The 7 SaaS Tools Every CFO Should Audit Before Q3 Renewals

A 10,000-employee SaaS company we modeled last month was paying roughly $13.9M of annual seats across seven contracts. Their active-user data said they needed closer to $9.4M. The other ~$4.4M — over 30% — was structural waste sitting in the seven vendors every enterprise over-buys from. Q3 is when most of those contracts hit their 90-day notice window. If you're not auditing in May, you're paying retail in October.

This is the short list. Seven vendors, seven waste signatures, and the audit step that gets you a real number per contract.

Try the free calculator — 15 seconds, no signup. Paste your headcount, see the typical waste range across these seven contracts.

Why Q3 is the renewal cliff

Most enterprise SaaS deals get signed in January or February — fresh budget, new fiscal year, account executives with quotas to close. Those contracts are almost always 12 months with auto-renewal and a 60-to-90-day notice window. A contract signed February 1 with 90-day notice means you have to give notice by November 1 or you're locked in another year at the same seat count. That puts the audit work in Q3.

The companies that lose this round wait until the renewal email lands in October, scramble for usage data they don't have, and accept whatever the account manager offers. By then you've lost the leverage of "we're prepared to walk." Vendr's contract-renewal guide makes the same point: buyers who open the conversation 90+ days out land in a different pricing band than buyers who don't.

Pull your contracts. Sort by renewal date. Audit the seven below first — they account for the bulk of the dollar value at most enterprise companies.

The audit framework: contracted vs active vs replaceable

For each contract, you need three numbers:

1. Contracted seats — what the contract PDF says you're buying. Don't trust the vendor admin console; that reflects what's provisioned, not what's committed.
2. Active seats — who actually logged in in the last 30 days. Pull from the IdP or the vendor's usage report.
3. Replaceable seats — active users whose throughput an AI agent can match at a fraction of the cost. Different from "unused."

The waste signature is contracted − active. The compression signature is replaceable × per-tool agent compression %. Both feed the renegotiation ask. The unused SaaS license playbook walks through the contracted-vs-active gap in more depth.

A note on prices throughout this post: we use vendor list prices (Salesforce $165, Outreach $130, etc.) because that's what shows up on the contract you're auditing. The catalog rates we model the calculator against ($100 / $100) are the negotiated enterprise rates you should be paying — the gap between list and negotiated is itself a renewal lever.

1. Salesforce — the $165/seat list ceiling

Salesforce Enterprise is the heaviest line item at most enterprise SaaS companies — list price $165/seat/month. The waste signature is two-headed. First, provisioned-but-inactive seats: marketers who got Sales Cloud licenses three years ago "just in case," support reps upgraded from Service Cloud Essentials who never used the Enterprise features. Typical waste: 22–35% of provisioned seats inactive for 30+ days.

Second — where most CFOs miss six figures — AI agent compression on active seats. Salesforce Agentforce's catalog compression on Salesforce is 30% (the prospecting and case-handling subset that the native agent runs end-to-end inside the CRM). 11x's Alice runs an alternative SDR overlay also at 30% catalog compression on Salesforce. You don't drop the AE; you drop the seat count on the SDR side because volume flows through the agent.

Audit step: Pull the Salesforce login report (Setup → Users → Login History). Cross-reference with your IdP. Anything inactive 30+ days is a renegotiation target. For active SDR seats, model the Salesforce Agentforce overlay at 30% on Salesforce. Note: AiSDR is a separate SDR agent that compresses Outreach and Salesloft at 30% each, Apollo and ZoomInfo at 20% each — it does not run on Salesforce directly, so don't double-count it here. The Salesforce renegotiation playbook has the full script.

2. Outreach — the "every AE needs one" myth

Outreach lists at $130/seat/month and gets bundled into onboarding for every new sales hire, nobody asking whether the rep is actually doing high-volume outbound. Inbound-focused AEs, customer success teams who got "just in case" licenses, sales engineers who log in twice a quarter — all paying full price.

Typical waste: 30–40% of provisioned Outreach seats send fewer than 20 emails/month. On top of the inactive-seat problem, AiSDR's catalog compression on Outreach is 30% for active reps, and Regie.ai compresses Salesloft at 45% if your stack runs there instead — sequence personalization, follow-up cadence, subject-line A/B all move to the agent.

Audit step: Export Outreach's user activity report. Filter for users sending fewer than 20 emails over the trailing 30 days — those are drop candidates. For high-volume reps, run the AI agent compression model before committing to the same seat count next year — apply 30% to Outreach specifically (AiSDR's catalog rate), not a higher blended SDR-stack number.

3. Slack — the "everyone gets one" tax

Slack Business+ runs $15/seat/month, which sounds cheap until you multiply by 10,000 employees and realize you're spending $1.8M/year. Every full-time employee gets a Slack seat in onboarding by default. Productiv's 2024 SaaS engagement data shows the average engagement rate across enterprise SaaS apps is roughly 45% — and Slack, which is one of the heaviest-usage tools in any stack, still leaves 30–45% of seats sporadic-or-dormant.

Slack is also where you should look hardest at tier downgrades. Most enterprise companies are on Business+ ($15) when Pro ($7.25 on annual billing; $8.75 monthly) covers the actual feature usage. The "we need Business+ for compliance retention" line is real but often overstated.

Audit step: Pull Slack's Analytics → Members tab. Anyone who's sent zero messages and read fewer than five channels in 30 days is a drop candidate. Then audit the tier. No clean AI agent compression on Slack itself — this is pure utilization audit.

4. Zendesk — the support-team-staffed-for-2019 problem

Zendesk Suite Professional sits at $115/seat/month for most enterprise support orgs. Support headcount got built up during 2018–2022 when ticket volume was growing and AI deflection didn't exist. Most companies haven't re-baselined.

The waste signature here is less about inactive seats (support agents are usually active) and more about replaceable seats. Sierra (60% catalog compression on Zendesk), Decagon (65%), and Intercom Fin (50%) all credibly deflect tier-1 tickets — but their flat-fee economics only pencil at scale. Sierra unlocks around 130 Zendesk seats year-1 (~87 steady-state); Decagon around 95 (~67 steady-state); Fin's economics are volume-driven, not seat-driven, and only pencil if your team would otherwise be hiring through a 30%+ growth curve.

Audit step: Pull last quarter's ticket volume by agent and category. Tier-1 categories (password reset, billing question, basic how-to) are the deflection target. If you're below ~80 support seats and not hiring aggressively, the honest answer is "renegotiate Zendesk down to active count and skip the agent layer until support headcount triples."

Try the free calculator — 15 seconds, no signup. Model the Zendesk + Sierra unlock thresholds before you renew.

5. Zoom — the post-pandemic over-provisioning bill

Zoom Business runs ~$20/seat/month and is the cleanest example of post-pandemic structural waste in the SaaS stack. In 2020–2021, every employee got a Pro or Business license because nobody knew who'd need to host calls. Five years later, those licenses are still provisioned by default — and roughly 40–55% go unused in any given month.

Second problem specific to companies with both Zoom and Microsoft 365: duplicate video conferencing seats. If sales runs on Zoom but internal meetings migrated to Teams, you're paying for two platforms across an overlapping user base.

Audit step: Pull the Zoom user activity report (Account Management → Reports → Usage). Anyone hosting fewer than two meetings/month drops or downgrades to Basic (free). Then audit the Zoom + Teams overlap. No meaningful AI agent compression on Zoom itself.

6. Notion — the "free trial that became a paid seat" classic

Notion Business runs $20/seat/month (Plus is $10) and is the textbook example of the trial-to-billable seat leak. Somebody sets up a free workspace, invites collaborators, and a year later you have 80 paid seats half of whom haven't logged in this quarter. Every workspace member counts as a paid seat regardless of usage.

Typical waste: 35–50% of paid Notion seats are inactive 30+ days. There's also a tier-mismatch problem — Plus ($10) covers most use cases, and many companies are on Business primarily for SAML SSO, paying the full delta on every seat for a feature only IT needs.

Audit step: Notion admin → Members → sort by "Last active." Anyone inactive 30+ days drops. If SAML SSO is the only Business-tier reason, run the math on Plus + a separate SSO tool. No AI agent compression that materially matters yet.

7. Adobe Creative Cloud — the license-tier mismatch

Adobe Creative Cloud All Apps Teams runs ~$90/seat/month for the full Suite. The contrarian take: Adobe's waste pattern isn't seat compression — it's license-tier mismatch. Most "Creative Cloud users" at non-design companies (marketing, product, sales ops) only use Photoshop and maybe Acrobat. Single App is $35/seat/month. Migrating two-thirds of your Creative Cloud users to Single App saves $660/seat/year with no productivity hit.

I'll admit a limitation here: Adobe Creative Cloud is more about license-tier mismatch than seat compression — a different lever. AI image tools compress some Photoshop workflows, but the overlap is messier than the Sierra/Zendesk math, and "replace the human" is rarely the right framing for creatives.

Audit step: Adobe admin → User assignments → look at which apps each user has actually opened in 60 days. Anyone who's only opened one or two apps is a Single App candidate. The savings here usually exceed the seat-count savings on cheaper tools.

A real example: 10,000-employee SaaS enterprise

Anonymized but real. Enterprise B2B SaaS, 10,000 employees, total SaaS spend ~$60M/year. The seven-tool audit numbers:

Total identified waste: $4,446,000/year, or 32% of these seven contracts ($13.86M contracted annual), concentrated in seven vendors. Some renegotiated down (Slack, Zoom — vendors won't fight on volume tiers); some were tier downgrades (Adobe). Zendesk had no compressible waste at this scale because Sierra/Decagon don't pencil under their unlock thresholds and the team was 100% engaged. Net realized savings after Q3 renewal cycle: ~$2.9M (about 65% of identified, the typical realization rate).

That's seven audits, four renewal calls, two tier downgrades. The work fits in 60 hours of finance + procurement time spread over the quarter. The output is seven figures.

How to apply this — the 90-day audit calendar

If you're reading this in May:

May (week 1–2): Pull contracts. Sort by renewal date. Identify which of the seven above hit their 90-day notice window between August and November.

May (week 3–4): For each, pull contracted seat count from the PDF, active seat count from the IdP or vendor admin console, and last-30-day usage report. Build the gap table.

June: For Salesforce and Outreach, model the agent overlay using per-tool compression rates (Salesforce Agentforce on Salesforce 30%, AiSDR on Outreach 30% — never a single blended rate, and never AiSDR on Salesforce since the catalog has no such mapping). For Zendesk, check whether your seat count clears Sierra's ~130 / Decagon's ~95 year-1 unlock; if not, skip the agent and renegotiate the contract instead. The seat compression primer is the cleanest math explainer.

July: Outreach to account managers with the gap data. Don't ask "what discount can you give us" — ask "we need to drop to X seats at renewal; what's the path." Most vendors counter at 30–50% of the gap.

August–September: Negotiate. The auto-renewal traps post covers the contract clauses that determine how much room you actually have to walk.

October–November: Sign at the new seat count.

The companies that lose this race skip May and start in September. By then no audit data, no agent pilots, no leverage. The negotiation becomes "please give me a discount" instead of "here's the seat count we need."

The bottom line

Seven contracts. Seven distinct waste signatures. The total dollar value at most enterprise companies sits in the $2M–$5M range — large enough to justify a quarterly audit, small enough that it falls through the cracks if nobody owns it.

Pick three to audit this month. The other four wait until next quarter. Whichever three you pick, you'll find waste — the question is whether you find it before the auto-renewal clock runs out or after.

Try the free calculator — 15 seconds, no signup. Paste your stack, see the typical compression number across all seven before you commit to next year's renewal.