All articles
9 min read

Win-Loss Analysis for a Small Phone Sales Team: 20-Call Audit

Win-loss analysis for a small phone sales team means tagging each lost or stalled call's reason from the recording, then counting. Twenty calls spot a common reason but cannot rank reps.

By Tej Pandya, Founder · Updated 9 Nov 2026

A sales manager wearing headphones listens to a recorded call while marking a paper grid beside a stack of call slips.

Key takeaways

  • Win-loss analysis for a small phone team is a count of reasons tagged from the call itself. Tag 20 lost or stalled calls with one of eight codes, set them beside what the rep logged, and act on the biggest code.
  • A rep's logged reason is a recollection, so measure the gap on your own calls. The audit sets the logged reason beside the reason the call shows and counts the matches.
  • If a reason sits behind one lost or stalled call in five, 20 tagged calls will show it at least once with about 99% probability (our model, assuming independent calls). Twenty deals per rep cannot separate a 40% win rate from 55%.
  • Keep undecided deals out of the win rate and give them their own reason code.
In this article
  1. 1What is win-loss analysis for a team that sells by phone?
  2. 2How do you run a 20-call win-loss audit on your own recordings?
  3. 3Does your audit pass?
  4. 4Why can't you rely on the lost reason your rep logs?
  5. 5How many lost or stalled calls do you need before the count means anything?
  6. 6What should you ask a lost lead when you call back?
  7. 7What do you change once the audit shows a pattern?
  8. 8What does it cost to record and summarise the calls with TalkEasy?
  9. 9Where to start this week
  10. 10Frequently asked questions

A rep who writes "lost on price" in the notes has recorded an opinion, not a finding. The recording of that call shows what the buyer actually said. A spreadsheet is enough to turn a month of those recordings into a count.

Win-loss analysis is usually taught for companies with a research budget. That means buyer interviews, a CRM (customer relationship management software) full of deal stages, and someone whose job is to run the programme. A small team of telecallers on one business number has none of that. That recording is the raw material those programmes pay for.

This page gives that team a method: eight reason codes, a 20-call audit, five questions for a call-back, and the arithmetic on how many calls you need before a percentage means anything. The arithmetic decides whether you act on a pattern or on luck.

What is win-loss analysis for a team that sells by phone?

Win-loss analysis is the practice of recording why each deal was won or lost, counting the reasons, and fixing the largest one. For a team that sells by phone, the evidence sits in the call. You tag one reason on each recording and compare the tags with what the rep wrote down.

Two numbers frame it. Win rate is wins divided by decided deals, where a decided deal is one you won or lost. Win-loss ratio is wins divided by losses.

Suppose a quarter closes with 20 won, 30 lost and 15 where the buyer never decided. The win rate is 20 ÷ 50 = 40%, and the ratio is 20 ÷ 30 = 0.67. Counting the 15 undecided as losses would give 20 ÷ 65 = 31%, which mixes two different problems into one figure.

Keep undecided deals out of the win rate and give them their own reason code, "timing or no decision", because the fix is different. Our lead conversion rate guide shows how the choice of denominator changes the answer.

How do you run a 20-call win-loss audit on your own recordings?

Pull your last 20 lost or stalled calls into a sheet. Tag each with one reason code taken from the recording, write the rep's logged reason beside it, and count both columns.

Audit 20 lost or stalled calls; use your decided-deal count, not these 20, for win rates.

Hands placing coloured tags on twenty paper cards laid out in a grid on a desk beside headphones.
Each recording gets exactly one reason tag, so the counting at the end is simple.
  1. 1
    List the calls.
    Use these columns: date, lead source, rep, reason the rep logged, reason the call shows, and the sentence or timestamp that proves it. Include more than one rep and more than one lead source. If one rep supplies all 20, the result describes that rep.
  2. 2
    Find the moment.
    Open the recording and find the sentence where the buyer's position became clear; a call summary or flagged objection can point you to that moment.
  3. 3
    Pick one code.
    Tag the reason still standing when the call ended.
  4. 4
    Write the evidence.
    Copy the sentence or note the timestamp.
  5. 5
    Count and compare.
    Count each column. Agreement is the number of calls where the rep's reason and the call's reason match, divided by 20.
CodeTag it when the call shows
PriceThe buyer compares cost or asks for a lower figure, and the call ends there
CompetitorThe buyer names another provider they will use or are comparing
Timing or no decision"Later", "not now", or "I need to ask someone", with no date agreed
Went quiet after quoteA quote or promise was given and later calls were not answered
Wrong fitThe need, location or budget sits outside what you sell
Slow or missed callbackThe buyer mentions waiting, or the first conversation followed a missed call
Unresolved objectionThe buyer raised a doubt about trust, quality or terms and the call ended without an answer
OtherNone of the above; write the buyer's words

If you already log outcomes with call dispositions (the outcome codes a call is logged with), our call disposition guide lists the outcome codes you can put these reasons beside.

Does your audit pass?

These three checks are our recommendation, and each has a pass mark.

  • Evidence check. Every tag has a sentence or timestamp. Pass: 20 of 20. Any tag without one becomes "Other".
  • Second-listener check. A second person tags the same 5 calls without seeing the first person's tags. Pass: at least 4 of 5 match. If fewer match, tighten the code definitions before going further.
  • Spread check. The 20 calls include at least two reps and two lead sources.

Then read your agreement figure. If the rep's note matches the call on fewer than 15 of 20 (three in four), stop treating the rep's reason field as a finding and use the call-based tags instead.

Why can't you rely on the lost reason your rep logs?

A rep's logged reason is a recollection, often the easiest one to write down, so measure the gap on your own calls before you trust it. The audit does this: it sets the reason the rep logged beside the reason the call shows and counts how often the two match.

A 2014 study in Industrial Marketing Management (Elsevier) examined key-account sales failures from the buyers' side. It studied large B2B accounts, not phone sales teams.

Our inference is that "price" is over-recorded. It sits outside the rep's control, and it is also the answer that ends a conversation politely for the buyer. That is why the audit compares the rep's note with the call. Your own agreement figure is the one that matters.

A recording shows what the buyer said to your rep. It does not show what they weighed afterwards or told colleagues. The audit is the first step, and the call-back below is the second.

How many lost or stalled calls do you need before the count means anything?

Twenty tagged lost or stalled calls are enough to find a reason that sits behind about one loss in five. They are not enough to rank reps or to call a win rate precise. For comparing two reps, you need about 40 decided deals each, and even then only a large gap is reliable.

Finding a common reason. Our model assumes each lost or stalled call independently has a 20% chance of carrying a given reason. The chance of seeing it at least once in n calls is 1 − 0.8ⁿ.

Tagged lost or stalled callsChance of seeing a 20% reason at least once
567%
1089%
1596%
2099%

Seeing a reason once is detection, not proof that it matters. We would act on a code that appears in about a quarter of the 20 calls (5 or more), then re-test it the next month.

Trusting a win rate. If you observe a 40% win rate, the true rate is plausibly anywhere in this range. It is a 95% Wilson score interval, a standard way to put a margin of error on a percentage from a small count. Our calculation, November 2026.

Decided dealsWins at 40%Plausible true win rate
10417% to 69%
20822% to 61%
502028% to 54%
1004031% to 50%

Ranking reps. Fisher's exact test asks whether a gap between two rates could be chance; a p-value, the chance of seeing a gap this big if there were no real difference, below 0.05 is the usual bar. Our calculation, November 2026:

  • 8 wins of 20 deals (40%) against 11 of 20 (55%) gives p = 0.53. That is noise.
  • 12 of 40 (30%) against 22 of 40 (55%) gives p = 0.041. That is a real difference.

So a monthly league table built on 20 deals per rep ranks noise. Use the tagged reasons to coach, and wait for 40 decided deals per rep before comparing win rates.

What should you ask a lost lead when you call back?

Ask five short questions in this order and treat the first answer as a lead, not the truth. A recording cannot tell you what the buyer decided afterwards, so call a sample of the 20 within a week or two and ask:

  1. "What did you decide in the end?"
  2. "What mattered most in that decision?"
  3. "Was there anything we said or did that made it harder to say yes?"
  4. "How did the other option compare on that point?"
  5. "What would make you look at us again?"

If the answer to question 2 is price, ask what they compared it with. As our judgement, price is often a stand-in for a competitor or a doubt about trust. Our cold call objection handling replies cover what to say next.

A salesperson in an office listens attentively on a phone call with a notepad and cup of chai on the desk.
A short, patient call-back lets the buyer explain the decision in their own words.

What do you change once the audit shows a pattern?

Act on the one code that appears in about a quarter of your 20 calls, make one change, and run a fresh 20 after a month. These changes are our recommendations:

  • Price: ask what the buyer is comparing against before offering a discount.
  • Slow or missed callback: set a callback rule and count missed calls weekly.
  • Unresolved objection: write and rehearse one reply to the top objection.
  • Wrong fit: move the disqualifying question to the start of the call.
  • Went quiet after quote: fix a dated follow-up before ending the call.
  • Competitor: if one name recurs, write a one-page comparison.

Once the habit holds, our voice of customer guide shows how to turn tagged calls into themes for the whole business.

What does it cost to record and summarise the calls with TalkEasy?

As of November 2026, TalkEasy Pro costs ₹999 + GST (Goods and Services Tax) a month. It includes 3 hours of talktime (minutes spent on calls) a day, call recording and history, unlimited users and contacts, team management and an analytics dashboard. Extra talktime is ₹49 + GST an hour. You can cancel anytime with no long-term lock-in.

The dashboard reports answer rates and missed calls, which feed the "slow or missed callback" code. Our AI Call Summaries list the customer's requirement, objections raised, recommended follow-up and call outcome within a couple of minutes. Our AI Call Insights flag objections, competitor mentions and pricing talk in the recording. You start from the flag and confirm it by listening at that point.

Keep the won or lost column in your own sheet, and keep the tagging decision with your manager. The summaries and flags point you to the moment to listen to, but you decide the code.

Customisation and integrations, including custom CRM integration, are on our quote-only Enterprise plan.

Where to start this week

Pull your last 20 lost or stalled calls and fill in the sheet. Talk to Our Expert and book a consultation to walk through the audit on your own recordings. Questions can also go to support@groweasy.ai.

Run the 20-call audit on your own recordings

Talk to our expert to walk through the audit on your own call recordings.
Talk to Our Expert

Frequently asked questions

No. Any team that wins and loses sales can run a smaller version. It needs a record of each outcome and its reason, not a research budget.
Win rate is the share of decided deals you won, shown as a percentage. Win-loss ratio is the number of wins you get for each loss, shown as a single number. Use the same denominator every month so the two stay comparable.
No published benchmark fits a small phone team, so compare your own rate month to month. Judge a change only once you have 40 decided deals in each period you compare.
No. A spreadsheet is enough for the first audit. A CRM helps later if you want to compare reasons by deal stage.
No. A recording captures what the buyer said to your rep, not what they decided afterwards. Call back a sample of lost leads to fill that gap.

Sources

  1. 1.Why are you really losing sales opportunities? A buyers' perspective on the determinants of key account sales failures, Industrial Marketing Management (Elsevier), 2014
  2. 2.TalkEasy AI Call Summaries, TalkEasy
  3. 3.TalkEasy AI Call Insights, TalkEasy
  4. 4.Our calculations: the detection probabilities (1 − 0.8ⁿ), the 95% Wilson intervals for a 40% observed win rate, and the Fisher exact tests (8/20 vs 11/20; 12/40 vs 22/40). All are computed from the stated counts, and you can redo them with any statistics package., November 2026

Keep reading

Get in touch — we'd love to help.

Talk to Our Expert