Understanding Wasted Seats

Wasted seats are one of the most actionable numbers in The GAiGE. They show you, in dollars, exactly how much of your AI spend is going to seats nobody's actually using.

What counts as "wasted"

A seat is wasted when:

1. A user is assigned to a tool in The GAiGE (so we count them toward the seat count)
2. They had zero "yes" responses to Q0 ("Did you use [Tool] today?") in the period

In other words: they had access, they were prompted, and they never reported using it. The cost of those seats — unused users × cost per seat — is the wasted spend.

Where you'll see it

• Dashboard — the Wasted spend card, with the dollar figure and the count of unused seats
• Reports → By Tool — a per-tool Unused column, with amber chips when the count is high
• Insights — automatically called out when one tool has a particularly large amount of wasted spend

What to do about it

Three reasonable responses, in increasing order of severity:

1 · Reassign

Take the tool away from the unused users and give it to someone on a waiting list. Common when a tool is in demand and you've got idle seats blocking access.

2 · Train

If the unused users should be benefitting from the tool but aren't, the issue is often "they don't know how". Q5 ("Why didn't you use it today?") will tell you — look for the "I don't know how to use it" response. The Insights panel automatically flags this when it crosses 15%.

3 · Cut

Reduce your seat count with the vendor. If the same 12 ChatGPT seats have been dormant for three months running, you don't need them. The wasted-spend number is the dollar saving you'll get by cutting.

A worked example

Imagine the dashboard says:

Wasted spend: $1,649 / mo across 52 unused seats

That's roughly $20,000 a year going to seats producing no measurable value. Even acting on half of it — say cutting 25 seats — would be a $10k+ annual saving.

Drill into Reports → By Tool to see which tools are responsible. Often it's one or two tools dominating the figure. Those are your targets.

A note on response rates

If your response rate is low overall (say <30%), the "active" count understates reality — some users may be using the tool but not answering pulses. In that case the wasted-spend number is upper bound rather than absolute truth.

Two ways to address this:

• Improve response rate — usually a Chrome extension rollout issue. See Why is response rate low?
• Sanity-check with users directly — if 12 ChatGPT seats look unused but you happen to know one of those people definitely uses it, you may need to investigate why their pulses aren't firing

But generally, even with imperfect response rates, the relative picture is reliable — the tools with the most wasted seats really are the most over-licensed.

Related: Setting tool costs and why it matters · Reading your dashboard at a glance · Why is response rate low?