The thirty-year capture gap.
For three decades, industrial distributors have known their reps were hearing things in the field that never made it back to anyone. The competitor someone's switching to next quarter. The fleet expansion that's about to hit. The pricing pressure on a top-50 customer. The reps could've told you — they did tell their dispatcher, their spouse, the person on the next bar stool — but the system that was supposed to record it was a typed form they didn't fill.
So the field intelligence problem stayed unsolved. Not because the data wasn't there. Because the capture step was broken. CRM tried to fix it with structured fields. Mobile CRM tried to fix it with smaller structured fields. Both produced the same number: 20–40% rep participation, generating data that managers didn't trust, surfacing intelligence too late to be useful.
"The field team is doing real work — visiting accounts, hearing what's coming, picking up on competitor moves — and 90% of it never makes it back to anyone."
This guide is about what the other 90% looks like once you actually capture it — what kinds of signals fall out of voice notes, how they get processed into something a manager can act on, and what happens to your business when you close the gap. The Bauer Built story below is the cleanest example we've got, but the pattern is the same at every customer once voluntary capture is real.
What "signal" means at Voze.
A signal is a structured claim extracted from an unstructured voice note. A rep says, in a 30-second parking-lot recording: "Just left Hendricks. Sam mentioned Samsara reached out again — third time in 60 days. Wants to talk fleet rates Thursday before quarter close." Voze pulls that apart into something the system can act on.
Out of those two sentences, Voze captures:
- Account: Hendricks — matched against the DMS account list.
- Contact: Sam — matched against the contacts known at Hendricks.
- Competitor mention: Samsara — third Hendricks mention in 60 days, which is a frequency pattern.
- Action signal: Fleet-rate conversation requested before quarter close — both an action item and a deal-risk indicator.
- Follow-up: Thursday meeting, calendared on the account.
Five structured records out of one breath of voice. None of them required the rep to type. None required them to pick from a dropdown. The rep talked the way they would've talked to their dispatcher anyway, and the structure happened in the cloud while they were driving to the next stop.
Voze classifies signals into four action tiers — Urgent, Action, Watch, and Pattern. Urgent means the data is telling us delay costs the deal. Watch means we're tracking. Each flag is held to a separate accuracy bar, and we publish how often each is right because a flag a manager can't trust isn't a flag.
That last point is the one most "AI for sales" products fail at. Calibrated confidence — not "we extracted competitor mention with 0.83 probability," but "Urgent flags fire only on patterns we've measured at 90%+ accuracy across customers." The flag has to mean something the manager can stake a Monday decision on. Otherwise it's noise wearing a confidence badge.
Bauer Built: 97 signals, ~15% closed.
Bauer Built, a commercial-tire distributor across the upper Midwest, ran Voze across a regional team for a year. Over that period, the system surfaced 97 distinct signals worth escalating — competitor-switch warnings, fleet expansion mentions, at-risk-account flags, product-gap hints from buyer conversations. Of those 97, roughly 15% closed as additional business inside 12 months.
The 15% number is worth sitting with for a second. It's not a conversion rate on inbound leads. It's a conversion rate on signals that came from inside the existing customer base — conversations that, in the old model, would've evaporated by Monday. A rep happens to mention something in a voice note. The system flags it. A manager acts on it. Some weeks or months later, that conversation has produced real revenue. Without Voze, the signal doesn't get captured. Without the capture, the conversation doesn't escalate. Without the escalation, the deal doesn't happen.
The other 85% didn't all evaporate. Some were patterns that didn't materialize this year. Some were earlier than we thought. Some led to saves rather than expansions — a top-50 account stayed instead of churned, which doesn't show up as new revenue but is real value. The 15% is the floor on visible upside, not the ceiling.
"Account spend is dropping at one of your top 50 customers. Your DMS shows you that. Voze shows you that your rep heard them mention a competitor twice in the last 60 days. You knew something was wrong before the quarter ended."
The pipeline a signal travels.
Here's the four-stop trip a piece of field intelligence makes from rep voice to revenue impact. Each stop is where most legacy systems lose the signal.
Stop 1 — Capture.
Rep records a voice note within 60 seconds of leaving the account. Voze transcribes, tags account/contact/product/competitor/follow-up, writes it to the account record. Failure mode in legacy systems: the rep never logged it, or logged it three days later with the texture stripped out.
Stop 2 — Classification.
The signal gets categorized: at-risk, expansion, competitive, product-gap, fleet-event, pricing-pressure, follow-up. Action-flag accuracy is tracked separately from classification accuracy, so a misclassified note is a different quality problem than a missed urgency call. Failure mode in legacy systems: even when capture happens, everything is "general note," so nothing rises.
Stop 3 — Surfacing.
Voze Web surfaces signals to the right person in time. Account owner sees their own. Sales manager sees the territory feed ranked by urgency. RevOps sees the cross-territory patterns rolling up. Failure mode in legacy systems: the data sits in a database nobody reads, or it goes to a dashboard nobody opens.
Stop 4 — Action.
Someone does something. Rep makes the call. Manager assigns a save play. RevOps escalates the at-risk account to the GM. The action gets logged back against the originating signal, so we can measure win rates by signal type over time and feed the loop. Failure mode in legacy systems: even when the signal lands, nobody owns it, so nothing happens.
Stop 4 is where most field-intelligence pilots quietly die. Capture works, classification works, signals surface — but nobody owns the action. Without a clear handoff and a logged outcome, the loop doesn't close, and after three months the team stops trusting the feed.
The dashboards that matter.
There are four views in Voze Web that do most of the work for sales leadership. Everything else is decoration. If you set up a pilot, these are the four you should be opening every week.
The territory signal feed.
Every signal captured in the territory in the last 7 days, ranked by urgency. This is the regional manager's Friday-afternoon home base. Click a signal, see the source note, decide whether to escalate.
The competitor velocity view.
How often each competitor name is being mentioned in your territory, by week, ranked by trend slope. Not the level — the slope. A new entrant showing up three times this week versus zero last week is the leading indicator of a switching cycle.
The at-risk account list.
Top-50 accounts where the conversation pattern has shifted — competitor mentions climbing, pricing-pressure flags firing, visit frequency dropping. Cross-referenced against the DMS for revenue trend. This is the leading-indicator view that beats the lagging-indicator quarterly business review.
The pattern roll-up.
Themes emerging across reps and territories that no single rep would surface alone. A fleet-customer cluster asking about EV transition. A product family losing share to a specific competitor across regions. This is the executive-sponsor view — the one that ends up in the board deck.
Sizing this for your team.
The 15% number from Bauer Built isn't a guarantee. It's a reference point. Here's how to size what Voze field intelligence is plausibly worth at your business — without doing the kind of hand-wave ROI math we'd both rather not.
- Count your top-50 customers. The signals worth escalating mostly live there. If you've got 200 accounts, you're really managing 50 — and that's where field intelligence pays.
- Pick an at-risk rate. What percentage of your top-50 customers are you worried about each year? At most distributors, this is 10–15%. So 5–8 accounts where conversation signal could materially change the outcome.
- Pick an average save/expand value. What's a typical save or expansion at a top-50 account worth in annualized revenue?
- Multiply. If field intelligence helps you save or expand a third of those at-risk accounts in year one, that's the year-one impact ceiling.
You'll arrive at a number that's a multiple of what Voze costs. That's not a sales claim — it's the math. The reason it works is that the cost of capture went from "20–40% of reps logging structured forms" to "30 seconds of voice per visit," and the cost of escalation went from "what someone remembered from Monday" to "the system already flagged it." The economics shifted because the capture step shifted.
Field intelligence works once capture stops being the bottleneck.
- A signal is a structured claim extracted from an unstructured voice note. Voze pulls account, contact, product, competitor, action, and follow-up out of a 30-second recording.
- Action-flag accuracy is tracked separately and held to a higher bar than classification. If Urgent flags start missing, we treat it as a quality issue and fix it.
- Bauer Built: 97 signals → ~15% closed as business in 12 months. Floor on visible upside, not the ceiling.
- Four dashboards do most of the work: territory feed, competitor velocity, at-risk account list, pattern roll-up.
- Stop 4 (action and ownership) is where most pilots leak. Capture is solved; ownership is the next bottleneck.