Case Study · 06Insurance · NationalEinstein AI · Copilot + Generative

$780K saved on an Einstein AI add-on package.

A national personal-lines insurer renegotiated an Einstein Copilot, Prompt Builder, and Conversation Insights bundle proposed alongside the Sales Cloud renewal. Credit-pool sizing reset to a defensible pilot scope, consumption guardrails added, $780K three-year reduction at 36% off the initial vendor quote.

$780K+
3-Year Savings
36%
Reduction vs. Initial Quote
3
Einstein SKUs Restructured
10wk
Engagement Length
The Situation

An AI proposal bundled into a renewal under pressure.

The client was a national personal-lines insurance carrier with a 3,400-seat Sales Cloud deployment supporting agency-channel sales. The Sales Cloud renewal was eight months out. Salesforce had proposed an Einstein AI package — Copilot for all 3,400 seats, Prompt Builder for 200 power users, and Conversation Insights for 600 inside-sales seats — bundled into the renewal at an attractive headline discount.

The internal champion for AI was the Chief Revenue Officer, who had seen Copilot demoed at a Salesforce industry event and wanted a path to deployment in the upcoming fiscal year. The package was priced against a credit pool sized for full-population, sustained use across all three SKUs. The proposed annual outlay for the AI bundle alone was $1.1M.

The Head of Procurement engaged SalesforceNegotiations with a specific brief: this was the firm's first material AI commitment with Salesforce; the contract structure would set a precedent for every subsequent AI add-on. The objective was not to refuse the deal — the CRO had executive air cover for the investment. The objective was to right-size the credit pool, build consumption guardrails, and avoid signing a multi-year commitment to a pricing model the firm did not yet understand.

Diagnostic Findings

The economics of the proposed credit pool.

Einstein generative SKUs price against a credit pool. Each Copilot turn, each Prompt Builder execution, and each Conversation Insights call consumes credits at a published rate. The diagnostic phase modeled the proposed pool against three usage scenarios.

Salesforce's pool was sized for full-population sustained use. The proposed credit pool would cover every one of the 3,400 Copilot-licensed users transacting at the high-end usage profile published by Salesforce. The firm's analogous pilot data showed actual adoption tracking at 40–60% of licensed seats and per-user consumption at the lower published profile. The pool was over-sized by an estimated 55–70% against realistic year-one consumption.

Annual commit with no carry-forward eliminated the value of right-sizing. The proposed contract structure committed the full annual credit pool with no carry-forward of unused credits and no true-down at renewal. A buyer who over-bought credits in year one would carry the over-buy forward indefinitely.

Prompt Builder usage requires deliberate operational rollout. The Prompt Builder SKU is the most expensive Einstein component on a per-user basis. The proposed deployment to 200 power users was speculative — no internal use cases had been defined, no prompt templates had been built, and no rollout plan existed. The SKU was a candidate for deferred deployment, not year-one commitment.

Consumption pricing principle

Every consumption-priced AI SKU should be sized against a realistic year-one adoption curve, not the vendor's full-utilization scenario. The buyer carries 100% of the over-sizing risk; the vendor carries 0%. Right-sizing at signature is the only practical lever.

Our Approach

How the AI add-on case was reset.

01

Usage scenario modeling

Three usage scenarios modeled against the proposed credit pool: low (40% adoption, low-end consumption), expected (60% adoption, mid-range), aggressive (80%, high-end).

02

Pilot data benchmarking

Six months of internal Copilot pilot data from a 200-user proof-of-concept pulled in to validate the realistic-consumption case.

03

SKU-by-SKU sequencing

Copilot deployed at signature; Conversation Insights deferred to month 6; Prompt Builder deferred to year 2 pending operational readiness. Each SKU sized against its deployment window.

04

True-up over true-down

Contract structured around a smaller initial commit with a documented expansion clause — buyer pays for additional credits when consumed, not when contracted.

05

Carry-forward language

12-month rolling carry-forward of unused credits negotiated into the contract. Eliminates the year-one over-buy risk.

06

Annual right-of-renegotiation

Mid-term right-of-renegotiation language for the AI SKUs specifically — preserves leverage against a pricing model that is still evolving.

Levers Pulled

Where the $780K came from.

Lever3-Year ContributionMechanism
Copilot credit pool right-sizing$340KPool sized to 60% adoption / mid-range consumption with documented expansion clause.
Prompt Builder deferred to year 2$220KSKU removed from year-one commitment pending operational readiness.
Conversation Insights deferred 6 months$110KDeployment phased to match Sales Operations rollout calendar.
Carry-forward credit language$60K12-month rolling carry-forward eliminates over-buy risk in year one.
Annual right-of-renegotiation on AI SKUsNon-cashPreserves leverage against an evolving pricing model. No immediate dollar value.
Mid-term price-cap on credit unit price$50KPer-credit unit price capped at the negotiated rate for the contract term.
What did not work

An initial request for a 50% discount on the Copilot per-credit rate was rejected; Salesforce holds the per-credit rate firm in early-adopter cycles. The lever moved to pool sizing, carry-forward, and SKU sequencing — where Salesforce had more concession authority — and the headline rate was conceded.

"

Our CRO wanted Copilot. The question was never whether to commit; it was how not to over-commit. We ended up with the AI capability we wanted, at a credit pool we will actually consume, with structural protection against a pricing model that's still moving.

Head of Procurement
National Personal-Lines Insurer
Timeline

10 weeks alongside the Sales Cloud renewal.

WEEK 1–2
Usage scenario modeling
Three scenarios modeled against the proposed credit pool. Pilot data pulled and validated.
WEEK 3
SKU sequencing case
Per-SKU readiness assessment. Copilot ready at signature; Conversation Insights at month 6; Prompt Builder year 2.
WEEK 4
Strategy memo and CRO/CFO sign-off
Strategy memo delivered. Target reduction quantified. CRO and CFO sign-off before vendor engagement.
WEEK 5–9
Negotiation execution
Four counter-cycles. Pool right-sizing accepted cycle two. SKU deferral and carry-forward closed cycle four.
WEEK 10
Close and close memo
AI bundle signed alongside the Sales Cloud renewal. Written close memo with consumption guardrail dashboard for ongoing monitoring.
Five Takeaways

What this Einstein deal establishes.

01

Vendor-proposed credit pools are sized for full-population sustained consumption.

Salesforce sizes Einstein credit pools against the high-end published utilization profile. Real-world year-one adoption rarely reaches that profile. Every buyer should model three scenarios and commit to the realistic middle, with documented expansion language for upside.

02

SKU-by-SKU sequencing is the most effective AI lever.

Deferring Conversation Insights by 6 months and Prompt Builder by 12 months produced $330K in savings on its own. Operational readiness — not vendor demand — should drive deployment timing.

03

Carry-forward language is the missing default in most Einstein contracts.

Without carry-forward, unused credits in year one are lost. With 12-month rolling carry-forward, the buyer's pool sizing risk drops materially. Salesforce will agree to this in the negotiation but rarely proposes it.

04

Annual right-of-renegotiation preserves leverage against a moving pricing model.

Einstein pricing is evolving. Per-credit rates have moved twice in the last eighteen months. Contractual right to re-negotiate AI SKU pricing mid-term preserves leverage when the rate environment shifts.

05

Headline per-credit rate is the weakest lever; pool sizing and structure are the strongest.

Salesforce holds per-credit rates firm in early-adopter cycles. Buyers who push hardest on rate concede ground on pool sizing and carry-forward. The reverse trade — accept the rate, win on structure — produces a better outcome by a substantial margin.

Einstein pricing is negotiable.

If Salesforce has proposed a Copilot, Prompt Builder, or Conversation Insights bundle, we model the realistic pool sizing within 30 days.

Contact Us →Einstein AI Negotiation

The Salesforce Negotiation Brief