The ROI of Sunsetting: How to Cancel 20% of Your SaaS in 90 Days

Every finance team that runs a serious first cleanup finds the same thing. About 20% of the SaaS bill is being paid for nothing. Not underutilized: literally nothing. The tool is not logged into, the seats are stale, the business owner left, or the capability is duplicated by another tool the company also pays for. That 20% is the fastest ROI available to any finance leader.

Ninety days is enough to capture most of it.

What is realistic to cut in 90 days?

The 20% target holds if the company has never run a systematic cleanup. If a program has been running for a year already, the achievable cut drops to 5 to 10%. For most mid-market companies with $1M to $10M in annual SaaS spend and no prior program, 20% is the middle of the range.

The savings come from four sources, in this rough distribution:

• Cancellation of unused tools. 40 to 50% of savings. Fastest. Most straightforward.
• Cancellation of duplicated tools. 20 to 25%. Requires deciding which of two overlapping tools to keep.
• Downgrade of oversized contracts. 15 to 20%. Same vendor, fewer seats.
• Renegotiation on renewals inside 90 days. 10 to 20%. Depends on renewal cadence.

The dollar mix varies. A company with lots of annual contracts and few month-to-month subscriptions will see less immediate cancellation and more renegotiation. A company with a lot of individual card-based subscriptions will see the opposite.

What does the week-by-week plan look like?

Twelve weeks, one owner (usually FP&A or the controller), four phases.

The heavy lifting is in weeks 3 to 6. That is when the finance owner needs the most cover from the CFO because cancellations require business owners to accept the loss of tools they may still be defending emotionally.

Weeks 1 to 2: what does discovery produce?

Two deliverables by end of week 2.

Deliverable 1: complete vendor list. Pulled from AP, card feeds, and SSO. Reconciled by vendor domain. Owned by one finance person. Every row has vendor name, annual contract value, billing frequency, business owner (or "unassigned"), and renewal date.

Deliverable 2: utilization overlay. For every vendor above $5K annual, active users in the last 90 days divided by paid seats. Pulled from SSO login activity or the tool's admin console.

The output is ranked by "waste score": annual value multiplied by (1 minus utilization). Vendors with high value and low utilization float to the top.

If the discovery does not finish by end of week 2, do not proceed to cancellation. A partial list produces partial results and creates political capital problems when the missing 20% shows up later.

Weeks 3 to 6: how do you execute cancellations without breaking anything?

Cancellation is procedural, not judgmental. Apply the same rubric to every candidate.

• Utilization under 20% and no logins in last 60 days. Immediate cancel candidate.
• No named business owner or business owner departed. Immediate cancel candidate.
• Duplicate of an approved tool with clearly better coverage. Cancel candidate, with 30-day migration window.
• Month-to-month billing. Can execute cancellation this week.
• Annual contract with expiring term in next 60 days. Send non-renewal notice this week.
• Annual contract with expiring term more than 60 days out. Log for future action; do not attempt mid-term cancellation unless usage is truly zero.

Every candidate goes through the same three-step process:

1. Data notice to business owner. Send the utilization data and the cancellation intent. Deadline: 5 business days to object with a documented use case.
2. Grace period. If no objection, send cancellation to vendor with a 30-day active grace period. Tool stays live during grace; if someone screams during grace, reverse cancellation and reevaluate.
3. Formal termination. After grace period, contract ends. Users are offboarded. Savings hit the P&L.

Silence is consent. This is the operating principle. Business owners who care about a tool defend it with data. Ones who do not are showing you the tool is not needed.

Weeks 7 to 10: how do you consolidate overlapping tools?

Consolidation is harder than cancellation because it requires deciding which of two tools to keep and executing the migration.

The decision rubric:

• Utilization advantage. If tool A has 70% utilization and tool B has 30%, keep A.
• Feature coverage. If tool A covers 90% of tool B's use cases but not vice versa, keep A.
• Cost per active seat. All-in cost divided by active users. Lower wins, if the other factors are close.
• Migration cost. Estimate the labor cost of migrating from the loser to the winner. If migration cost exceeds one year of loser's contract value, keep both for now and revisit at renewal.

Execute consolidation in this order:

1. Announce the decision. Business owner communicates to team which tool wins.
2. Migration period. 30 to 60 days depending on complexity. Both tools stay live.
3. Cutoff date. Loser is deprovisioned. Contract ends at next renewal or immediately if month-to-month.

Only run consolidation on tools with meaningful overlap and meaningful spend. Consolidating two $5K annual tools takes as much effort as consolidating two $50K annual tools and returns 10x less. Focus on top 20 by combined value.

Weeks 11 to 12: what can you actually renegotiate in two weeks?

Any contract renewing in the next 60 days is fair game. Renegotiation follows a compressed cycle.

• Utilization-based ask. If utilization is 50%, ask for 50% price reduction or seat reduction. Vendor pushback is expected; the anchor matters.
• Cross-vendor benchmark. Have alternative pricing quotes ready. Vendors respond faster when they know they have competition.
• Contract term flexibility. Offer a longer term (e.g., 2 years instead of 1) in exchange for reduced annual price. Sometimes the vendor's incentive is total contract value, not annual rate.
• Feature scope. If you use only the base tier features, negotiate down from enterprise tier.

Realistic outcomes in a two-week window:

• 10 to 15% price reduction on renewals where utilization is above 70%.
• 25 to 40% price reduction on renewals where utilization is below 50% (achieved through seat cuts).
• Longer notice windows on future renewals.
• Cap on rate escalation.

What does the savings math look like?

Concrete example. Company at $2M annual SaaS spend, 87 vendors, no prior program.

• Discovery finds: 22 tools with utilization under 20% ($185K annual). 6 duplicate pairs where one can be cut ($95K annual). 14 tools with utilization between 20 and 50% eligible for seat downgrades ($120K reduction possible). 8 contracts renewing in next 60 days with renegotiation potential ($60K reduction possible).

• 90-day execution:

• Kills 22 unused tools: $185K annualized (roughly $46K captured in the 90 days as month-to-month tools stop billing and annual notices trigger)
• Consolidates 6 pairs: $95K annualized (captured over 30 to 60 day migration windows within the 90 days)
• Downgrades: $120K annualized, mostly landing in months 4 to 9 as renewals cycle

• Renegotiations: $60K annualized, captured immediately on the 8 renewing contracts

• 90-day P&L impact: roughly $200K to $250K annualized savings realized within the quarter, with another $200K to $300K committed to future renewal cycles.

Program cost during those 90 days: 0.25 FTE of finance time plus any tooling. Typical net return: 8 to 12x the program cost, captured before the next quarter-end.

What tools should you never cut fast?

Some categories should not be part of a 90-day sunset even if utilization looks low.

• Identity and access management. Cutting SSO breaks everything downstream.
• Accounting, payroll, benefits. Regulatory and payroll timing risk.
• Core CRM and revenue systems. Revenue continuity trumps cost savings.
• Security and compliance tooling. SOC 2, EDR, backup. Cutting these creates larger downstream cost.
• Production infrastructure. Different discipline; needs its own optimization cycle.

Even if utilization data looks low for one of these, do not cut in a fast sunset. Any real optimization on these needs a longer runway and more cross-functional review.

The mistake to avoid

Most finance teams that attempt a fast sunset stall at week 4 because they treat every cancellation as a debate rather than a procedure. The debate consumes more cost than the cancellations save. What actually works is a rubric applied consistently, a 5-day objection window with data attached, and silence-is-consent as the default. Not every cancel will land clean, and about 5 to 10% will need to be reversed. That is a rounding error against the 20% you captured. The discipline that makes the sunset work is the same discipline that keeps spend flat in year two, so run it as an operating procedure, not a one-time project.