Taming SaaS Sprawl: How to Cut Software Licence Costs 30-50%

How the Sprawl Happens

Nobody decides to waste a third of the software budget. It accrues: a team card-swipes a tool in 2021, the champion leaves, the tool auto-renews. Marketing and product each buy their own analytics. A "temporary" project workspace enters its fourth year. Seats provisioned for leavers are never reclaimed because deprovisioning belongs to nobody. Industry benchmarks consistently find 30-40% of SaaS spend delivering little or no value — our audits regularly find worse, because benchmarks only count the tools companies know they have.

Phase 1: Discovery — You Can't Cut What You Can't See

Build the inventory from four sources, because each one misses things: finance data (AP and expense/card feeds — where shadow IT lives), your identity provider's SSO and OAuth logs (which reveal actual usage, not just purchase), contract records, and a short manager survey for anything on personal cards. The output is a register: tool, owner, cost, renewal date, seat count, and — critically — active usage in the last 30/90 days from SSO telemetry. This register alone changes renewal conversations forever, and pairs beautifully with a contract-extraction pipeline that reads renewal terms out of the PDFs nobody has opened since signature.

Phase 2: The Four Cuts

• Zombie seats (fastest money): licences assigned to leavers or 90-day-inactive users. Typical recovery: 15-25% of seats on collaboration, design and dev tooling. Fix the root cause by wiring deprovisioning to your joiner-mover-leaver process, or it regrows within a year.
• Tier inflation: everyone on Enterprise when 70% of users touch only Standard features. Match tiers to actual feature telemetry; most vendors' own admin consoles will convict them.
• Duplicate capability: three survey tools, two e-signature platforms, four project trackers. Pick one per category on data, not diplomacy: usage depth, integration fit, and exit cost.
• Zombie tools: whole products with single-digit monthly active users. Kill with a 30-day objection window; archive the data; cancel.

Phase 3: Negotiate Like You Mean It

Renewals are where the register becomes money. Start 90+ days out (auto-renew windows are traps — calendar them the day you sign). Bring usage data: "we're paying for 400 seats and 210 log in" is a price conversation the vendor cannot deflect. Benchmark against list-price discounts peers achieve (routinely 20-40% at enterprise scale), trade things that cost you little — term length, case-study rights, payment timing — for percentage points, and always have a credible alternative, even if switching is unlikely. A vendor who knows you can't leave prices accordingly.

The renewal date is the only day of the year the vendor's incentives align with yours. Miss it, and you've agreed next year's price by silence.

Phase 4: Consolidate — Sometimes by Building

The strategic end-game: when three or four adjacent subscriptions each solve 60% of a workflow, a single custom system that fits it exactly often costs less than two years of the combined licences — with no per-seat maths punishing your growth. We've replaced tool-stacks for scheduling, client onboarding and reporting this way; the build-vs-buy logic is the same one we set out for HR systems. Run the five-year total honestly and let the number decide.

Sustained results need light governance: an owner per tool, a default renewal calendar, SSO as the purchasing gate, and a quarterly 30-minute review of the register. That cadence is the difference between a one-off cut and a permanently smaller line.