Fictional data. Every customer, person, product, dollar figure and date on this page is invented to show the shape of the operating model. Nothing here is real customer information.
What we sold, and how much of it the customer has measured.
Fourteen accounts. Each one carries the value hypothesis the seller and customer agreed to, the telemetry since go-live, and the evidence tier behind every dollar claimed.
Accounts
Sold value, measured value, usage and renewal
| Customer | Stage | ACV | Measured of sold | Realized | Weekly active | AI score | Renewal | Next action |
|---|
Sold versus measured
Annual value, one scale
Measured and accepted Directional Sold, not yet measured
Forecast value is deliberately not drawn. A forecast is the sales case restated, not evidence.
How we would know it worked.
Eight measures a leadership team agrees to before the first account is scored. Each one names the question it answers, the formula, the threshold, and who owns it. Live values are computed from the portfolio.
Evidence tiers
What a dollar of claimed value is worth
Realization stages
Every account is in exactly one
Rules
Three rules the measures depend on
Only measured counts. Directional and forecast value appear on every page so nobody hides them, but the realization rate uses accepted measurements only. The customer's finance or executive owner accepts a number, or it is not realized.
The promise is the denominator. Sold value is the sum of the objectives written into the success plan at signature. If the seller promised it, Valerie tracks it. A promise that was never written down cannot be measured, which is the point.
Both sides keep a ledger. Commitments belong to us or to the customer, with a date. A late customer commitment is a risk to the value, not an excuse for it.
Weekly value realization rollup.
Generated from the portfolio, in the order a leadership team reads: the headline, renewals inside 180 days, accounts to watch, late commitments, product themes.
About Valerie
The work after the sale, in code.
Valerie is a value realization engine I built on my own time. It is a concept prototype with fictional data, and it is where my operating model for enterprise AI lives in a form you can click on.
The problem
Enterprise AI adoption is an operating model problem before it is a deployment problem. Most organizations can get a pilot started. Fewer can say whether the customer is technically ready, whether adoption is happening in real workflows or in isolated experiments, and where measurable value is showing up. Almost none can answer the question that decides the renewal: of what we sold them, how much did they get?
The thesis
The value hypothesis the seller and customer agree to at signature is the contract that matters. Write it down as objectives with a baseline, a target, an owner and a dollar figure. Then hand it, unchanged, to the people who deliver. Every week after that, the only honest question is how much of that hypothesis has become an accepted measurement.
I ran a version of this at Salesforce and called it the Value Handshake: when a Success Plans opportunity neared close, presales and sales relayed the use cases, personas and expected value to onboarding, the CSM and renewals, so delivery started with the expectation the customer bought instead of a fresh discovery. Valerie is that relay with the arithmetic attached.
What I built
- A data model where every account carries what was sold (products, quantities, ACV, term, success plan), the value hypothesis (objectives with baseline, target, current, promised and claimed dollars, and an evidence tier), six weeks of usage telemetry, commitments on both sides, and the events that connect them.
- Derived measures, not hand-entered scores. Realization rate, evidence coverage, time to first measured value, value multiple, entitlement utilization, objectives measured, renewals without evidence, late commitments. Adoption and value confidence are computed from telemetry and the ledger.
- Three lenses on one model. The success team, the seller and the customer's AI owner see the same account and ask different questions of it. The questions are answered from the data.
- A weekly rollup generated from the portfolio in the order a leadership team reads it.
Decisions I made on purpose
- Only measured value counts toward realization. Directional and forecast value are shown everywhere and counted nowhere.
- The sold value is the denominator. A generous seller makes their own account look worse. That is the right incentive.
- Commitments have an owner and a date, and the customer's late commitments are visible to the customer.
- Every score can be opened. A score is a starting point. The account team still needs the evidence behind it.
Limitations
The data is invented and the thresholds are opinions. There is no integration, no authentication, no history beyond six weeks of usage, and no way to edit an account in the page. The value formulas per objective (hours saved times loaded cost, tickets deflected times cost per ticket) are assumed, not shown. A production version would need those formulas agreed per objective and a place for the customer to accept a number.
Where it goes next
- Objective-level value formulas, editable, with the customer's acceptance recorded as an event.
- A monthly history so realization rate has a trend, not a snapshot.
- An account brief generator: one page per account for the renewal conversation.
- Benchmarks across the portfolio: median time to first value by industry and success plan tier.