Obviously these customers are hashing with the same hash and seed Google are using; they have to be or the whole exercise would be pointless. These organisations may not have the nous to prevent Google from reassembling the original data, so there are no guarantees. Also, they're not anonymising when they're matching two separate data sources. That's not anonymisation in anyone's book. That's pure sophistry, bollocks and misleading bullshit to cover insidious mining and profiling of people's PII.
All of this becomes irrelevant, however, when you realise that, to Google, your identity is that pesky hash. Talking about anonymity at that point becomes pointless.