Measure internal tools before you redesign them
Measure internal tools before you redesign them
Measure internal tools before you redesign them
The short answer
Internal and enterprise tools rarely come with revenue or retention numbers, so good design work on them goes unseen. Before the project starts, pick two to four measures that matter to people who must use the tool, such as errors, time on a key task or time to learn it. Record a baseline, then measure again after release.
Internal and enterprise tools rarely come with revenue or retention numbers, so good design work on them goes unseen. Before the project starts, pick two to four measures that matter to people who must use the tool, such as errors, time on a key task or time to learn it. Record a baseline, then measure again after release.


Illustrative concept, not client work or measured results.
Why the usual numbers don't fit
Much metric advice starts from Google's HEART framework, which maps product goals to signals and metrics. NN/g's Page Laubheimer points out that its engagement, adoption and retention dimensions are "meaningful only for consumer products, where the user can decide whether to use the product." When the tool is part of someone's job, retention just tracks staff turnover. His CASTLE framework measures cognitive load, advanced feature usage, satisfaction, task efficiency, learnability and errors instead.
Why the usual numbers don't fit
Much metric advice starts from Google's HEART framework, which maps product goals to signals and metrics. NN/g's Page Laubheimer points out that its engagement, adoption and retention dimensions are "meaningful only for consumer products, where the user can decide whether to use the product." When the tool is part of someone's job, retention just tracks staff turnover. His CASTLE framework measures cognitive load, advanced feature usage, satisfaction, task efficiency, learnability and errors instead.
Set the baseline first
NN/g's Alita Kendrick lays out benchmarking: choose about 5 to 10 top tasks, "aim for 2–4 metrics," measure before the redesign, then repeat with the same method. Pilot the study first. She warns that "bad numbers are worse than no numbers," so pick a plan your team can keep running. Laubheimer prefers time on task from controlled usability tests over analytics, since interruptions skew it.
Start with what frustrates people
Jared Spool's tip in a 2015 talk is simple: "Count the error messages." Errors are cheap to track and point at real pain. Numbers only say whether something moved, not why, so pair them with research. If past work was never measured, say so plainly and describe what you observed. Don't invent a number.
Set the baseline first
NN/g's Alita Kendrick lays out benchmarking: choose about 5 to 10 top tasks, "aim for 2–4 metrics," measure before the redesign, then repeat with the same method. Pilot the study first. She warns that "bad numbers are worse than no numbers," so pick a plan your team can keep running. Laubheimer prefers time on task from controlled usability tests over analytics, since interruptions skew it.
Start with what frustrates people
Jared Spool's tip in a 2015 talk is simple: "Count the error messages." Errors are cheap to track and point at real pain. Numbers only say whether something moved, not why, so pair them with research. If past work was never measured, say so plainly and describe what you observed. Don't invent a number.





Before your next internal-tool project
- Name the top tasks for each group of users.
- Pick two to four measures that fit those tasks.
- Record a baseline with a method you can repeat.
- Count error messages from day one.
- Measure again after release and write down what changed.
Before your next internal-tool project
- Name the top tasks for each group of users.
- Pick two to four measures that fit those tasks.
- Record a baseline with a method you can repeat.
- Count error messages from day one.
- Measure again after release and write down what changed.
References
- Page Laubheimer, "CASTLE Framework for Productivity/Workplace Applications", Nielsen Norman Group (29 January 2023; read 19 November 2026).
- Alita Kendrick, "7 Steps to Benchmark Your Product's UX", Nielsen Norman Group (13 September 2020; read 19 November 2026).
- Jared Spool, "Is Design Metrically Opposed?", talk transcript, UIE (UXIM, April 2015; read 19 November 2026).
- Kerry Rodden, Hilary Hutchinson and Xin Fu, "Measuring the User Experience on a Large Scale: User-Centered Metrics for Web Applications", CHI 2010, Google Research (2010; read 19 November 2026).
References
- Page Laubheimer, "CASTLE Framework for Productivity/Workplace Applications", Nielsen Norman Group (29 January 2023; read 19 November 2026).
- Alita Kendrick, "7 Steps to Benchmark Your Product's UX", Nielsen Norman Group (13 September 2020; read 19 November 2026).
- Jared Spool, "Is Design Metrically Opposed?", talk transcript, UIE (UXIM, April 2015; read 19 November 2026).
- Kerry Rodden, Hilary Hutchinson and Xin Fu, "Measuring the User Experience on a Large Scale: User-Centered Metrics for Web Applications", CHI 2010, Google Research (2010; read 19 November 2026).