When sales folk (in any industry, not just publishing) talk about PSA, they're referring to the number of units sold, on average, by any given store in a chain. For example, if Joe's House o' Books has 100 store locations and sells 1,000 units of a given title over a given length of time, their PSA for that interval is 10 units per store (1,000 / 100).
PSA is important to the sales force because it allows us to make direct comparisons between national accounts that have different store counts; it's not exactly fair to tout Book-O-Rama's superior net sales and market share against Joe's House o' Books if Book-O-Rama has 500 store locations and Joe's only has 100. Compare:
Book-O-Rama: 2,000 net copies sold of Disaffected Teen Vampires.
Joe's House o' Books: 1,000 net copies sold of the same title.
Assuming these two are the only competitors in the market, B-O-R outsells Joe's 2:1 and has superior market share (67% to 33%), but their PSA is suprisingly low: only four units per store (2,000 / 500) compared to Joe's ten units per store (1,000 / 100). Why the discrepancy, you might ask? If so, congratulations: you're thinking like a salesperson. We'd be asking the exact same question. Low PSA can be indicative of anything from problems with store location (the stores are in low-traffic or low-population areas, or are in areas not heavily populated by their target client) to insufficient market penetration (though not always, as we've seen that low PSA can be accompanied by high market share).
What does this mean for you? As with market share, not a huge amount, since these comparisons are designed to provide information to publishers and retailers, not individual authors. However, a low PSA could be the result of your title not being in promotion at a given account, so it's always good to ask your agent a few weeks or months before on-sale whether (s)he knows where your book will be placed in-store at various locations.
Tomorrow: PW reports and another management shakeup... what's in store for Borders?