The 90-day SaaS cost cleanup plan

June 16, 2026
SaaS cost optimization 90-day plan
The 90-day SaaS cost cleanup plan

You know you're overpaying for SaaS. Everyone is. The question is how much and where to start. Here's a 90-day plan that focuses on the highest-impact cuts first and builds sustainable habits so the waste doesn't come back.

Days 1-7: find out what you're paying for

Before you can cut, you need to see. Pull your company's credit card and expense reports for the last 3 months. Flag every recurring SaaS charge. You're building a simple spreadsheet: tool name, monthly cost, number of seats, who owns it.

Most companies find 10-20% more SaaS subscriptions than IT knows about. Marketing signed up for a design tool. Sales added a prospecting service. Someone in engineering is paying for a monitoring tool that duplicates what you already have.

For IT cost optimization, this visibility step is foundational. You can't optimize what you can't see. IBM's cost optimization framework makes the same point: discovery comes before everything else.

Days 8-21: audit your biggest per-seat tools

Sort your SaaS list by total annual cost. The top 3-5 tools are where most of the waste hides. For support teams, Zendesk is almost always in the top three.

For each tool, answer: how many seats do we pay for, and how many are actively used? The definition of "actively used" varies, but "logged in within the last 90 days" is a reasonable threshold.

For Zendesk, LicenseTrim generates this report in two minutes. For other tools, you may need to dig into admin panels or API data. Some cost optimization guides suggest starting with the tool you understand best, which is good advice.

Flag inactive seats. Calculate the waste. This is your baseline number.

Days 22-45: clean up and downgrade

Now act on what you found. For Zendesk: downgrade inactive agents to End User (not suspend, downgrade). Convert ticket-viewers to light agents. Review add-on assignments and remove AI/WFM from agents who don't use those features.

For other tools: similar process. Remove inactive users. Downgrade permissions where possible. Cancel subscriptions for tools with low adoption.

According to Zylo's research on SaaS cost savings, the biggest barrier at this stage isn't finding the waste, it's getting internal approval to act on it. Department heads will push back. "We might need those seats later." The answer: most SaaS tools let you add seats back mid-term. You can only remove at renewal.

Keep track of every change you make. You'll want the total savings number for the next step.

Days 46-75: negotiate renewals

Check which tools renew in the next 90-180 days. For each one, prepare a renewal brief: current seat count, actual usage, proposed new seat count, projected savings.

Walk into renewal conversations with data, not requests. "We've used 62 of our 85 Zendesk seats for the past 6 months, we'd like to renew at 65" is more effective than "can we get a discount."

For cloud infrastructure specifically, the same principle applies: usage data is your leverage. Reserved instances, committed use discounts, right-sizing recommendations all depend on knowing what you actually use.

Days 76-90: build the monitoring habit

The cleanup is the easy part. Keeping it clean is harder. In this final stretch, set up the recurring processes that prevent waste from coming back.

Monthly: review agent activity in your biggest per-seat tools. Flag anyone inactive for 30+ days. This takes 15 minutes if you have the right data.

Quarterly: full audit of seat counts vs. active usage across all SaaS tools. Compare to last quarter. This is the operational efficiency part that most companies skip.

At renewal time: pull the data you've been tracking and use it to negiotiate. You'll walk in with 6-12 months of utilization data instead of scrambling for numbers the week before.

For Zendesk, LicenseTrim automates the monitoring. Monthly drift reports show new inactive agents and the cost impact. For everything else, a simple spreadsheet updated quarterly is better than nothing.

With AI-driven cost analysis becoming more accessible and GenAI tools getting cheaper to run, the tooling for this kind of optimization is only getting better. But the fundamentals haven't changed: know what you pay, measure what you use, cut what you don't.