Case study · Customer name withheld
Eleven days of outbound calls, checked against the record.
A B2B software company's outbound sales team: 51 reps on Salesloft. Every dial was matched to its carrier record and every recording was read twice. Here is what the audit found, and what it cost.
- Customer
- Customer name withheld
- Window
- Sep 18 – Oct 2, 2026 · 11 working days
- Source
- Salesloft, read-only
- Audit
- KPI Shield
01
The question for every logged call: could it have reached anyone?
Like most outbound teams, this one's activity reports counted what the dialer logged: dials, calls and conversations. A dial cut while it was still ringing counts the same as one that rang out. A call dispositioned “Connected” counts as a conversation whether or not anyone picked up.
The audit took two weeks of that activity and checked it against what the phone network and the recordings actually show. Nothing in Salesloft changed, and no rep installed anything.
02
Every dial, every recording, twice.
| Reps | 51 | on one outbound team |
|---|---|---|
| Dial attempts | 44,506 | placed by the dialer, each matched to its carrier record |
| Logged calls | 41,656 | the activity the reports count |
| Logged conversations | 9,046 | calls dispositioned as connected |
| Recordings | 4,882 | 232 hours of audio, transcribed and read twice |
| Window | 11 days | working days, Sep 18 – Oct 2, 2026 |
Each dial was matched to its carrier record: did it ring, for how long, who hung up, and what was logged. Each recording was transcribed and read twice for what it actually held. A finding was kept only when the rep's own logged activity showed it.
03
One call in twenty-two had no real chance of reaching anyone.
of 41,656 logged calls: rang out in under 6 seconds, sat inside bursts and rapid redial runs, or went to malformed numbers.1
The typical rep sat at 2.3%:2 misdials and early hang-ups happen to everyone. The rest was concentrated. The three highest reps were at 42%, 35% and 25%.
| Typical repthe floor: median over reps with 100+ logged calls | 2.3% | |
|---|---|---|
| Whole teamof 41,656 logged calls | 4.6% | |
| Highest repof their logged calls | 42% | |
| Secondof their logged calls | 35% | |
| Thirdof their logged calls | 25% |
calls logged as connected conversations were contradicted by the carrier record or the recording: no answer, a busy line, a hang-up during ringing, or a voicemail greeting.
Measured· Sep 18 – Oct 2, 2026What it looked like
Five patterns the audit found, redrawn with synthetic timings in the same shape. No customer row, number or time appears here.
A run of 42 dials in 30 minutes
About one dial every 43 seconds for half an hour, each counted as a dial. Dials inside a run like this had no real chance of reaching anyone.
42 dials in 30 minutesMeasuredSynthetic data- Inside the run: no real chance of reaching anyone
3 dials to the same number within 63 seconds
One number, three counted calls in about a minute. A redial seconds after the last attempt has little chance of a different result.
3 dials in 63 secondsMeasuredSynthetic data55 calls logged "Connected" that never connected
The disposition said a conversation happened; the carrier record said nobody picked up. Each one entered the funnel as a conversation.
55 callsMeasuredSynthetic dataCalls logged Connected, with what the carrier record shows Call Logged as What the carrier record shows Wed 10:12 AM Connected No answer · rang 29 s Wed 10:31 AM Connected Busy signal Wed 1:48 PM Connected Rep hung up after 4 s of ringing One malformed 9-digit number dialed 4 times
A number with a missing digit cannot ring. The dialer fails it in zero seconds, and the attempt can still be logged as a call.
1 number, 4 dialsMeasuredSynthetic dataNumber as stored(312) 555-0199 digitsEach attempt failed in 0 s and was logged “No Answer”: 4 dials on the rep's day.
49 dials to 25 malformed numbers
The first dial to a bad number is a list problem. The dials after it are activity that could not reach anyone.
49 dials to 25 numbersMeasuredSynthetic data- First dial to the number: a list problem
- Dialed again after it had failed
“Measured” on each panel is the pattern and its count; the drawing uses synthetic timings.
04
What it cost: one number measured, one estimated.
Kept deliberately narrow. Only the 10 flagged reps, and only the part of their calls above the typical rep's floor of 2.3%. The share of a rep's logged calls that could not reach anyone is the share of the pay for that activity that bought no chance at a conversation. Pay assumed at a $50,000 OTE.
$63,000
a year if the rate holds3 · about $2,400 every two weeks
- The three highest reps
- ≈ $47,000 of it
- The other 41 reps
- are not in this number.
What that number leaves out
These cannot be measured from the dialer. Each figure rests on the assumption next to it, meant to be replaced with a team's own.
| Cost | Assumption | A year |
|---|---|---|
| Manager time | 1 hour a week per flagged rep at $50 an hour | ≈ $25,000Estimate |
| Burned leads | about 50,000 low-quality dials a year at this rate, at $1 of list and enrichment spend each | ≈ $50,000Estimate |
| Rep turnover | 2 of the 10 flagged reps leave or are managed out in a year, at $20,000 each to recruit and ramp | ≈ $40,000Estimate |
| Tooling | seats and dialer minutes spent on dials that could not connect | ≈ $3,000Estimate |
| Decisions made on bad numbers | connect rates, forecasts, list purchases and hiring decided on activity that did not happen | not priced |
| Measured, above | the 10 flagged reps at a $50,000 OTE | $63,000Measured |
| Together | with the assumptions doing most of the work | About $181,000Estimate |
770–800
people, roughly, the questionable dials would have reached at the team's 40% reach rate, had each of them had a real chance.4
Estimate· Sep 18 – Oct 2, 202605
What a weekly check changes.
The audit ran once, on eleven days. KPI Shield runs the same checks every week on the week before, for every rep, so the patterns above stop being a one-time discovery.
A short list per manager
Each finding with the evidence behind it and the other explanations already checked. Nothing to build or export.
The team's own rules, learned once
A shared main line, a support rep, a pattern the team considers fine: suppressed with an owner, a reason and an expiry, so the next run already knows.
Evidence, not a score
No rep score and no black box. A finding is a list of dials, logs and recordings a manager can open and a rep can answer.
06
Keep only what the record proves.
Two sources, for every call. The carrier's record of the dial (did it ring, how long, who ended it, what was logged) and, where a recording exists, two independent reads of it.
Set aside, not counted. Outages, busy lines, redials bundled by the dialer and voicemail pickups. A finding stands only when the rep's own logged activity shows it, and every number opens to the dials behind it.
What this page shows. Aggregates only. Rep-level results went to the customer and nowhere else. Customer name withheld. Figures marked measured are counts from the audit's records; figures marked estimate rest on the assumption printed next to them.
Notes
- 1Window: Sep 18 – Oct 2, 2026, 11 working days, one outbound team of 51 reps on Salesloft. Measured figures are counts from the dialer's records, the carrier records and the recordings in that window.
- 2The typical rep's floor is the median share over reps with 100 or more logged calls in the window.
- 3A year at the rate measured in the window. Two weeks is a short window: any one rep can run high or low in a fortnight.
- 4Reach rate: the share of the team's logged calls that reached someone in the window, applied to the questionable dials. An estimate, not a count.
Your numbers
Price it with your own inputs.
The same method as above, on team-level inputs. The defaults come from this one audit; your results will differ, and only your own calls can measure them.
Pay for activity above the floor
Estimate$59,000
a year · about $2,300 every two weeks
About 10 reps above the floor, at roughly 14% of their logged calls each. Their share above 2.3%, times their pay.
- Manager time 1 hour a week per flagged rep at $50 an hour
- $26,000
- Rep turnover 1 in 5 flagged reps a year at $20,000 each
- $40,000
Assumes every rep logs about the same number of calls, so the flagged share sets how concentrated the cost is, not how big. On the audited team this arithmetic gives $59,000; the figure measured rep by rep was $63,000.
Measure it on your own calls.
Book a demoBook a demo
Find out what your team's record shows.
A walkthrough on demo data, then what your dialer's data lets us check. The same audit is available for your last 30 days.
- The review queue and one finding's evidence, end to end
- What your source supports: the capability report
- Trial, Integrity Audit or a plan: what fits your team
Rather try it yourself? Start a 14-day free trial.