Value Selling
Personalized AI Value Proposals: How to Build One Without a Value Engineer

A buyer who asks "what is this worth to us?" is no longer asking for a slide. They want a figure built from their own situation that a colleague in finance can open and check. A personalized value proposal is the answer to that question, and AI has made it fast to produce one. Fast isn't the hard part. The hard part is producing one the buyer can trust on a deal where you don't have a value engineer or a long record of similar wins.
46%
of software decisions now include finance, up from 31% a year earlier. Buyer self-report; it describes the market, not any vendor's customers.
G2 2026 Buyer Behavior Report, online survey of 1,038 B2B software decision-makers, June 2026 (published July 22, 2026)
That is the setting for every value proposal now. More of them will be read by someone who wasn't on the call and has no reason to take a vendor's number on faith.
Key takeaways
- 1A personalized value proposal is a buyer-specific projection, not a result. Label it that way.
- 2AI can draft one from almost anything. What makes it credible is the model underneath: approved claims, visible conditions, and arithmetic that gives the same answer every time.
- 3You don't need a value engineer or a long deal history to start. You need your own product material, someone willing to approve claims, and a buyer willing to correct your starting assumptions.
- 4Write one case for the whole buying group, with one set of totals, and show the range rather than a single number.
What a personalized value proposal is, and what it isn't
The terms get used loosely, so here is how we use them. The difference that matters is the question each one answers and who it is built for.
Four terms, four questions
Term
ROI calculator
The question it answers
Roughly what could a tool like this return for a company like ours?
Usually public and self-serve. Generic by design.
Term
Personalized value proposal
The question it answers
What could this purchase be worth to this buyer, using their situation?
Built during the sales conversation for one account. A projection, not a promise.
Term
Business case
The question it answers
Can the buyer justify this spend to everyone who has to agree?
The fuller document: problem, proposed change, benefits, costs, assumptions, evidence, and open questions.
Term
Value model
The question it answers
What does our product change, for whom, under what conditions, and on what evidence?
The seller's reusable knowledge. Every proposal and case is a projection of it.
Most teams have the first three in some form. The fourth is usually missing. The reasoning behind why customers buy gets rebuilt on each deal by whoever has time, and it leaves when they do. That is why two reps can give two buyers in the same situation very different numbers.
Two ways AI can personalize the proposal
Every AI value proposal tool has to answer one question: what does the AI ground its numbers in? There are two broad answers.
The first grounds the proposal in history. It ingests CRM fields, call transcripts, and past closed-won and closed-lost deals, then matches each new opportunity to a value framework that a value or sales engineering team configures and maintains. As deals close, outcomes feed back in. This can work well for a company that already has a large deal record and a team to own the framework.
The second grounds the proposal in a value model built from your own material. That is the approach Serre takes.
What each approach needs before the first proposal
| Question | History and framework first | Value model first (Serre) |
|---|---|---|
| Starting material | History and framework firstCRM data, call transcripts, and a record of closed deals | Value model first (Serre)Material you already have, such as a solution brief, discovery notes, or an existing calculator |
| Who sets up the logic | History and framework firstA value or sales engineering team configures the framework | Value model first (Serre)Serre proposes claims, conditions, and evidence; a person on your team approves what enters the library |
| When it gets useful | History and framework firstImproves as closed deals accumulate | Value model first (Serre)Usable on the next deal, and reused on every deal after it |
| How a buyer's numbers get in | History and framework firstPulled from CRM fields and transcripts | Value model first (Serre)Guided Discovery proposes a starting hypothesis; the buyer's corrections become the inputs |
| How the math is done | History and framework firstVaries by product | Value model first (Serre)Fixed formulas: the same inputs give the same result, shown as conservative, expected, and upside |
| Who it's written for | History and framework firstVaries by product | Value model first (Serre)One case for the whole buying group, with one set of totals |
How the value-model approach works
From discovery to a business case
Guided Discovery
Serre enriches your company, your solutions, and the prospect with context for value selling. The result is a starting hypothesis about what this buyer might be trying to change. It is not buyer confirmation.
Value model
Approved claims, the conditions each depends on, and the evidence behind them. The rep applies this shared logic to the buyer and replaces estimates with what the buyer confirms or corrects.
Business case
One projection of the model for this buyer: visible assumptions, fixed arithmetic, a conservative, expected, and upside range, and one set of totals for the whole buying group.
Guided Discovery: research becomes a hypothesis, not an answer
Guided Discovery enriches three things with context for value selling: your company, your solutions, and your prospects. Enrichment supplies context and hypotheses. Your team reviews them, and the buyer confirms what is true.
Here is what that looks like concretely, using the fictional legal-software example from our sample case. The initial account context suggests the firm wants to collect fees sooner, that its finance lead cares about cash flow and the time spent resolving billing issues, and that the seller's product supports billing preparation and follow-up. That is a hypothesis and a reason to ask better questions. It doesn't establish where this firm loses time or what a change would be worth.
So the rep asks where the process slows down, who does the work, and what the team has already tried. In this example, the buyer points to repeated invoice preparation work rather than slow collections. That correction changes the case. A general collection-speed claim would have missed the change actually under discussion, so the inputs now come from the preparation work the buyer described.
The value model: no value engineer required
You point Serre at material you already have. It proposes claims, the conditions each one depends on, and the evidence behind them. A person decides what enters the library. That person might be a value engineer if you have one, or an enablement lead, a sales manager, a product marketer, a founder, or an AE. Approving a claim is a judgment call, not a specialist skill set.
Because the model starts from your material rather than from your deal record, it doesn't wait for a history of closed deals before the first case. It also lasts. The reasoning a rep earns on one deal stays with the company when that rep moves on.
The business case: one projection, one set of totals
Every number in the case comes from a formula with inputs you can read. The same inputs give the same result, and an approved case is frozen, so new evidence never silently changes a number a buyer already has. The delivered case shows the annual benefit three ways: conservative, expected, and upside. Where stress-testing is available, those come from running the case across the plausible range of each input.
The case is written once for the whole buying group. Operational, budget, and financial readers can each see the measure they care about, but those views describe the same benefit and are never added together.
That matters more than it sounds. Gartner found that tailoring for buying-group relevance had a 20% positive impact on buying-group consensus, while content focused on individual-level relevance had a 59% negative impact on it (Gartner, May 2025). Those are relative effects on consensus, not effects on whether a deal is won. They are still a good reason not to send four versions of the same proposal to four people.
A worked example
Illustrative projection, fictional company
This continues the fictional invoice-preparation example from our sample case. The inputs are invented for explanation. It is not a customer result or a benchmark.
Continuing the discovery above, the buyer confirms that five people prepare invoices. The rep estimates that the product would release two hours a week for each of them, over 46 working weeks, at a blended rate of $45 an hour.
- Hours released: 5 people × 2 hours × 46 weeks = 460 hours a year
- Capacity value: 460 hours × $45 = $20,700 a year
The weakest input is the two hours a week. It is an estimate, not something the buyer confirmed. To show what it does to the answer, hold everything else fixed and let it run from one to three hours a week:
- Conservative (1 hour a week): $10,350 a year
- Expected (2 hours a week): $20,700 a year
- Upside (3 hours a week): $31,050 a year
That spread is the honest answer at this stage. It also tells the rep what to ask next: how long invoice preparation actually takes today. This hand calculation varies one input for illustration. A real case would vary every uncertain input, not just this one, so it would show different figures.
Hours recovered aren't cash saved
All three figures are capacity value: time released and valued at a labor rate. They become a cash saving only when the buyer names a capture action and an owner, such as not filling an open role. Without that, present the figure as capacity and let the buyer decide what the time is worth.
A checklist before you send a personalized value proposal
- Can you name where each number came from: the buyer, an estimate, or an outside source?
- Has the buyer confirmed or corrected the inputs that move the answer most?
- Is it labeled as a projection?
- Does it show a range, with the conservative figure in view?
- Are time savings labeled as capacity unless a cash capture action exists?
- Is there one set of totals, with no benefit counted twice across operational, budget, and financial views?
- Would the same inputs give the same answer if someone else ran it tomorrow?
- Could your champion explain it to a colleague who wasn't on the call?
Where this helps and where it doesn't
It helps when your reps keep rebuilding the same value argument for different accounts, when there's no value engineer to queue for, or when a new product or segment means there's no deal history to learn from yet. It also helps when buyers ask for numbers they can check.
It doesn't help if nobody on your team will approve claims. Without that, the library is storage. It doesn't replace discovery: research gives you a hypothesis, and only the buyer can confirm it. And a better proposal can't create budget, urgency, or fit where none exists. A useful case can lead to yes, no, not yet, or a request for better evidence. Each of those is a decision.
If your team already has a large deal record and a value engineer who maintains a framework, a history-based approach may suit you well. The two aren't exclusive. The question is what your next proposal can stand on today.
Next step
If you want to build the value model your proposals project from, get started. If you'd like to talk through how it would work for your team, contact sales.
Frequently asked questions
What is a personalized value proposal?
It is a buyer-specific projection of what a purchase could be worth to one account, built from that buyer's situation during the sales conversation. It sits between a generic ROI calculator and a full business case.
What is an AI value proposal?
It is a personalized value proposal that AI helps produce: drafting claims, suggesting inputs, and writing the narrative. The credibility still depends on what the AI works from and whether a person reviews the result. In Serre, AI proposes and drafts, a person approves what enters the library, and fixed formulas do the arithmetic.
Do I need a value engineer to build one?
No. Serre proposes the structure, drafts claims from your own material, and asks what each claim depends on. Someone on your team approves what enters the library. If you have a value engineer, they are the natural person to own it.
Do I need a history of closed deals?
Not to start. The value model begins from material you already have, such as a solution brief, discovery notes, or an existing calculator. Each later case starts from what the team has already approved.
How is a value proposal different from a business case?
A value proposal answers what the purchase could be worth to this buyer. A business case is the fuller argument the buyer uses internally: the problem, the proposed change, benefits and full costs, assumptions, evidence, and open questions. In Serre, both are projections of the same value model.
Should I send a different version to each stakeholder?
Generally not. Write one case with one set of totals, and let each reader find the view that matters to them. Gartner's research links individual-level tailoring to a negative effect on buying-group consensus.
Is the number in a value proposal a promise?
No. It is a projection based on stated assumptions. Show the range, label estimates as estimates, and separate capacity from cash.
Turn every deal into a defensible case for change.
Value engineering for every seller. A person approves what enters the library.

