaign has the least use for. Where it does matter is at launch, where a bundler is trying to acquire supply across many wallets in one block and a partial fill genuinely changes the outcome, which is a launch mechanic rather than a campaign one.

Private submission and what it actually prevents

This is the property worth paying for, and it is worth being precise about what it does.

Broadcasting a transaction widely means more parties observe it before it settles. Anyone who can see an order before it lands can position around it, buying ahead of your buy and selling into it afterwards. Private submission removes the easy version of that observation.

What it does not do is make your activity private in any general sense. Once the block is produced, everything is permanent and public. Private submission is about the window before settlement, not about the record afterwards. Anyone who reads your transfer history tomorrow sees exactly the same thing whether you bundled or not.

For campaigns specifically, the value is proportional to how predictable your flow is. Repetitive activity in a recognisable rhythm is the most anticipatable kind, which is why the private path matters more here than for a single retail swap. The sandwich explainer covers the mechanism this defends against and the three other defences that pair with it.

How tips work, and how to size them

A tip is not a priority fee and confusing the two leads directly to overpaying.

The sizing rule is relative rather than absolute. What matters is your tip compared to what else is competing for the same block, which moves constantly. Live tip percentiles are published, and reading them takes seconds. Copying a number from a months-old guide is how people end up paying launch-grade tips on routine swaps, which is one of the more expensive mistakes in this category and shows up clearly in a campaign's fee breakdown.

One practical consequence of the last row deserves emphasis. A priority fee is charged whether your transaction succeeds or fails, while a tip is only charged when the bundle actually lands. That asymmetry means the two behave very differently during congested periods: an aggressive priority fee on a run with a high failure rate is money spent on transactions that produced nothing, whereas an aggressive tip on a bundle that never landed costs you nothing beyond the attempt. Understanding which of the two you are escalating changes what a bad hour costs you.

When bundling earns its cost

• Repetitive flow that is easy to anticipate. Volume campaigns qualify, which is the main case on this site.
• Anything with a deadline. A swap that must land in a specific block has no alternative.
• Operations that must not partially complete. The atomicity case, which is rare outside launch mechanics.
• Contested moments generally. When many participants are competing for the same block space, the private path is worth more than usual.

A useful way to think about the spend is as insurance with a variable premium. What you are buying is the removal of a specific extraction path, and the value of that removal scales with how contested the moment is and how predictable your flow would otherwise be. On a quiet afternoon with well-varied campaign settings, that value is genuinely small. During a launch hour with a metronomic schedule, it is large. Paying the same tip in both situations means overpaying in one and underpaying in the other, which is why a fixed tip figure copied from a guide is worse than reading the live percentiles each time.

When it does not

Being honest about the limits is more useful than the sales pitch.

Quiet conditions. When block space is not contested, ordinary submission with a sensible priority fee lands reliably and the tip buys very little.

Slippage-driven failures. If your swaps are failing because they move the pool more than your tolerance allows, no submission path fixes that. The cause is depth and swap size, and the failure guide covers how to tell which error you are actually hitting.

As a substitute for spreading. Bundling a campaign that still routes everything into one pool at identical sizes on a regular clock has protected the delivery of a pattern that remains completely obvious. The submission path is one layer; varied sizing, uneven spacing and multi-venue routing are the others, and they matter more.

The practical summary for a token team: you want the private path, you have almost no use for atomicity, and you should be paying tips sized to current conditions rather than to a launch scenario. Any service quoting bundling as a premium add-on is charging you for the half you do not need. On this site it is on by default and its cost sits inside the flat fee, which you can see as an exact SOL figure in the console on the home page before anything is submitted.