Geofencing & Dwell-Time Verification for Field Sales Coverage
By Sameer Joshi
A GPS-based geofencing and dwell-time engine that tells a sales leader, for every assigned customer, every day, whether a field rep genuinely made the call, without relying on the rep's own word for it.
Platform
Manager Web Dashboard + GPS Field App
Duration
Proof-of-concept engagement
6
Distinct visit outcomes captured
<5 min
Minimum dwell time defining a real visit
2
Independent signals verified per visit
Project overview
The proof of concept showed that a coverage percentage alone is not a trustworthy metric: a rep can show 100% self-reported coverage and still be the most serious integrity concern on the team once GPS evidence is checked, while reps with legitimate tracking gaps were correctly cleared instead of penalized, proving the value of independent, evidence-based verification over self-reported coverage.
Platform
Manager Web Dashboard + GPS Field App
Duration
Proof-of-concept engagement
Type
Field Sales & Location Intelligence
Stack
8 technologies
The challenge
Field sales is one of the hardest functions to manage remotely. A sales leader assigns a rep a beat plan, a list of customers to visit, in order, with an objective for each stop, and then has almost no reliable way to know what actually happened out in the field until the numbers come in at month-end. Visit reporting is self-reported, coverage percentages can lie in both directions, drive-past visits go undetected, route and territory compliance is unverifiable, tracking gaps look identical to deliberate misconduct in naive systems, and manual audits cannot scale across a distributed sales force.
Visit reporting is self-reported. A rep files a check-in with an outcome note, and the manager has no independent way to confirm the rep was actually at that customer when they filed it.
Coverage percentages can lie in both directions. A rep can look like a top performer on paper while skipping real work, and a rep who did everything right can look like a poor performer because of a missed check-in or a dead phone battery.
Drive-past visits go undetected. A rep can pass within sight of a customer's shop without stopping long enough to actually make a call, and a simple "was the rep nearby" system will still count it as a visit.
Route and territory compliance is unverifiable. Without location evidence, managers cannot confirm reps are covering their assigned territory in the planned sequence rather than cherry-picking convenient stops.
No distinction between misconduct and bad luck. A tracking gap caused by a dead battery or a basement showroom with no signal looks identical, in a naive system, to a rep who deliberately went dark.
Manual audits don't scale. Spot-checking a handful of reps by phone call or surprise visit cannot cover a distributed sales force across dozens of territories every day.
What we set out to do
- 01
Build a geofencing and dwell-time engine that can tell a genuine customer visit apart from a drive-past, using GPS evidence alone.
- 02
Cross-verify that GPS evidence against the rep's own manual check-in, so agreement between the two signals, not either one alone, is what counts as a confirmed visit.
- 03
Give every disagreement between the two signals its own distinct, explainable status, instead of collapsing everything into a single pass/fail flag.
- 04
Make the underlying policy (geofence radius, minimum dwell time, coverage threshold) a business decision managers can see and tune, not a hidden black box.
- 05
Separate genuine integrity concerns (spoofed location, impossible travel speed, a check-in filed from a customer the rep was never near) from honest operational noise (tracking gaps, low battery, an unfiled check-in on a day the rep was clearly on site).
- 06
Fit into the sales organization's existing customer and territory data rather than becoming a second, disconnected system of record.
How we solved it
Two Independent Signals, Compared
Every assigned customer visit is evaluated against two signals that are collected completely independently of each other: the rep's GPS trail (geofencing + dwell time) and the rep's manual check-in (with an outcome note). Neither signal is trusted on its own: the system looks for agreement between them.
Key decision
Require two independent signals to agree, rather than trusting either the GPS trail or the self-reported check-in alone.
Result
A rep cannot fabricate a visit by filing a check-in alone, and a rep cannot be penalized on GPS data alone for a legitimate check-in the location system simply couldn't see (e.g., a signal-dead basement store).
Dwell Time as Continuous Presence, Not Just Proximity
Dwell time isn't a simple "was the rep ever inside the radius" check. The engine looks at the rep's full GPS trail inside a site's geofence and finds the longest continuous stay. If a rep leaves and comes back later, that counts as two separate stays, not one long one, so a rep can't inflate dwell time by looping past a location multiple times.
Key decision
Measure the longest continuous in-geofence stay, with a configurable tolerance for short GPS gaps within a single visit (a rep briefly losing signal mid-call shouldn't split one visit into two).
Result
The system reliably tells a genuine sales call (a real, continuous stay) apart from a drive-past (entered the geofence, left seconds later).
Six Outcomes, Not One Pass/Fail Flag
Because two independent signals can agree or disagree in different ways, every visit is classified into one of six distinct outcomes, each with its own meaning and its own business implication: Verified, No check-in, Unverified, Check-in without presence, Drive-past, and Not visited.
Key decision
Treat cases with no evidence either way as fundamentally different from cases with evidence of deliberate misreporting, rather than lumping both into one generic failed bucket.
Result
Managers get an explanation for every gap, not an accusation. Tightening the policy can flag a visit as unverified, but never as check-in without presence: only direct proximity evidence can produce that status, protecting reps from false claims.
A Device-Integrity Layer, Separate from Visit Scoring
Alongside visit-level verification, the system runs a separate integrity pass over each rep's daily GPS trail: mock-location app detection, physically impossible travel speeds between consecutive pings, tracking gaps, unusually late starts or early finishes relative to the working window, and low-battery context. These are reported as their own signals rather than silently folded into the coverage number.
Key decision
Keep device-integrity findings distinct from visit-coverage scoring, and always attach a plausible, human-readable explanation (e.g., "battery fell to 9%, a likely cause of this tracking gap") rather than a bare flag.
Result
A rep with a legitimate, explainable gap in their day is never treated the same as a rep whose trail shows a physically impossible 26 km jump in under two minutes.
A Live, Manager-Tunable Policy
The geofence radius, minimum dwell time, and the coverage percentage that triggers a flag are not fixed constants: they're parameters a sales leader can adjust, and every report recomputes from the raw GPS and check-in data the moment a parameter changes. Nothing is cached or pre-decided.
Key decision
Make the verification policy a visible, adjustable business rule rather than a hard-coded threshold buried in the system.
Result
A manager can see, in real time, the trade-off between a stricter policy (fewer false positives, more review) and a looser one (fewer reviews, more risk of counting a drive-past as a visit), and own that trade-off deliberately.
Measurable impact
6
Distinct visit outcomes captured, not just "visited / not visited"
2
Independent signals cross-checked on every visit
5
Real-world failure modes independently and correctly identified
<5 min
Minimum on-site dwell time to count as a genuine visit
Tech stack
What we learned
Field sales has always run on trust: a rep says a visit happened, and a manager mostly has to believe it. Geofencing and dwell-time verification don't replace that trust; they give it something to stand on. By requiring two independent signals to agree, and by giving every kind of disagreement its own honest, explainable status instead of a single pass/fail flag, the system turns a distributed, largely invisible field sales operation into something a manager can actually see, measure, and manage fairly.
- 01
Coverage percentage is meaningless without independent verification: it can be perfect and still hide the most serious problem on the team.
- 02
Dwell time (continuous presence, not just proximity) is what separates a genuine sales call from a drive-past.
- 03
Every disagreement between GPS and self-reported data deserves its own explanation, not a single generic failure flag, because "unprovable" and "contradicted by evidence" are very different situations with very different consequences for the rep involved.
- 04
A verification policy that is visible and tunable builds trust with both managers and field reps: nobody has to take a black box's word for it.
Want a system like this one?
Bring us the problem behind it - the constraint, the load, the deadline. You get an architecture opinion and a cost view from the engineers who built the work on this page, not a sales team.
Prefer we come to you? Fill in the form below.
Tell us where to reach you
Takes under a minute. An engineer reads it, not a lead-scoring tool.
More case studies
Ready to build something that matters?
We solve problems that don't have Stack Overflow answers. Let's talk.
Book a Discovery Call