Almost always.

I like to call this "latency", rather than "delay" because the offset is deliberate, not some sort of forehead smacking oversight. The 70ms allows time for building data blocks and reacting to dynamic conditions on the network. This is why the Sonos system is so robust. Time alignment between players is within two milliseconds over the long term -- even if the network topology changes in mid flight. These "dynamic conditions" could be the result of a transport node going down or a blast of wireless interference.

The exception is for surround speakers associated with PLAYBAR. In this case the latency is reduced to about 35ms in a compromise to minimize lip sync issues. This can result in some difficulty if another player is Group'd with a PLAYBAR driven surround room. The Group'd rooms will use the standard 70ms latency while the surround room is using 35ms. If this creates an issue, one can adjust the surround room's delay as one would adjust a lip sync delay in most surround systems and align Group'd rooms with the surround room -- at the risk of impacting the surround processing lip sync offset.

Some people, particularly competitors, criticize Sonos for adding this latency. Is it necessary? An interesting experiment is to start simultaneous play with two identical conventional CD players and identical CD's. Since the clocks in the CD players are not absolutely running at the same rate, the players will drift out of sync over time. How long must the players run before the offset becomes obvious? Typically, by the end of the first track there will be obvious misalignment. In a Sonos system, after days of Group play, the alignment will remain within the two millisecond window. Sonos is actively managing the time alignment. If one is VERY perceptive, one might be able to point to an "adjustment event", but they are infrequent, very gentle and you'll miss it if you blink.