Songlinks › Comparison
Three ways to look up a songwriter
Every tool that answers “what did this writer register?” picks one of three approaches, and the approach decides what it can and cannot get right. Here is what each does with the same registrations — and, at the end, a head-to-head whose questions were fixed before anyone answered them.
The same registrations, three ways
The numbers below come from one measured run: a hundred real registered names sent through as a list, which reached 281 registrations that belonged to a person. 109 of them belonged to the writer the list asked about; 172 belonged to somebody else who shares a name with one of them.
Why this is the whole difficulty
A registration is a name as somebody once typed it, with the works registered under it. Nothing in the data says two records are the same human, and nothing says two humans are different. Those two absences are not symmetrical problems — a tool that solves one by assumption creates the other — which is why the approach matters more than the interface.
- About six in ten songwriter records carry no IPI name number at all, so anything keyed on the identifier cannot see them.
- 8% of songwriter records that do carry one share their exact registered name with a record under a different number, so anything keyed on the name merges people.
Measured 2026‑09‑15 on 2,826 registration records drawn at random; the guides show the working.
Measured against the leading public lookup
Approaches argue; measurements settle. In September 2026 we asked Songlinks and the leading public songwriter lookup about the same 240 writers, drawn at random in three separate draws, and checked both against the registry’s own records. The questions and the pass mark were fixed before either answered, which is what stops a comparison from being a list of the things we happen to do well.
Figures from the comparison run 2026‑09‑15.
The study measured four questions in all, and the chart shows two of them. The others were telling apart different people who share a name, where the two lookups came out level, and recognising one person across several registrations — the area we are actively extending, using records we can check.
What a comparison like this cannot tell you
It was run on writers drawn at random from one body of registration data, on one day, and both systems have moved since. It says nothing about which tool suits your workflow, what either costs, or how they behave on a catalogue like yours rather than on a random sample. Songlinks is in beta, and its groupings are a model’s judgement that no person has checked.
The measurements are re-run and the figures here move with them, or this section comes down. That is the only way a comparison on a web page stays worth reading.
See it on your own writers
The fastest way past an argument about approaches is a name you already know the answer for. Look one up — free, nothing to sign up for — or send a list and we will run the whole thing: bulk songwriter lookup.