Is Salesforce Becoming What It Once Replaced

“Tech didn’t beat us. Incentives did.”

[Hero image placeholder: Yellow-dominant abstract editorial artwork showing a once-simple Salesforce contract map folding into a dense grid of credit meters, agent workflows and renewal numbers. Avoid generic corporate stock photography.]

What Agentforce Buyers Are Missing.

A former Siebel Systems executive once said that to Anders, SaaSed’s co-founder. Siebel did not lose because the product was useless. It lost because the way it was sold created unhappy customers at scale.

Two decades later, the Agentforce push creates a familiar commercial pattern. Agentforce is not just another Salesforce add-on. It marks a structural shift in the Salesforce SaaS model, from relatively predictable per-seat economics to workload-driven consumption pricing.

In a seat-based model, Finance can forecast spend from headcount, licence mix and negotiated unit price. In a consumption model, cost follows the volume, frequency and design of agent activity. The new equation is simple to write and hard to govern:

workload → agent action → credit usage → cost

That is where the risk begins. Most organisations cannot yet map that chain with confidence. Many Salesforce account teams cannot either. Complexity is a tax on the unknown. If they cannot convince you, they will confuse you, often without meaning to.

The Hidden Pricing Shift: Seat-Based vs Consumption

For most large Salesforce estates, the commercial centre of gravity has historically been seats. Sales Cloud users. Service Cloud users. Platform users. Add-ons layered by role, geography or business unit.

This model is imperfect. It creates shelfware, over-provisioning and renewal pressure. But it is still relatively legible. A CFO can ask how many users need access. A CIO can challenge whether a role needs a full licence. Procurement can benchmark unit pricing, discount logic and uplift exposure.

Agentforce changes the control surface.

The question is no longer only, “Who needs a licence?” It becomes, “Which work is being delegated to agents, how often does that work run, which systems does it touch, and what does each interaction consume?”

That is a very different procurement muscle.

A seat is assigned. A workload behaves. It changes with adoption, process design, data quality, integrations, customer demand, seasonality and internal governance. The commercial risk is not simply that people use Agentforce. The risk is that nobody can explain, before scale, how usage turns into spend.

This is why consumption pricing needs to be treated as a risk model, not a modernity badge. It can be efficient when the organisation understands the unit economics. It can be expensive when the organisation buys credits before it understands the work.

The problem is rarely the technology in isolation. Agentforce may prove technically strong. The issue is the commercial structure wrapped around it, especially when AI agents, Data Cloud and flex credits arrive inside a renewal conversation that already has time pressure.

The Incentive Problem: Why Salesforce AEs Push Agentforce

Sales incentives do not need bad intent to create bad outcomes.

If a sales organisation is trained to maximise annual contract value, and it is given a strategic AI product with board-level attention, the direction is obvious. Bring Agentforce into the renewal. Attach it to transformation language. Bundle it with adjacent products. Create urgency before the customer has evidence.

At the time Anders wrote the original note in late 2025, Salesforce was reporting roughly 9,500 customers live on Agentforce, while Agentforce represented around 1.3% of total revenue. Those numbers matter less as a static snapshot than as a signal. Adoption was visible, but revenue contribution was still early. For a strategic AI narrative, that creates internal pressure.

For current board papers, procurement teams should refresh those figures against Salesforce’s own Investor Relations materials before using them in negotiation strategy.

[Editorial outbound source placeholder: Insert the specific Salesforce earnings call, shareholder letter or investor presentation that confirms the latest Agentforce customer count and revenue contribution before publication.]

The buyer-side implication is plain. When a vendor needs a product category to grow quickly, the renewal becomes the cleanest distribution channel. Existing customers already have security reviews, procurement routes, executive relationships and dependency on the platform. That makes the AI upsell much easier to introduce than a net-new sale.

Again, this is not a claim that Salesforce AEs are acting maliciously. It is a claim that incentives behave exactly as designed.

If the account team is rewarded for early ACV commitment, the temptation is to sell future usage before actual usage is understood. If the customer is rewarded internally for AI progress, the temptation is to approve the commercial structure before the operating model is ready.

That is the Siebel problem in new clothes.

How the Agentforce Risk Manifests in Your Renewal

From the client side, Agentforce risk rarely arrives as a single dramatic line item. It usually arrives as a set of small framing moves that make the eventual quote feel natural by the time it lands.

The pattern is familiar to anyone who has managed a large Salesforce renewal: the commercial frame gets set before the commercial document appears.

• Bundled Expansions: Agentforce and Data Cloud can appear inside broader renewal or expansion narratives, making it hard to isolate what is driving the uplift. This is the SELA Bundle Trap: a broad commercial wrapper makes the customer feel they are buying flexibility, while the real effect may be reduced price transparency. If your organisation is considering this structure, first get clear on what a Salesforce SELA actually is and where AI-related commitments sit inside it.
• The Flex Mirage: Flex credits can look efficient because they promise adaptability. The risk is the Flex Credit Mirage, where flexibility is presented as control even though nobody has modelled the workload, trigger volume or governance path. Credits are not inherently bad. Unmodelled credits are the issue.
• Premature Commercials: Commercial structures are often discussed before the customer has proven value with baseline entitlements, pilots or included credits. This creates a negotiation built on aspiration rather than evidence. Once the quote lands, the conversation has already moved from “Should we?” to “How much?”

There is a fourth pattern that deserves its own name: Agentic Integration Gravity.

AI agents become more commercially sticky as they connect deeper into CRM records, service processes, sales workflows, customer data, case routing, knowledge content and downstream systems. The more useful the agent becomes, the harder it may be to unwind later. That is not just technical lock-in. It is process lock-in, data lock-in and behavioural lock-in.

This is why procurement should not treat Agentforce as a small add-on if the use case touches core revenue, service or operational workflows. The first commitment may look modest. The second order effect is what matters.

The Structural Divergence

The difference between traditional Salesforce licensing and Agentforce consumption pricing is not cosmetic. It changes the questions Finance, IT and Procurement need to ask.

Dimension	Traditional Seat-Based Pricing	Agentforce Consumption Pricing
Predictability	Spend is usually forecast from user counts, licence types, contract terms and known growth assumptions.	Spend depends on workload volume, agent triggers, automation design, data calls and adoption patterns.
Primary risk	Shelfware, over-licensing, renewal uplifts, unnecessary editions and unused add-ons.	Runaway usage, poorly governed triggers, unclear credit burn, premature commitments and weak value evidence.
Vendor incentives	Increase seat count, expand editions, attach clouds and protect renewal uplift.	Seed AI usage, bundle credits, expand consumption commitments and create future dependency.
Buyer control point	Licence assignment, user governance, role mapping and SKU rationalisation.	Workload modelling, trigger governance, usage telemetry, pilot design and commercial guardrails.
Best negotiation evidence	Current users, assigned licences, login activity, business unit demand and historical discounting.	Use-case economics, baseline usage, observed credit consumption, avoided labour or service cost, and scenario modelling.

This divergence also affects contract architecture. A standard agreement, SELA or AELA can each behave differently once consumption and AI commitments are added. Before treating these as interchangeable wrappers, compare the consequences of each in Salesforce commercial structures such as standard agreements, SELA and AELA.

The table points to a simple truth: the old problem was often paying for people who did not use what they had. The new problem may be paying for work you did not fully understand before you automated it.

That does not make consumption wrong. It makes it less forgiving.