What your trade computes that others do not
Support load is not proportional to the number of clients, it is proportional to the number of questions already answered elsewhere. A large share of tickets has a written answer somewhere: in the documentation, in a ticket closed six months ago, in a colleague's message. The cost is not answering, it is finding — and that cost is paid by the most expensive person available at the moment the client presses.
Then, a publisher's support mixes two kinds of request that nothing distinguishes on arrival: those about usage, and those revealing a product defect. The first are settled by an answer, the second must become a roadmap item. A ticketing tool that does not sort them pushes everything at engineering and turns developers into the front line.
Finally, version matters. The same question has a different answer depending on the release installed, the option enabled or the client's deployment. A knowledge base that ignores that dimension produces answers that are right in general and wrong in particular, which costs more than no answer at all.
What most publishers work around by hand
Nothing exotic: these are the normal frictions of a team outgrowing its internal tools.
- The same question returns weekly and gets rewritten from scratch weekly.
- The useful knowledge sits in closed tickets, where nobody goes looking.
- A developer is interrupted for a question that was never technical.
- The documentation exists but no longer matches the client's installed version.
- The client chases by email because they have no way of seeing where their request stands.
- Nobody knows which feature generates the most tickets.