My usual estimate is the ability to generate 102dB peaks. At 16 watts (12dBW) that means speakers of 90dB efficiency.
This approximation comes from a compilation of Stereophile speaker reviews over a few years, in the late 90s; it is the minimum of their recommended power range. So I say this is good for "most audiophiles, most of the time."
For the median audiophile, you might get away with 6dB less efficiency, so 84dB minimum. This is mostly a combination of not needing exceptional loudness, and the fact that SETs overload really gracefully, so minor clipping is not objectionable. (High feedback and/or solid state amps have very hard clipping, so even minor amounts are quite objectionable.) Room size and absorptive treatments also play a role.
Sorry, that puts you in unclear territory - it's a judgement call.
