Everything in this category lists the same capabilities, and the capability lists are accurate, which is what makes them useless. The differences show up in about an hour of use with your own articles, and the hour is worth more than a week of comparison because the thing that varies is not what a product can do but how well it does the one thing readers experience.
Load twenty of your own articles
Not the sample content. Your own, including the two that are awkward: the long one with a table and the one whose title uses a word your company invented. Ten minutes of pasting buys you a test that means something, and sample content is chosen by the vendor to flatter the search.
Ask three colleagues to find five answers
Give them the questions in the words a real asker would use, not the words in your titles, and watch. Count how many of the five they found in the first three results. That number is the product, and it varies far more between products than anything on the grid.
Then check three things that are not search
Can an article carry an owner and a review date as fields. Can you see, in one view, everything overdue. Can you export the twenty articles and read what comes out. Those three decide whether the base is still true in two years and whether you can ever leave.
Questions people ask about knowledgebase software
Does the spelling matter, knowledgebase or knowledge base?
Not to the product. Both spellings are used by vendors and by searchers, and every serious search in this category treats them as the same word. Use whichever your own people use and put the other one in the body once.
What if the trial period is shorter than our decision?
Run the hour test on the last day of the trial rather than the first. The rest of the trial is exploring; the test is the decision, and doing it early means you spend the remaining weeks rationalising.
Is the cheapest option usually fine?
Often, if it passes the three checks that are not search. The price differences in this category are mostly about seats and support rather than about the reading experience, which is why the hour test is worth more than the price comparison.
What is the most common mistake?
Choosing on the editor. The editor is used by a handful of people occasionally and the search is used by everybody constantly, and the editor is what gets demonstrated.
How much should reading cost?
Nothing per person, ideally. Check whether readers are billed as seats, because the number of people who should be able to read your base is everybody in the company and a per-reader price quietly caps that.