Several factors over many years contributed to the decline and demise of 1-2-3:
- What eventually became Lotus Symphony was originally planned to be 1-2-3 Release 2.0. Lotus fixed this by releasing Symphony as a separate product; Symphony did very well, but the more tightly focused 1-2-3 out-sold it. However...
- Lotus's first product for the Mac, Lotus Jazz, was a GUI implementation of Symphony, not 1-2-3, and was relatively unsuccessful because it was more than many customers wanted (and too much for the 512K Mac of its day. So, where Microsoft had a successful implementation of a GUI spreadsheet on the Mac, Lotus perceived that it had a failure (although Jazz's implementation was quite nice, if perhaps ahead of its time). (An implementation of 1-2-3 for the Mac wouldn't appear until 1991.)
- Lotus's Intel-based GUI efforts were targeted at OS/2, which turned out to be a dead-end. It should be noted that Microsoft strongly encouraged Lotus to follow that path even as Microsoft was pursuing Windows.
- The first major effort to "modernize" the 1-2-3 codebase with a re-implementation in C (vs. assembler), Release 3, took a long time and was targeted specifically at DOS, lacking any accommodation for GUIs. Rudimentary graphics support was later bolted on the side with acquired (as opposed to in-house) software, requiring an awkward auxiliary file format.
- Trying to catch up in the Windows space, Lotus rushed an implementation for Windows on top of the awkward implementation of Release-3-with-graphics. This implementation was as weak as one might expect. (1-2-3 for Mac was built on top of the same infrastructure; necessarily, much more time and effort was spent adapting the code base for a new platform, with arguably better, but still compromised, results.)