Five AI Questions Before Your Next SaaS Renewal: What Changed Since You Last Signed

Executive Summary

• Every mid-market company renews 10-20 SaaS contracts annually. In 2026, nearly all of those renewals include AI features that did not exist when the contract was originally signed. Vendors are embedding AI into existing products, pricing it into renewals, and in some cases activating it by default — creating cost, governance, and data exposure the buyer never evaluated.
• 78% of IT leaders report unexpected charges tied to consumption-based or AI pricing models (Zylo, 2026 SaaS Management Index, 40M+ licenses analyzed). The AI premium on enterprise SaaS runs 30-110% above base pricing, and 60% of vendors deliberately bundle AI features to mask price increases.
• The license audit card (the companion to this tool) finds what AI the organization is already paying for. This card prevents the next surprise: five questions to ask before auto-renewing any contract that now includes AI capabilities the buyer did not evaluate at the last signing.
• A 300-person company that auto-renews without asking these questions can expect $40,000-$120,000 in AI-related cost increases across its SaaS stack over 12 months — money that is neither budgeted nor governed.

Why SaaS Renewals Changed in 2025-2026

The SaaS renewal used to be a pricing negotiation. Now it is a governance decision.

Three things happened simultaneously. First, Microsoft, Google, Salesforce, and Zoom embedded AI into base product tiers, raising prices 6-22% to fund the capability (Microsoft, December 2025; Google, January 2025; Salesforce, August 2025). Second, vendors introduced consumption-based AI pricing alongside traditional per-seat models — meaning the invoice can change after the contract is signed based on how much AI the organization uses (Zylo, 2026 SaaS Management Index). Third, 92% of SaaS companies have launched or plan to launch AI features (High Alpha, 2025 SaaS Benchmarks), which means the next renewal of almost any business application will include AI terms that were not in the last contract.

The mid-market buyer who treated the last renewal as a checkbox exercise — “same terms, same price, auto-renew” — is now exposed on three fronts: cost (prices went up), data (AI features access company information), and compliance (new regulatory obligations attach to AI-processed data under the Colorado AI Act, the EU AI Act, and state privacy laws).

The Five Questions

Question 1: What AI features were added since the last contract, and are they on or off by default?

Why this matters: Vendors add AI features between contract cycles. Some activate by default. Google Workspace Gemini turned on automatically for Business Standard and above in March 2025. Zoom AI Companion is on by default for all paid plans (Zoom, December 2025). If the IT team never changed the settings, AI has been processing company emails, documents, and meeting transcripts since the feature launched — without a governance review.

What to ask the vendor:

• List every AI feature added to this product since the last renewal, including features in beta or preview
• For each feature, state whether it is on by default, opt-in, or opt-out
• For each feature that is on by default, state the date it was activated for existing customers

What to check internally: Pull the admin console for the product. Confirm whether AI features match what the vendor disclosed. Vendors have a pattern of activating features in phases — the sales team may not know the feature is already live for existing customers.

Question 2: What data does the AI feature access, and where does it go?

Why this matters: AI features require data to function. Meeting summaries need meeting transcripts. Email assistants need email content. Document tools need document text. The question is what data the AI feature accesses, whether that data leaves the organization’s tenant, and whether the vendor or its subprocessors can use it for model training.

What to ask the vendor:

• What specific data types does each AI feature process (email, documents, chat, meeting audio, CRM records)?
• Does the data stay within the customer’s tenant, or is it sent to external AI model providers?
• Name every subprocessor that handles AI-related data processing
• Does the vendor use customer data to train or fine-tune AI models? If so, is there an opt-out, and is opt-out the default or does the customer need to request it?

Context: For paid enterprise accounts, most major vendors (Microsoft, Google, Salesforce) state they do not train on customer data by default. But the terms vary by product tier. Google’s Gemini in Workspace does not train on enterprise data — but the same user’s personal Google account uses a different policy (Google Workspace Terms, January 2025). Salesforce’s Agentforce processes data through its Einstein Trust Layer, but custom agent deployments may route data differently (Salesforce Trust Layer documentation, 2025). The details matter, and “we don’t train on your data” is not a sufficient answer without knowing which tier, which feature, and which subprocessors apply.

Question 3: What does the AI component actually cost — and how is it priced?

Why this matters: AI pricing takes three forms in 2026, and vendors often mix them within the same contract:

What to ask the vendor:

• Separate the AI cost from the base product cost. What is the AI premium per seat per month?
• If consumption-based: what is the unit of consumption (tokens, queries, agent actions)? What are the overage rates? Is there a cost cap?
• What is the total contract cost with AI features versus without? Can AI features be removed to reduce the price?
• How does this pricing compare to what the organization paid 12 months ago for the same number of seats without AI?

Benchmark: Microsoft 365 E3 rises from $36 to $39/user/month in July 2026, a $3/seat AI premium. Google Workspace Business Standard went from $12 to $14/user/month, a 17% increase. Salesforce Agentforce starts at $125/user/month on top of the base Enterprise license. For a 300-person company on all three platforms (with 50 Salesforce users), the annual AI cost premium is approximately $93,000 — before anyone decides to use the features.

Question 4: Can the organization deactivate, downgrade, or opt out of AI features without changing the base contract?

Why this matters: The emerging vendor playbook is to bundle AI features so tightly that they cannot be removed without downgrading the entire product tier. Microsoft 365 E3 includes Copilot Chat starting July 2026 — the buyer cannot decline Copilot Chat while staying on E3. Google Workspace includes Gemini at all Business and Enterprise tiers — there is no “Workspace without Gemini” option.

What to ask the vendor:

• Can AI features be disabled at the admin level for the entire organization?
• Can AI features be disabled selectively by user group, department, or role?
• If AI features are disabled, does the contract price decrease? By how much?
• If the organization does not want AI features, what is the alternative product tier and its price?

What this reveals: If the vendor cannot separate AI from the base product and cannot reduce the price, the organization is paying for AI regardless of whether it uses it. This is the vendor’s intent — bundling makes AI adoption frictionless but also makes AI opt-out impossible. The buyer’s leverage is in the renewal negotiation, not after the contract is signed.

Question 5: What changed in the terms of service, data processing agreement, or acceptable use policy since the last renewal?

Why this matters: AI features create new contractual obligations — and new risks — that did not exist in the previous agreement. The EU AI Act’s requirements for high-risk AI systems take effect August 2, 2026. The Colorado AI Act took effect February 1, 2026, covering AI systems that make consequential decisions. State consumer privacy laws increasingly include AI-specific provisions. The organization’s existing DPA may not cover AI-related data processing, especially if the vendor added subprocessors for AI inference.

What to ask the vendor:

• Provide a redline comparison of the current terms versus the terms at last renewal
• Identify every clause that references AI, machine learning, automated processing, or algorithmic decision-making
• List every new subprocessor added since the last renewal, with the specific service each performs
• Does the DPA cover AI-specific data processing, or does a separate AI addendum apply?
• What is the vendor’s liability allocation for AI-generated errors, hallucinations, or data processing failures?

What to bring to the GC: The redline comparison and the subprocessor list. The GC contract red lines card (the companion to this tool) specifies five non-negotiable clauses for AI vendor contracts: data ownership, model training opt-out, breach notification, liability allocation, and audit rights. Every renewal that now includes AI features should be evaluated against those five red lines — even if the vendor is an incumbent the organization has used for years.

Key Data Points

What This Means for Your Organization

The renewal conversation changed. A year ago, the CIO’s renewal checklist was: confirm the seat count, negotiate the price, sign. In 2026, every renewal is an AI procurement decision whether the buyer treats it that way or not. The vendor added features, raised prices, changed data handling terms, and in some cases activated AI capabilities before the renewal conversation even started.

The five questions above take 30 minutes per vendor. A 300-person company with 10-15 core SaaS contracts can run the full exercise in a week. The return is immediate: the organization identifies AI costs it did not budget, AI features processing data it did not authorize, and contractual terms it did not review. That visibility converts a passive renewal into a governance decision — and gives the CFO, CIO, and GC the information they need to negotiate from a position of knowledge rather than surprise.

The license audit card finds what AI the organization already bought without realizing it. This card ensures the organization does not buy more of the same. Together, they form the procurement half of AI governance — the counterpart to the acceptable use policy and vendor contract red lines that govern how AI is used after the contract is signed. If your next renewal cycle raised questions worth discussing, I am available at brandon@brandonsneider.com.

Sources

1. Zylo, 2026 SaaS Management Index (40M+ licenses analyzed across enterprise customers, 2026). Reports 54% license utilization, $19.8M average annual waste, 78% of IT leaders reporting unexpected AI/consumption charges, 60% of vendors bundling AI to mask price increases, 61% of IT leaders forced to cut projects due to unplanned SaaS costs. Credibility: Independent platform data, large sample, consistent methodology year-over-year. https://zylo.com/reports/2026-saas-management-index/

2. High Alpha, 2025 SaaS Benchmarks (survey of SaaS companies, 2025). Reports 92% of SaaS companies have launched or plan to launch AI features; 64% embed AI as a supporting feature; 36% describe AI as core to their product. Credibility: VC firm with SaaS portfolio — directional, potential selection bias toward AI-forward companies. https://highalpha.com/resources/

3. Microsoft, “Advancing Microsoft 365: New capabilities and pricing update” (December 2025). Announces E3 increase from $36 to $39/user/month and E5 from $57 to $60/user/month effective July 1, 2026, bundling Copilot Chat into base plans. Credibility: Primary source, vendor announcement. https://www.microsoft.com/en-us/microsoft-365/blog/2025/12/04/advancing-microsoft-365-new-capabilities-and-pricing-update/

4. Google, “The future of AI-powered work for every business” (January 2025). Announces Gemini bundled into all Workspace Business and Enterprise plans with 17-22% base price increase effective March 2025. Previously offered as a $20-30/user/month add-on. Credibility: Primary source, vendor announcement. https://workspace.google.com/blog/product-announcements/empowering-businesses-with-AI

5. Salesforce, “Updated Product Packaging and Pricing” (2025). Agentforce add-on at $125/user/month for Enterprise and Unlimited editions. Agentforce replaces Einstein add-ons and Einstein 1 Editions. Average 6% price hike for Sales and Service Cloud effective August 2025. Credibility: Primary source, vendor announcement. https://www.salesforce.com/news/stories/pricing-update-2025/

6. Zoom, AI Companion 3.0 launch (December 2025). AI Companion included with all paid plans at no additional cost. On by default. Custom AI Companion add-on at $12/user/month. Enterprise connectors (Salesforce, ServiceNow, Box) launching spring 2026. Credibility: Primary source, vendor announcement. https://news.zoom.com/zoom-launches-ai-companion-3-0/

7. SaaStr, “The Great SaaS Price Surge of 2025” (2025). Documents average 8.7% annual SaaS price increases, with range of 10-25% for major platforms. Identifies AI feature embedding as the primary justification for above-inflation increases. Credibility: Independent SaaS community platform, broad industry coverage. https://www.saastr.com/the-great-price-surge-of-2025-a-comprehensive-breakdown-of-pricing-increases-and-the-issues-they-have-created-for-all-of-us/

8. Zscaler, ThreatLabz 2026 AI Security Report (2026). Documents that AI features increasingly operate through embedded capabilities in SaaS applications, with capabilities that often activate by default and interact with enterprise data without being labeled as AI. Credibility: Independent security firm, enterprise-focused research. https://www.zscaler.com/blogs/security-research/ai-now-default-enterprise-accelerator-takeaways-threatlabz-2026-ai-security

9. National Law Review, “SaaS Contracts: Trends Reshaping Negotiations in 2025 and Beyond” (2025). Documents emerging requirement for AI-specific clauses in SaaS agreements: model training opt-out, subprocessor disclosure, AI liability allocation. Credibility: Independent legal publication, practitioner-authored. https://natlawreview.com/article/trends-negotiating-software-service-providers

Brandon Sneider | brandon@brandonsneider.com March 2026