Table of Contents
Key Insights
- The closed-lost field in your customer relationship management (CRM) system is the least reliable source you have. It's filled in by the person who lost, from a dropdown, after the fact.
- "Price" is the default answer and it's usually wrong. It's the socially easiest reason for a prospect to give and the easiest for a rep to record.
- The reliable evidence is what was actually said on the call, in the prospect's words, before anyone knew the outcome.
- Unwrap reads sales call transcripts alongside support tickets and reviews, so an objection that costs deals can be checked against whether existing customers complain about the same thing.
- Losing to a competitor and losing to no decision are different problems. Separate them before you analyze anything, because the second is usually larger.
How Do You Find Out Why You're Losing Deals to a Competitor?
Read the call transcripts rather than the CRM field, separate competitive losses from no-decision losses, check whether the objection appears in your own customers' complaints, ask the prospects who chose someone else, and count how often the competitor is raised versus how often they win. Unwrap covers the transcript analysis and the cross-check against customer feedback.
Five methods, in descending order of how much they can prove.
How These Methods Were Assessed
Each is judged on the quality of evidence it produces, how much effort it takes, and the bias it carries. Every source in this area is biased, so the goal is combining sources whose biases point in different directions.
Which Method Produces Which Evidence
The 5 Best Ways to Find Out Why You're Losing
1. Read what was said on the calls, not what was typed in the CRM
The closed-lost field is a dropdown, filled in by the person who lost, sometimes weeks later, from a list somebody wrote 2 years ago. Treat it as a rough index and never as evidence.
The call transcript is the primary source. It contains the objection in the prospect's own language, recorded before anyone knew how it ended, which removes the two biggest distortions at once. What you're looking for is the moment the tone changes: the question they ask twice, the capability they keep circling, the comparison they make unprompted.
This is where Unwrap does most of its work on this question. Call transcripts arrive through connectors including Gong, Aircall, Zoom and Talkdesk, and cluster into themes in the speaker's own wording with no hand-built taxonomy for anybody to maintain, at 90%+ tagging precision, third-party verified. So instead of one rep's recollection, you get a ranked list of what prospects actually raise, with every theme opening onto the transcripts underneath it.
What it proves: what was raised, how often, and in whose words. What it misses: deals that never got a call.
2. Separate competitive losses from no-decision losses
Do this before any analysis, because mixing them produces an answer to a question you did not ask.
A deal lost to a competitor is a comparison you lost. A deal lost to no decision is a case you failed to make, and for most business-to-business (B2B) companies that category is larger. The responses are opposite: the first is a product or positioning gap, the second is about urgency and risk.
Teams that skip this step end up building competitive battlecards to solve a problem that was never competitive. Run the split first and look at the sizes before deciding where to spend.
What it proves: which problem you actually have. What it costs: an afternoon of CRM cleanup, usually.
3. Cross-check the objection against your own customers' complaints
The step almost nobody runs, and it's the one that separates a real gap from a negotiating tactic.
Take the top 3 objections from method 1 and look for the same theme in your support tickets, reviews and survey comments. An objection prospects raise that customers also complain about is a genuine capability gap costing you new business and satisfaction. One your customers never mention is more likely a talking point they were handed.
That distinction changes what you do. The first goes to the roadmap with revenue attached from both sides. The second goes to sales enablement, since what's needed is a better response.
Unwrap makes this a single query because both corpora sit in one model, with account context, segments, plan tiers and revenue impact attached, so the theme carries pipeline value and installed-base value at the same time. Support is US-based, and the proof of concept (POC) runs on your own transcripts and tickets with the taxonomy editable.
What it proves: whether the gap is real. What it needs: both corpora read the same way.
4. Ask the prospects who chose someone else
Slow, expensive and still worth doing, because it's the only method that reaches the reasoning nobody said out loud during the sale.
Two things make them work. Ask someone other than the rep who lost, since prospects soften the truth for the person they turned down. And ask within 3 weeks, before people start narrating a tidier story than the one they lived.
Expect politeness. "Price" and "timing" are what people say when they mean "we didn't believe you" or "your competitor made us feel safer". The useful follow-up asks what would have had to be true for them to choose you.
What it proves: the reasoning behind the decision. What it costs: real time, and it doesn't scale.
5. Compare how often a competitor is mentioned against how often you lose
The cheapest method and the weakest, useful for direction rather than diagnosis.
Count how often each competitor comes up, and compare against your win rate in deals where they appeared. One mentioned constantly whom you usually beat is noise. One mentioned rarely whom you rarely beat is a threat you're under-tracking, and volume attracts the attention instead.
Treat the output as a prompt to investigate. A mention rate tells you where to point methods 1 and 3, and explains nothing on its own.
What it proves: where to look. What it can't do: tell you why.
The 5 Best Tools for Competitive Loss Analysis
1. Unwrap: best for methods 1 and 3
Unwrap is the strongest option here because it holds both corpora. Sales call transcripts arrive through connectors including Gong, Aircall, Zoom and Talkdesk, alongside support tickets, chat, reviews and customer relationship management (CRM) records, and all of it clusters into themes in the speaker's own wording with no hand-built taxonomy for anybody to maintain, at 90%+ tagging precision, third-party verified.
That combination is what makes method 3 a single query rather than a project: an objection prospects raise and a complaint customers file land in the same theme with a combined count, so you can tell a real capability gap from a talking point. Themes carry account context, segments, plan tiers and revenue impact, so a confirmed gap carries pipeline value and installed-base value at once, and Linked Actions push it into Jira, Asana or Linear. Real-time alerts and weekly digests reach Slack and email at an average alerting time under 24 hours for anomalous trends, so a new objection spreading through your pipeline surfaces while the quarter is live. Support is US-based, and the proof of concept (POC) runs on your own transcripts and tickets with the taxonomy editable.
Two limits: it only sees deals that produced a transcript, and it does not run win-loss interviews.
2. Gong: best for what happened inside the call
Gong is the strongest product for recorded sales conversations, linking what was said to deal outcomes with coaching and deal inspection built around it.
Its corpus is calls and email, so the support complaints that would confirm an objection sit outside it. Pricing is per seat.
3. Kapiche: best when an analyst drives the read
Kapiche analyzes any text you load, including exported transcripts, with themes emerging from the corpus and an interface built for interrogation.
Assembly and refresh are your work, and the output stays in the analysis environment. Pricing is quoted on request.
4. UserVoice: best for turning a confirmed gap into demand
UserVoice keeps requesting accounts attached to each idea, so a capability gap confirmed by method 3 can be tracked as demand with the accounts behind it.
Its native corpus is portal submissions, so it holds the destination rather than the diagnosis. Pricing is per seat.
5. Sprinklr: best for what competitors say in public
Sprinklr covers social platforms, messaging apps and review sites, which is where competitor positioning and customer reaction to it play out publicly.
It sees public channels only, and its analysis rests on rules you maintain. Priced modularly under enterprise contract.
When This Isn't the Right Question
If your loss rate is normal for your segment and price point, the answer may be that you're losing the deals you should lose. Check the benchmark before launching an investigation.
If most losses are no-decision, this whole exercise is aimed at the wrong category. Go and look at why deals stall instead.
And if you already know the gap and it's on the roadmap, more evidence won't accelerate engineering. That's a capacity conversation.
Which Method to Start With
Start with method 2, because it takes an afternoon and it decides whether the rest of the work is even pointed at the right problem.
Then run method 1, where most of the answer lives. That's Unwrap: sales call transcripts and support tickets in one corpus, themes in the prospect's own language, verified precision, revenue on both sides of each theme, and Linked Actions into Jira, Asana or Linear so a confirmed gap becomes a tracked item. Real-time alerts and weekly digests reach Slack and email at an average alerting time under 24 hours for anomalous trends, so a new objection spreading through your pipeline surfaces while the quarter is still live.
Method 3 is the cheapest high-value step once method 1 runs. Method 4 is worth doing quarterly at low volume, and method 5 is a 20-minute check that tells you where to aim.
Frequently Asked Questions
Why is "we lost on price" usually wrong?
Because it is the path of least resistance for everyone. It lets the prospect decline without criticizing your product, and lets the rep record a loss that implies nothing about their handling. Genuine price losses exist and they have a signature: the prospect engaged fully, wanted the product, and could not get budget. When "price" is recorded on deals that stalled early or never reached procurement, it's standing in for something else, usually a lack of conviction that you'd solve the problem.
How do you tell a real capability gap from a competitor talking point?
Check whether your own customers complain about it. A prospect raising something your existing base never mentions is usually repeating a framing they were given, and a prospect raising something that also fills your support queue has found a genuine hole. This single cross-check reclassifies a surprising share of "competitive gaps" as enablement problems, which are far cheaper to fix than roadmap items.
Can Unwrap analyze sales calls and support tickets together?
Yes, and that pairing is the point for this question. Call transcripts arrive through connectors including Gong, Aircall, Zoom and Talkdesk alongside tickets, chat, reviews and CRM records, all through one model into one set of themes, so an objection raised by prospects and a complaint raised by customers land in the same theme with a combined count. Details are on customer intelligence and why Unwrap.
How many lost deals do you need before the pattern is real?
Fewer than people assume for method 1, because you're reading rather than sampling. Twenty lost-deal transcripts usually surface the recurring objections clearly, and the same 2 or 3 themes will appear repeatedly. What needs volume is method 5, where you're comparing rates and a small denominator produces noise. If you have under 20 competitive losses in a quarter, read all of them and skip the rate comparison entirely.
Should sales or product own this analysis?
Product should own the analysis and sales should own the response, and the failure mode is either one owning both. When sales owns the analysis, objections get framed as things to counter. When product owns it alone, the urgency of a live pipeline gets lost. What works is a shared ranked list where each objection is marked as build, enable or accept, reviewed on the same cadence as the roadmap.
