How to run a win/loss analysis from your Gong calls

By Ahmet Nuri Ozcelik · Product Marketing Leader & GTM Engineer · 8 min read

Quick answer: Win loss analysis software extracts loss patterns from closed-won and closed-lost deals by reading every call transcript with consistent criteria, instead of sampling three deals out of three hundred via interviews. Callmine runs this analysis directly on top of your Gong calls — no data leaves Gong — and produces a comparable report across every deal in the period. The output is a defensible answer, not a story.

You can run win/loss analysis from Gong calls by combining call transcripts, deal outcomes, and consistent analysis criteria. Instead of manually reviewing a few lost deals, teams can analyze patterns across many won and lost opportunities to understand which objections, competitors, pricing concerns, and discovery gaps influence revenue outcomes. The practice itself is high-value. The damage in most win/loss programs comes from how the practice is usually executed.

Most win/loss programs are doing more harm than good — not because the question is wrong, but because the answer is built on three deals out of three hundred.

The standard model: a manager remembers three or four notable losses, calls a few buyers who agree to talk, runs structured interviews, ships a quarterly readout. The output is a narrative built on a handful of selected stories, usually skewed toward the most dramatic deals — the ones that lost loudly, not the ones that lost quietly. The roadmap and headcount decisions that follow get made against that narrative. And then everyone wonders why the next quarter looks the same as the last.

The problem isn't effort. It's sampling. You can't run a credible win/loss program on three deals out of three hundred.

This post walks through how to run win/loss analysis the way it should be run — across every closed deal in the period, against consistent criteria, on the data you already have. The output is a defensible answer, not a story.

Why most win/loss is broken

Three failure modes show up in almost every program I've seen.

It samples the loud deals

The deals that get reviewed are the ones the team remembers — the big losses, the strategic ones, the deals where someone has a strong opinion. The quiet losses, where a deal stalls and dies without drama, almost never get reviewed. They're often the most informative.

It runs on memory and CRM notes

Even when you do interview the buyer post-loss, weeks have passed. The story they tell is reconstructed and rationalized. The actual moment the deal turned — the specific objection that didn't get handled, the question that wasn't answered, the feature gap that surfaced in week four — is gone.

It surfaces stated reasons, not real ones

"We lost on price" is the most common loss reason in every win/loss program ever run. It's also almost never the real reason. Pricing is the easiest thing to point at, the easiest thing for the buyer to say, the easiest thing for the rep to write in the CRM. Real losses usually come from something earlier and quieter — a discovery question that wasn't asked, a stakeholder who never bought in, an integration concern that never got addressed.

What you actually need is an analysis that covers every closed deal, uses the same criteria across all of them, and reads the actual conversations — not the post-mortem story.

The structural difference between the two approaches:

The five-step playbook

Here's the analysis I'd run if I were starting from scratch tomorrow.

Step 1: Define your filters

The point of filters is to get a meaningful, comparable set of deals — not too narrow to be statistically empty, not too broad to be incoherent.

A reasonable starting point:

• ·Time period: Last 90 days. Older deals have stale memory and may reflect a different product or market.
• ·Outcome: Closed in the period — both won and lost. (More on why won deals matter below.)
• ·Segment: One segment per analysis. Mid-market and enterprise buy differently; running them together blurs the signal. If you don't know which segment to start with, run the segment with the most volume.
• ·Deal size: Optional. Filtering for ACV > $25k removes noise from low-touch deals where the loss reason is usually "wasn't a real opportunity."

What you're aiming for is a set of 50–200 deals. Below that, patterns are unreliable. Above that, you can usually segment further.

Step 2: Include the wins

Most win/loss programs analyze losses. That's a mistake.

The most informative comparison isn't "why did this deal lose," but "why did this deal lose when a structurally similar deal won." If you analyze only losses, you can identify objections — but you can't tell which objections actually lose deals, because every deal has objections. The closed-won deals are your control group.

Add closed-won deals to your filtered set. Run the same analysis on both. Read the output as a comparison, not a list.

Step 3: Write the analysis brief

This is the part most teams underestimate. The quality of your analysis is determined by the quality of your brief.

A good brief asks specific, structured questions. Here's a starting prompt I'd run on the first analysis:

For each deal in the selected set, identify:

1. The primary stated reason the deal closed the way it did, in the buyer's exact wording where possible.

2. The underlying reason — distinct from the stated one — based on the actual progression of the calls. Look for: discovery gaps, stakeholder problems, competitive losses, timing/budget issues, product gaps, integration concerns, or champion failures.

3. The stage at which the deal was effectively decided — not necessarily when it closed. Many deals are won or lost in week three but don't close for another six weeks.

4. Specific objections raised on the call, how they were handled, and whether the buyer's response indicated the concern was actually addressed.

5. For closed-won deals: what made the difference. What did the rep do or say that the buyer responded to?

Write a one-paragraph summary per deal. Then aggregate patterns across the set.

The structure matters. You're asking for both the surface-level answer (stated reason) and the structural one (underlying reason). You're locating the inflection point in the deal cycle. You're capturing language. And you're forcing a contrast against won deals.

This is the brief you'd give a senior analyst. Programmable call analysis tools will run it across every call in the set the same way.

Step 4: Read the report

The report you get back will have three layers, and they should be read in order.

Aggregated patterns. The headlines. What loss reasons cluster? Where in the funnel are deals dying? Which objections show up most? Read these first to get the shape of the data.

Stage breakdown. Where did the deal turn? This is the question most win/loss programs never get to, because they don't have stage-level visibility into when a buyer actually decided. Look for whether deals are dying in discovery, in technical evaluation, in procurement, or in the final stretch. The stage matters more than the reason.

Won-vs-lost contrast. What did the closed-won deals do differently when the same friction surfaced? If 60% of lost deals raised an integration concern and 60% of won deals also raised an integration concern, the concern isn't the problem — the response is.

What you're looking for, ultimately, is one or two patterns that explain a meaningful share of your losses and that you can do something about. Not ten patterns. One or two.

Step 5: Act on it

This is the step most win/loss programs skip, and it's why the next quarter looks like the last.

The deliverable from win/loss analysis is not a slide deck. It's a list of two or three changes that will be made in response.

If pricing surfaces as the dominant loss reason and won deals had a quantified ROI conversation in late stage that lost deals didn't, the action is: build the ROI conversation into the standard demo and pressure-test it next quarter.

If discovery gaps show up as the underlying reason, the action is: rewrite the discovery framework and roll it out with mandatory shadowing.

If a specific competitor is showing up disproportionately and your reps are improvising the response, the action is: build the battlecard, train it, and re-run the analysis next quarter to see if the pattern shifts.

The point is to close the loop. Run the analysis. Make the change. Re-run it. Watch the pattern move (or not).

Common pitfalls

A few traps to avoid.

Acting on N = 3. One quarter of losses doesn't make a pattern. Wait until you have at least 50 deals analyzed. The first analysis is a baseline, not a verdict.

Trusting CRM loss reasons. They're written in 30 seconds, often by the rep who lost. Use them as a hypothesis to test in the actual calls, not as a finding.

Stopping at frequency. "Pricing was raised in 60% of losses" is interesting. It's not actionable. The actionable question is: in the wins where pricing was raised, what happened differently?

Running it once and shelving it. Win/loss is a recurring pattern recognition exercise, not a quarterly project. Schedule it.

What you actually need to run this

You need three things:

1. 01Call recordings with transcripts. If you have Gong, you have this.
2. 02Deal context. Which deals closed-lost vs. closed-won, in which segment, at what value, in what time period. If you have HubSpot or Salesforce, you have this.
3. 03A way to read calls at scale. This is the part that didn't exist until recently. Programmable call analysis tools — Callmine being one — let you run a structured analysis like the one above across hundreds of deals at once.

You can also run this manually. It just means a senior person spends three weeks reading calls instead of three minutes reading a report. Most teams don't, which is why most win/loss programs are running on three deals.

The point isn't the tool. The point is the analysis. If you only take one thing from this post, take the brief in Step 3 and the act-on-it discipline in Step 5. The rest is operational.

How to run this in Callmine

Callmine reads your Gong calls — it doesn't write to or modify Gong. The workflow:

1. Pick the cohort in Gong. Filter to deals closed in the last 90 days: Stage = Closed-Won OR Closed-Lost, Deal owner = AE roster, Deal size > $20k. This is a Gong-side filter; Callmine reads whichever Gong call set you select.

2. Run the saved "Win/loss diagnostic" template. The prompt:

Across these calls, identify the primary loss driver from this taxonomy: status quo, no champion, budget freeze, competitor loss, packaging mismatch, priority shift. Score each from 0–10 on confidence. Quote the call moment that supports the driver. For won deals, identify the deciding moment instead. Output as a CSV with columns: deal, outcome, primary driver, confidence, evidence quote.

3. Output lands as a CSV in your Callmine workspace, exportable to Slack, Sheets, or a Notion doc. Run weekly — the output is comparable across periods because the prompt is held constant.

Because Callmine analyzes every call in the cohort (not a sample), the output is the defensible answer your QBR needs. And because the prompt is reusable, week 2 of running this is faster than week 1.

Closing thought

The best win/loss programs aren't the ones with the prettiest decks. They're the ones with the shortest gap between "we noticed a pattern" and "we changed how we sell." The bottleneck is almost never the data. It's the willingness to read the data honestly and act on what it says.

You probably have a year of calls sitting in Gong. They are telling you, every week, why your deals close and why they don't. The only question is whether you're going to listen.

FAQ

Can you do win/loss analysis from Gong calls?

Yes. Gong calls contain the buyer objections, competitor mentions, pricing concerns, decision criteria, and deal context needed to analyze why deals are won or lost. The analysis runs across the call transcripts; Gong itself remains the system of record.

How many calls do you need for useful win/loss analysis?

A small number of calls surfaces anecdotes; patterns become reliable across dozens or hundreds of calls. The defensibility of the answer is a function of sample size, not effort — which is why running across every closed deal in the period beats a sampled interview program.

Should win/loss analysis include won deals too?

Yes. Comparing won and lost deals identifies which objections, messages, competitors, and deal patterns are associated with different outcomes. Loss-only analysis can't separate "things that happen in losses" from "things that happen in all deals."

How is AI win/loss analysis different from manual call review?

Manual review is limited by reviewer time and is inconsistent across reviewers. Programmable analysis applies the same criteria to every call, surfaces recurring patterns faster, and stays comparable across periods because the prompt is held constant.

Run a win/loss analysis on your own calls

Take the brief from Step 3. Bring 100 of your closed deals. We'll run the analysis.

[ Try Callmine free — 100 calls, 30 days ]

No credit card. After your trial, your workspace continues on the Free plan.