Never. If your product is generating revenue and profit, the profit the only wise thing is to invest in the product. If you do so, the product will have more features and becomes bigger. If you did not, your competitor will have done so, and will have innovated to a build a better business.
One release added a "tagging" feature (similar to Gmail tags) while the next release had a "hierarchical folders" feature (similar to Outlook folders).
The thinking was that tags and folders did pretty much the same thing, and folders were clearly superior since they were hierarchical. So we stripped the tagging and shipped with just the folder feature.
Perhaps unsurprisingly, there turned out to be customers that had built a whole lot of import/export/processing infrastructure around the tags, and were less than happy about having to reimplement it all using folders. Plus they had a ton of tagged data and we hadn't included a tool to convert the old format to the new one. (Because we thought noone were using the tagging feature, it had only been in the product for a single release.)
The lesson learned is that released features are similar to published APIs: If you remove or change them you will inevitably break something for someone.
I think that some operating systems have gone like that way - features are not always removed, but some feats are harder and harder to find; other parts are replaced with automatic parts, which will have hardcoded or automatic answers to questions, which were asked before; some tools are removed, because things get better and you dont need them anymore. I am sometimes really surprised when I think that nowadays OS is basically desktop, window and one toolbar and start/K/whatever menu - of course, there is a lot more, but it all orbits around this simple set of things, which clearly do not evolve into bigger complexity and more functions, but more like into elegance, simplicity and features over functions.
4 comments
[ 2.9 ms ] story [ 20.4 ms ] threadOne release added a "tagging" feature (similar to Gmail tags) while the next release had a "hierarchical folders" feature (similar to Outlook folders).
The thinking was that tags and folders did pretty much the same thing, and folders were clearly superior since they were hierarchical. So we stripped the tagging and shipped with just the folder feature.
Perhaps unsurprisingly, there turned out to be customers that had built a whole lot of import/export/processing infrastructure around the tags, and were less than happy about having to reimplement it all using folders. Plus they had a ton of tagged data and we hadn't included a tool to convert the old format to the new one. (Because we thought noone were using the tagging feature, it had only been in the product for a single release.)
The lesson learned is that released features are similar to published APIs: If you remove or change them you will inevitably break something for someone.