Duplicate work is the most expensive kind

Of all the forms of wasted time in a knowledge-work organisation, duplicate work is the most maddening because it's entirely unnecessary. The work has already been done. The research exists. The report was written. The decision was made and the reasoning was discussed. But nobody can find it, so someone does it again.

Bloomfire's research puts the figure at 14% of knowledge worker time, over four hours per week per person spent recreating information that already exists somewhere in the organisation. Asana's data is consistent: 209 hours per year per employee on duplicative work.

For a 100-person company at an average loaded cost of $65/hour, 4 hours per week of duplicate work costs roughly $1.35 million per year. The work is being done twice, sometimes three times, across different teams and different tools, each time producing slightly different results that create their own coordination overhead when they inevitably conflict.

The forms of duplication

Research redone. A team needs market data for a strategic decision. They commission or conduct research, not knowing that another team did essentially the same analysis three months ago. The original analysis is in someone's personal Google Drive folder, titled in a way that nobody else would recognise.

Reports rebuilt. A quarterly report is assembled from scratch because the previous quarter's report is in a former employee's documents, and the template was never saved to a shared location. The new report takes a full day. Finding and updating the existing one would have taken an hour.

Decisions relitigated. A team debates whether to use technology A or technology B, reaches a conclusion after a week of discussion, and implements it. Six months later, a different team faces the same question, debates it for another week, and may or may not reach the same conclusion, because the original discussion happened in a Slack thread that has long since scrolled away.

Solutions reinvented. An engineer solves a tricky integration problem. Another engineer on a different team encounters the same problem months later and solves it independently because the first solution was never documented anywhere searchable. Both solutions work. The time spent on the second one was entirely wasted.

Why process doesn't fix it

The typical response to duplicate work is a process: "always check the shared drive before starting new research," "always save templates to the team folder," "always document decisions in the wiki." These processes fail because they add friction to the workflow (checking the shared drive before starting anything is an extra step that's easily skipped under time pressure) and because the shared drive, team folder, and wiki are themselves fragmented and poorly searchable.

The fix is findability. If existing work is searchable by meaning across all the places it might live (Google Drive, Slack, email, Notion, meeting recordings), the duplication drops because people discover existing work before they recreate it. Not through a special process or an extra step, but through the natural act of searching before starting.

Self-writing documentation extends this further by capturing decisions, discussions, and solutions from the channels where they naturally occur. The Slack discussion about technology A vs B becomes a searchable decision record. The engineer's solution becomes part of the system documentation. The knowledge persists and is findable without anyone doing any additional work.

Frequently asked questions

How do we measure how much duplicate work we're doing? Track it informally for one month: ask team members to note whenever they discover that work they've done already existed elsewhere. Even rough tracking reveals patterns and quantifies the problem.

What's the most impactful type of duplication to eliminate first? Decision relitigation, because it wastes not just the time of the discussion but the time of everyone whose work is delayed while the decision is being re-debated. Decision logs have the highest ROI.

Can AI help detect duplication before it happens? An AI assistant with access to your full knowledge library can flag when a new research request or project brief overlaps with existing work. This is one of the most immediately valuable applications of AI in knowledge management.

Related reading: The cost of scattered knowledge, The search tax, Nobody reads the wiki. Related pages: Search, Connections, Decision log.