Gus Mueller, author of VoodooPad, writes a report on three years of building a small Macintosh software business.
One thing he misses -- or at least doesn't mention -- is that the software business is seasonal: sales decline in the summer. The effect can be swamped by other factors -- good news, major releases, hard work -- but it's real and significant.
In year one, Mueller got a huge publicity gift in July, which masked the summer slowdown, and figured the ramp-up year end carried good news about his product. Year two, spring comes, and with the approach of summer Mueller senses a product in decline. Maybe the falling sales were due to delays in releasing the product, but they might be the season, too. It's possible that neither the uptrend nor the downtrend said much about the product: the uptrend wasn't a sign that the product was good, nor was the subsequent downtrend a sign that the product had gotten stale.
You can always find an explanation for sales trends, but it's hard to know whether your explanation is right.