Product Positioning From Sales Calls: A PMM Playbook

By Ahmet Ozcelik, Product Marketing Leader & GTM Engineer — Published 2026-05-15

Quick answer: Product positioning from sales calls is the practice of deriving and testing positioning hypotheses — competitive alternatives, unique attributes, value, and market category — against the actual language buyers use in recorded sales and customer conversations, rather than against internal opinion or a handful of interviews. The method works because every step of April Dunford's positioning framework asks a question that the call corpus has already answered: what customers compared you to, which features they reacted to, what value they articulated in their own words. Running the analysis across hundreds of calls at once turns positioning from a once-a-year offsite into a hypothesis you can rerun every quarter.

Most positioning reviews produce a document everyone agreed to and no one believes. Product positioning from sales calls fixes the core defect: your buyers have already answered every question April Dunford's framework asks. The transcripts exist. The question is whether you've read them.

Why Positioning Workshops Produce Confident Answers to the Wrong Questions?

The premise of most positioning work is that positioning is an internal exercise — get the smart people in a room, walk through a framework, and converge on the right answer. The problem is the inputs.

When Product Marketing Alliance's State of Product Marketing data shows that most B2B SaaS teams refresh positioning every 18 months or longer and draw from only 4–6 customer interviews to do it, you have to ask: is the output a positioning statement or a well-formatted opinion? The answer matters because the CEO, sales leadership, and product will all push back on positioning they don't believe. If your defense is "we ran a workshop," you lose the argument.

April Dunford's framework in Obviously Awesome is structurally sound. Competitive alternatives, unique attributes, value, best-fit customers, market category — each step asks the right question. The problem is most teams treat those steps as brainstorm prompts rather than empirical questions. "What would customers use if we didn't exist?" is not a hypothetical — buyers have told your sales reps the answer on hundreds of calls. "What value do we deliver?" isn't something you vote on — it's what your best-fit customers articulated in their own words during discovery.

That's the dysfunction. The framework asks empirical questions. The process treats them as creative ones.

You can treat customer research drawn from sales calls as a nice-to-have, or you can recognize that your call corpus is a continuously updated research dataset. Every win call, loss call, demo, and renewal conversation contains a buyer answering at least one of Dunford's five positioning questions in their own language. The bottleneck isn't methodology. It's the cost of reading 300 transcripts to answer them honestly — which is why most teams skip it.

Mapping the Dunford Framework to Questions Your Call Corpus Can Answer

If you've read Obviously Awesome, you know the five components. What most positioning processes miss is that each one maps directly to a concrete question your call transcript data has already answered.

Competitive alternatives → "What would you use if you didn't buy us?"

Buyers name their alternatives in discovery and in objections. "We're also evaluating Salesforce," "We could probably do this in Excel," "Our engineers built something similar internally" — these quotes are in your transcripts. The call corpus answers this more accurately than any internal debate, because buyers name the real alternatives, including inertia, do-nothing, and internal builds that never appear in analyst competitive matrices.

Unique attributes → "Which capabilities surprised, relieved, or concerned the buyer?"

Listen for the moments in demos where a buyer's language shifts. "Oh, that's interesting," "Wait — can it actually do X?" or "That's going to be a problem with our security team." Those reactions are your differentiators and your risks, in rank order, weighted by frequency across the corpus.

Value themes → "What business outcome did the buyer say they're trying to achieve?"

This is where messaging mining from customer calls earns its place in the PMM workflow. Voice of customer isn't a research project — it's what buyers say in the first ten minutes of discovery, often more precisely than your website does. "We're losing ninety minutes per rep per day to manual CRM logging" is a sharper value driver than anything assembled in a positioning workshop.

Best-fit customers → "What firmographic and trigger patterns correlate with clean closes?"

When positioning lands cleanly, the deal moves differently — less objection handling, shorter sales cycles, fewer "let me check with legal" delays. The patterns in closed-won calls tell you who the ICP actually is, not who you wish it were.

Market category → "What frame did the buyer use before we offered one?"

Buyers arrive with a mental model. "We're looking for a better way to manage our analytics workflows," or "We need something like a project management tool but for compliance teams." These are category frames your buyers imposed on you. Whether your positioning reinforces or corrects those frames determines whether your messaging creates clarity or confusion.

Each of these is a single analysis prompt, not a six-week research engagement.

The Closed-Lost Corpus Is the Most Honest Positioning Data You Have

Won deals tell you what worked. That's useful, but incomplete. Buyers who chose you are diplomatically reluctant to revisit what almost didn't — they're invested in the relationship and not motivated to relitigate a closed decision.

Lost deals are different. Research on B2B late-stage buying behavior <!-- source: Gartner or Forrester on B2B buying behavior, or Bain win/loss research --> consistently finds that buyers articulate competitive alternatives and category reasoning more candidly in conversations where they didn't buy. That candor is precisely the signal positioning work needs.

A closed-lost corpus regularly surfaces categories you didn't know you were being placed in. Consider a company selling a revenue intelligence platform that discovers in its loss calls that buyers consistently compared it to BI tools — not because the product worked like a BI tool, but because every discovery call started with "reporting" and the category frame never shifted. The misframe is the finding. The fix isn't a new feature; it's a different conversation opening.

The right workflow pairs closed-lost calls with first-call discovery from the same segment. Closed-lost tells you where positioning broke down at the end of the evaluation. First-call discovery tells you what frame the buyer arrived with at the start. Together, you can pinpoint exactly where a competitive alternative entered the buyer's mental model or where a category misframe took hold and never recovered.

For the mechanics of structuring this systematically, the win/loss analysis against the closed-lost call corpus methodology covers the full approach.

One practical note: don't limit closed-lost analysis to deals where you lost to a named competitor. "Decided to deprioritize," "went with internal resources," and "couldn't get budget approved" are competitive alternatives too. Dunford's framework treats them explicitly that way. Your call corpus should reflect that.

A Worked Example: Rerunning a Dunford-Style Positioning Audit Against Gong Calls

This is the workflow for a PMM running a quarterly positioning audit. If your team uses Gong, this is the concrete path.

Step 1 — Filter to the right call set.

Start by segmenting Gong calls by deal stage: the last 180 days of calls filtered by HubSpot deal stage — closed-won, closed-lost, and active late-stage opportunities. Add a segment filter if your ICP has a defined tier (mid-market, North America, a specific vertical). This gives you the call population that reflects current market conditions, not deals from 18 months ago when your competitive set looked different.

Step 2 — Run the positioning prompt.

In Discera, run a custom prompt across all filtered calls in parallel:

"For each call, extract (1) any competitive alternative the buyer mentioned — including 'do nothing' and 'build it ourselves,' (2) the specific product capabilities the buyer reacted to positively or negatively, (3) the business outcome the buyer said they were trying to achieve in their own words, and (4) how the buyer described the category or problem space before we introduced our framing. Quote the buyer directly where possible."

Run this alongside the saved Competitive Intelligence and Messaging Validation templates for cross-reference. A corpus of roughly 1,000 calls finishes in about five minutes.

Step 3 — Review per-call findings and the roll-up summary together.

Discera returns a per-call view — which buyer said what, in which deal — and a roll-up executive summary that aggregates patterns across the full corpus. You're looking for convergence: when 40 of 180 closed-lost buyers named the same alternative, that's not anecdotal. When the value language in closed-won calls clusters around "time-to-value" but your website leads with "accuracy," that's your next messaging validation test.

Step 4 — Push to your team.

Schedule the output as a monthly DOCX to the PMM team and push the executive summary to your #pmm-positioning Slack channel. The positioning review meeting becomes evidence-based. When sales asks why you moved the positioning, you show them the quotes.

Making Positioning a Quarterly Hypothesis, Not an Annual Deliverable

Save the positioning prompt as a template. Schedule it to run monthly on the rolling 90-day call window. That's the full setup cost.

What you're watching for is drift. New competitive alternatives appearing in the corpus that weren't there six months ago. Shifts in value language — buyers who used to say "compliance" now consistently saying "audit readiness." New objections that cluster around a category frame you didn't intend to occupy. Any of these signals that something in the market has moved before it surfaces in your win rates.

The output feeds directly into messaging tests rather than static positioning documents. If the call corpus shows buyers consistently articulating a value theme you're not leading with on your website or in your sales deck, that's a test hypothesis — not a workshop opinion. Run the analysis, surface the language, test the message, rerun the analysis. Close the loop.

When the field shifts — a new entrant reframes the category, a large platform announces adjacent functionality, an analyst firm redraws the market map — the rerun cost is zero. You're not commissioning research or scheduling interviews. You're running the same prompt against the new call population. The broader product marketing use cases this supports range from analyst briefing prep to campaign messaging to sales enablement refresh.

Positioning stops being a deliverable and starts being a hypothesis. That's the shift worth making.

Limits of Call-Derived Positioning (and Where You Still Need Interviews)

Calls are a strong signal, but they overrepresent buyers who engaged. The segment you're invisible to — the buyer who encountered your category, formed an impression, and never booked a demo — isn't in your call corpus. Neither is the competitor who wins deals before you're ever evaluated.

Use call-derived positioning as the hypothesis, not the final answer. A focused round of 8–12 interviews with non-buyers or churned accounts tests whether the language patterns hold outside the population that already found you. The call corpus tells you exactly what to probe in those conversations; it doesn't replace them.

Quantitative survey work still matters for sizing. Calls tell you what buyers care about and how they articulate it. They don't tell you how many buyers share that profile across the addressable market. If you're using positioning to anchor investment decisions or category claims in external communications, pair the qualitative signal from calls with structured survey data.

The right division of work: calls for pattern recognition, interviews for language validation with non-buyers, surveys for market sizing. The call corpus makes the first step faster, cheaper, and more honest than a whiteboard session. That's not a small thing — it's where most positioning work collapses.

FAQ

Can I derive product positioning from sales calls if I don't use Gong?

Yes. Any call recording platform that produces transcripts works as the source. Gong is the most common in B2B SaaS at the $10M–$100M ARR range, and Discera connects to it directly, but the methodology applies to Chorus, Salesloft, Clari, or any transcript set you can analyze at scale. The binding constraint is transcript quality and volume, not the specific recorder you use.

How many calls do I need before positioning analysis is statistically meaningful?

A minimum of 30–50 calls per segment gives you enough signal to see patterns rather than anecdotes. At 100 or more calls per segment, competitive alternative frequency, value theme clustering, and category frame patterns become reliable enough to act on. Below 30 calls, treat the output as a directional hypothesis, not validated positioning — useful for generating the questions your next interview round should probe.

How is positioning analysis different from a standard win/loss program?

Win/loss programs typically focus on deal attribution — why did we win or lose this specific deal? Positioning analysis uses those same calls to answer a different question: what do these patterns, in aggregate, tell us about how buyers frame the problem and how our product fits into their mental model? The call set overlaps. The analytical prompt is different. Both are worth running; they answer different questions.

How do I get sales leadership to trust positioning that came from analyzing their own calls?

Show them the quotes. When the evidence for a positioning shift is "37 buyers in the last 180 days described the problem this way, and here are 12 direct quotes from your reps' calls," it's substantially harder to dismiss than a slide deck assembled in a PMM offsite. The per-call drill-down exists precisely for this conversation — pull the specific calls that support any claim you make in the positioning review.

How often should I rerun a positioning audit against the call corpus?

Quarterly is the right default cadence for stable markets. If you're in a fast-moving category — AI tooling, security, anything where a major platform announcement can shift the competitive landscape in weeks — monthly reruns on a rolling 90-day window are worth the negligible marginal cost. The scheduled analysis takes the decision out of it: the prompt runs, the summary lands in Slack, and drift surfaces before it shows up in win rates.

If your team uses Gong and you want to run this analysis against your own call corpus, Start a free trial at discera.ai.