Before the grid: which problem are you solving?
"Knowledge management software" covers tools that do different jobs. A comparison that ranks them on a single scale compares different things. So start by naming your problem:
| Your problem | The right family of tools |
|---|---|
| Answering the same questions fast, internally or for customers | A knowledge base or help centre |
| Writing, organising and finding things together | A wiki or collaborative workspace |
| Keeping, versioning, proving | A document management system |
| Not losing know-how held by a few people | Knowledge mapping, with measured mastery and risk |
The four do not exclude each other: many organisations already run three of them. KnowledgeCapital belongs to the fourth family and connects to the others rather than replacing them — the guide knowledge management software: how to choose describes these families. The grid below is mostly for telling apart tools that claim to solve the fourth problem.
The grid: twelve criteria, one question, one proof
Each criterion is asked as a question and checked with a concrete action during the trial. An answer you cannot check does not count.
| Criterion | The question to ask | The proof to ask for |
|---|---|---|
| 1. Tacit knowledge | Does the tool know what is written nowhere? | Declare a piece of knowledge with no document and see whether it shows up anywhere. |
| 2. Measured mastery | Are levels declared, or measured? | Take a test and watch the person's level change. |
| 3. Computed risk | Is criticality computed, and is the method published? | Remove a holder and watch the score move by itself. |
| 4. Link to real work | Is knowledge tied to processes and tasks? | Open a task and read what it requires, and who holds it. |
| 5. Validation | Who gives a piece of content its authority? | Submit a sheet and follow its route to publication. |
| 6. Freshness | How is outdated content spotted? | Find the review cadence, and what happens when it is overdue. |
| 7. Existing systems | Which systems are read, in which direction, with which rights? | Connect a real site or folder read-only, or read the step-by-step procedure. |
| 8. Data and AI | Where does the data live, where does the AI run, what leaves the network? | Read the list of allowed outbound connections. |
| 9. Security and traceability | Customer isolation, audit log, corporate directory? | Find in the log who performed a given action. |
| 10. Reversibility | Can you get everything back and re-import it elsewhere? | Read the reversibility plan and the export format. |
| 11. Languages | Multilingual interface and search, right-to-left writing? | Switch language mid-task and search in another language. |
| 12. Adoption and total cost | Who administers it, for how long, at what real price? | Count the admin hours during the trial; read the published prices. |
Weighting without fooling yourself
Score each tool from 0 to 3 on each criterion, then give each criterion a weight from 1 to 3 according to your problem. Two profiles to make it concrete:
- "Our experts are retiring": weight 3 for tacit knowledge, measured mastery, computed risk and the link to real work; weight 1 for content freshness.
- "Nobody can find anything": weight 3 for existing systems, freshness and adoption; weight 1 for computed risk.
Add one simple rule: a 0 on a weight-3 criterion rules the tool out, whatever its total. That is what stops a very pleasant tool from winning on a problem it does not address. And have the grid filled in by two people who do not confer: the gap between their scores shows where the demo was vague.
A fourteen-day trial protocol
- Days 1 and 2 — Explore a demonstration space: look for the four answers (what do we know, who knows it, what will we lose, what are we doing about it).
- Days 3 to 5 — On a blank space, map one process with the five or six people who run it. The guide on building a knowledge management system from scratch gives the order.
- Days 6 and 7 — Assess mastery, look at which knowledge rests on one person, and check that the result surprises you at least once.
- Days 8 to 10 — Test the link with existing systems: a real site, library or folder, read-only.
- Days 11 and 12 — Run a small assessment campaign and compare declared with measured.
- Day 13 — Read the commitments: security, backups, reversibility, where the data lives.
- Day 14 — Fill in the grid as a pair, and decide.
Questions to ask any vendor
- Show me a piece of knowledge held by a single person: how did the tool find it?
- What happens to the risk score when a holder leaves, without anybody editing it?
- What leaves our network, to which servers, and where is it logged?
- Does the AI run on your side, a third party's, or ours?
- What do you read from our systems, and what do you write to them?
- Which connectors are available today, and which are only announced?
- If we leave, what do we get back, in which format, and how fast?
- What does a year really cost for our scope, training and administration included?
What KnowledgeCapital answers — and what it does not do
Applied to itself, the grid gives this. Tacit knowledge, mastery, risk: knowledge is tied to tasks, mastery is scored from 0 to 5 then measured through assessment campaigns, and criticality is computed with a published method. Validation: sheets and lessons learned go before committees, and validated sheets are reviewed at a pace set by their criticality. Existing systems: Confluence, SharePoint and Microsoft 365, document management systems through CMIS or WebDAV, Notion, intranets and the inbound API, all read-only; only Confluence can also receive validated sheets, if you ask for it. Data and AI: local AI, confined outbound connections, a hosted edition with one isolated database per customer, or a sovereign edition on your servers. Languages: French, English, Spanish and Arabic, right-to-left included. Cost: priced per tracked person per month, from €6 to €22 depending on the plan, sovereign edition on quotation, 14-day trial with no card.
What it does not do, said just as plainly: it is neither a wiki nor a document management system, and it does not replace your writing tool; Google Drive, GitLab and ServiceNow connectors are not available today; SharePoint, document management systems, Notion and intranets are read without anything being written to them; and connectors are included from the Professional plan upwards. If your main need is a help centre for your customers, a dedicated knowledge base tool will serve you better.
Confluence is a trademark of Atlassian; SharePoint and Microsoft 365 are trademarks of Microsoft Corporation; Notion is a trademark of Notion Labs, Inc.; Google Drive is a trademark of Google LLC; GitLab is a trademark of GitLab Inc.; ServiceNow is a trademark of ServiceNow, Inc. KnowledgeCapital is neither affiliated with nor endorsed by these companies.