Skip to content
Curious Context Labs

case study · IT · Custom app

Finding the software you pay for but nobody uses

How card and accounting data, matched to sign-ins and contracts, can produce a ranked list of savings for owners to approve.

business
Real estate
build
Custom app
timeline
2–3 weeks
read
3 min

01 the business

Architecture practice

70 staff, around 90 software subscriptions

Software is bought on company cards by whoever needs it. Seats stay paid after people leave, two teams pay for the same tool, and renewals auto-charge before anyone asks whether the tool is still used.

02 before

The problem

Software gets bought on company cards by whoever needs it, so nobody has the full list. Seats stay paid for after people leave, and two teams pay for the same tool under different names. Renewals auto-charge before anyone asks whether the tool is still used.

  • No complete list of tools exists.
  • Leavers' seats keep being billed.
  • Renewal dates are missed.

03 the build

What the AI does

  1. Step 1: Pulls software charges from Xero or QuickBooks and card feeds from Ramp or Brex, and groups them by vendor.
  2. Step 2: Matches each tool to sign-in data from Okta, Google Workspace or Microsoft 365 to see who actually uses it.
  3. Step 3: Flags seats with no recent logins, tools that overlap, and plans that could drop a tier.
  4. Step 4: Reads contract and invoice PDFs to record renewal dates and notice periods.
  5. Step 5: Shows owners a ranked list of savings to review, with the evidence behind each one.

paused: waiting for a person to approve

How it was put together

  • Grouped software charges from accounting and card feeds by vendor.
  • Matched each tool to sign-in data to see who actually uses it.
  • Flagged unused seats, overlapping tools and plans that could drop a tier.
  • Recorded renewal dates and notice periods from contracts and showed a ranked savings list.

04 human in the loop

Where a person stays in charge

Each tool owner confirms whether a flagged seat or subscription can go, and finance approves any cancellation. The app never cancels or downgrades anything on its own.

  • Tool owners confirm whether a seat or subscription can go.
  • Finance approves every cancellation.
  • The app never cancels or downgrades anything itself.

05 rollout

From kickoff to live in 2–3 weeks

  1. Week 1Connect accounting, cards and identity data.
  2. Week 2Produce the first inventory and savings list.
  3. QuarterlyRe-run and review upcoming renewals.

06 after

What changes

You have one list of every tool you pay for, who uses it and when it renews. Cutting unused seats becomes a short monthly review rather than a once-a-year spreadsheet hunt.

worked example · your numbers will differ

Assume 70 staff and software spend of $2,500 per employee a year, with 15% of it unused or duplicated.

70 × $2,500 = $175,000 a year. 15% of that is about $26,000 a year of avoidable spend.

What to measure

  • Annual software spend.
  • Unused seats removed.
  • Renewals reviewed before their notice date.
  • Duplicate tools retired.

07 evidence

What the research says

  1. Organisations waste an average of $21 million a year on unused SaaS licences, and SaaS spend averages $4,830 per employee.

    Zylo, 2025 SaaS Management Index (opens in a new tab) · 2025

  2. Gartner research put the average cost of poor data quality at $12.9 million a year per organisation.

    Gartner, Data quality: why it matters and how to achieve it (opens in a new tab) · 2020

08 lessons

Mistakes to avoid

  • Cancelling on login data alone. Some tools are used rarely but matter.
  • Missing notice periods. Many contracts need 30 to 90 days' notice.
  • One-off audits. Spend creeps back without a quarterly review.

09 questions

Common questions

How much waste is typical?
Zylo's 2025 index found organisations waste an average of $21 million a year on unused licences, with SaaS spend averaging $4,830 per employee. Smaller firms waste less in total but often a similar share.
Does it need single sign-on?
It works best with it, but it can also use usage exports from individual tools.

keep reading

All case studies ->

Tell us how your business runs. We'll tell you where AI fits.

A free 30-minute call. You talk, we ask questions, and you leave with a few concrete ideas — or an honest answer that AI isn't the right fix yet.

Working with teams around the world.