Yep, that's 'increase', not 'improve'. We're talking about optically increasing review scores in order to satisfy the marketing needs of both booking websites and their suppliers (hotels). Would you stay in a hotel reviewed with an average 7? I definitely wouldn't. That's not because it's a low 'mark' in itself, but because it's a very low review score for a hotel.
Why is that? Let's take the review system of Booking.com as an example. They currently have 7 dimensions for review: cleanliness, comfort, facilities, staff, value for money, free wifi and location. Even if staff is rude and the rooms are hardly cleaned, the other dimensions may up the score considerably (assuming that they all weigh equally, which seems to be the case).
A colleague of mine once stepped into a Brussels hotel room, finding hair all over the place. Its review score was 7. He also stayed in a very well reviewed hotel in Athens, finding a gun store next door and a neighbourhood that felt about as safe as the slums of Naples. Yes, the hotel was excellent but he feared for his life when he stepped outside.
Booking.com has the added benefit of being able to see both the distribution of scores and the distribution of visitor types. The latter is useful because it allows you to see if the people reviewing match your type of visitor. Let's say that if 'groups of friends' are the main visitors, it's probably best to avoid such a place if you're an older couple.
A general rule of thumb to follow is that if you want a really good hotel, don't settle for a score less than 8,5. Be aware though, that outside the realm of Booking.com, which uses only confirmed reviews, there are many sites using Tripadvisor scores. These need not be unreliable per se, but they are unconfirmed and more open to manipulation.