· Checked November 18, 2026

Parsed transaction APIs or your own decoder: what is true

A parsed transaction API returns human-readable instructions plus execution results such as status, logs and balances, because it reads transactions after they run. Your own decoder over raw shreds sees transactions earlier but without any of those results, and only understands the programs whose layouts you teach it. Neither is strictly better.

What a parsed API returns, as a category#

Parsed APIs sit on top of nodes that have executed the block. That is why they can include success or failure, error codes, log lines, inner instructions and balance changes: all of these come from execution metadata, the structure RPC calls TransactionStatusMeta (E4 tech facts X-02). The RPC method getTransaction shows the protocol-level version of this with its jsonParsed encoding, which expands well-known programs into named fields. Commercial parsed APIs go further with their own schemas for swaps, transfers and mints.

None of that metadata exists in shreds. An entry carries signed transactions as the leader ordered them and nothing more (X-01).

Belief and what actually holds#

BeliefWhat holds
A parsed API understands every programIt understands the programs its maintainers have added. A new or niche program appears as raw instruction data until they do.
Parsed data and raw data arrive togetherParsed data needs an executed block, so it comes after the transaction is processed. Raw shreds carry the transaction earlier, without the outcome.
Your own decoder is only for high-frequency tradingIndexers, research datasets and alerting on programs no API covers are just as common reasons.
Decoding yourself is just parsing bytesTransactions are the easy part. v0 transactions need lookup table state to recover addresses, and programs need their layouts or IDLs.
A raw decoder can tell you whether a swap succeededIt cannot. The result is execution output, so you confirm it later through RPC or a confirmed source.
Anchor programs always use sha256-based discriminatorsThat is the default, but since Anchor 0.31 a program can set custom discriminators, so read its IDL rather than assume (E4 tech facts A-05).

What a decoder over raw shreds must add#

Three things a parsed API hands you for free become your job. First, lookup table resolution: a v0 transaction lists table indexes, and the addresses come from the table account's state, fetched or tracked by you (T-06, T-07). Second, program knowledge: reading an IDL for Anchor programs and hand-written layouts for the rest, which the Anchor decoding guide turns into code. Third, outcomes: every decoded row stays provisional until a status check says what happened, a gap described in what your decoder cannot see.

Choosing between them#

Use a parsed API when you need outcomes on every row, broad program coverage without writing layouts, and you can wait for execution. Build your own decoder when timing before the block settles matters, when you track programs an API does not parse, or when you want full control of the schema and its history. Many teams run both and key them on the same transaction signature, so early and settled views join cleanly. The build or buy comparison puts the full workload of the second path in one table.

Parser questions#

Does a parsed transaction API include failed transactions?#

Typically it can report them, because it reads execution results. A raw decoder sees failed transactions too, but cannot tell they failed.

Is jsonParsed the same as a parsed API?#

It is the RPC's own parsed encoding for well-known programs. Commercial APIs add their own schemas on top of similar execution data.

What is the hardest part of decoding yourself?#

Usually lookup tables and program layouts, not the transaction format. Both need data from outside the transaction.

When your own decoder is the wrong investment#

If you need outcomes immediately on every row, track only popular programs, and nobody on the team will maintain layouts as programs upgrade, a parsed API is the sensible choice. A raw decoder is a long-term commitment, and its value is only realised if early timing or custom coverage actually changes what you build.

Try it on real data#

The quickest way to judge the effort is to decode a day of real traffic yourself. A Standard feed gives you the raw stream, and the shred parser library cards show which pieces already exist.

Sources#

Raw shreds on your own server

Ultra Low Latency Shreds is $200 and Standard Shreds is $100. You pay in crypto, tell us which IPv4 address and UDP port should receive the stream, and we wire it up by hand.