In short: People share what they know when they get something back in the same moment. A promise of future collective value is a public-goods problem, and public goods are under-supplied. Make the return private, immediate and visible.
Key points
- Showing the three most comparable prior decisions and their outcomes is the strongest same-moment return.
- The best capture is a by-product of a memo, call or email the person had to produce anyway.
- Tell contributors when their record is reused; it is cheap to build and motivates experts.
- Backfill ten to twenty past decisions by interview before launch so precedent exists on day one.
- Cut any capture prompt that has no same-session return.
The smartest capture designs are two-way. The system gives the person something useful at the moment of work and takes their knowledge in return. Both halves have to land in the same interaction. When they do not, the economics are simple: the contributor pays now, the organisation collects later, and rational people stop paying.
This is not a culture problem to be solved with a launch campaign. It is a pricing problem.
Four returns that clear on the spot
Precedent. The next screen shows the three most comparable prior decisions and how they turned out. This is the strongest return because it helps with the decision the person is making right now.
A gap they had not seen. "The last two decisions in this segment were reversed after the regulatory change in March." A warning delivered at the right moment is worth more than any summary.
Work done for them. If the record is drafted from the memo, call or email they were going to produce anyway, and the record in turn drafts the next memo, capture costs almost nothing. The best capture is a by-product of work the person had to do regardless.
Credit. Tell people when their record is reused. "Your note on the loan waiver was surfaced in four decisions this quarter, and two of them changed course." Few things motivate experts more than evidence that their judgment travels. It is also cheap to build.
Give first, then ask
The sequence matters. If you launch by asking for reasons and promise precedent later, people experience the system as a tax. Launch with the give side working, even if thin, and ask for contribution only once people are already coming back for it.
The cold-start problem is real, because precedent needs records. Solve it deliberately. Before launch, backfill ten to twenty past decisions in the target class through short structured interviews with the people who made them. It takes a few days and it means the system has something to give on day one.
A check for every prompt
For each capture prompt in your design, write down what the person receives in the same session. If the honest answer is "it will help the team", expect the prompt to be ignored within a month. If it is "the next screen shows the three closest prior cases and their outcomes", it has a chance. If there is no same-session return, remove the prompt or move it to a moment where there is one.
In practice
- List every capture prompt and its same-session return. Cut any prompt without one.
- Backfill ten to twenty decisions by interview before launch, so precedent exists on day one.
- Build the reuse notification early. It costs little and turns contributors into advocates.
- Measure return visits per user per week. If they are not rising, the give side is not good enough yet.
What would prove this wrong. Sustained, high-quality contribution to a system that offers contributors nothing in the moment, driven by mandate alone. Some regulated settings achieve this; if they do so without quality decay, the trade is less essential than this piece claims.
Common questions
Why don't employees contribute to knowledge management systems?
Contributors pay now while the organisation collects later, so rational people stop paying. It is a pricing problem, not a culture problem.
How do I get people to share knowledge in a decision system?
Give a return in the same interaction: comparable precedents, a warning they had not seen, drafting done for them, or credit when their record is reused.
How do I solve the cold-start problem for a decision database?
Before launch, backfill ten to twenty past decisions in the target class through short structured interviews with the people who made them.
How do I know if a knowledge capture system is working?
Measure return visits per user per week. If they are not rising, the give side is not good enough yet.