Ghost Visits Are Costing FMCG Brands Millions. Here's How GPS and AI Are Finally Killing Them

A regional sales manager in Lahore showed me his dashboard last March. 312 outlet visits logged that week by his team of 14 reps. Looked great on paper. Then we pulled the GPS logs against the visit timestamps. 89 of those visits — 28.5% — never actually happened.

The reps had checked in from a chai dhaba two blocks from the actual kiryana store. Some had checked in from home. One guy had logged 11 visits in 40 minutes across a 14-kilometer stretch of Multan Road, which is physically impossible unless he was riding a rocket.

This is the ghost visit problem. And it's bleeding FMCG brands across South Asia, Africa, and Southeast Asia in ways most CFOs still underestimate.

What ghost visits actually cost you

Honestly, I used to think attendance fraud was a 5-10% problem. The kind of thing you accept as the cost of running a 200-person field force. Then I started looking at the numbers more carefully across distributors we work with.

The real figure sits closer to 22-31% in markets where reps are paid fixed salaries with weak supervision. In commission-heavy markets it drops, but only to about 12-15% because reps still fake visits to inflate their beat coverage metrics.

Do the math on a mid-sized FMCG distributor. 80 reps. Each costs PKR 65,000 monthly all-in. If a quarter of their reported visits are fake, you're effectively paying PKR 1.3 million a month for ghost work. That's before you count the opportunity cost — the actual outlets that never got visited, the SKUs that went out of stock, the competitor who walked in while your rep was scrolling TikTok.

And here's the thing that took me a while to accept: this isn't really about dishonest people. It's about broken incentive systems and broken supervision. Give anyone an unrealistic beat plan of 45 outlets a day in Karachi traffic and you've basically engineered the fraud yourself.

How GPS alone isn't enough (and what AI adds)

First-generation GPS sales rep tracking was a step forward. You knew where the rep's phone was. But reps figured out the workarounds fast.

They'd leave a second phone at an outlet they'd visited once. They'd check in early, then go home. They'd ask the shopkeeper to confirm a visit that never lasted more than a handshake. Pure location data tells you a dot was on a map. It doesn't tell you whether a real sales conversation happened.

The newer wave of field sales platforms — and I'll mention Zivni here because we've watched what their team is building closely — combines GPS with pattern recognition AI that flags things humans would miss. Things like:

• A rep whose visit durations cluster suspiciously around 90 seconds
• Outlets that always get visited at the same odd time (3:47 PM every Tuesday)
• Check-in geofences that are hit but with no corresponding order, no photo, no SKU scan
• Movement patterns that don't match the assigned beat plan
• Reps whose "productive visits" suddenly spike right before performance reviews

One distributor in Faisalabad ran this for six weeks. The system flagged 23 reps out of 67 for review. Eleven were let go. The remaining twelve got retrained and put on tighter beat plans. Productive outlet coverage went up 41% the following quarter without adding a single new hire.

That's the real ROI conversation. Not "catch bad reps." It's "figure out where your route to market for FMCG is actually broken and fix it."

The supervisor problem nobody talks about

Look, I'll say something uncomfortable. The technology isn't the hard part anymore. GPS works. AI pattern detection works. The hard part is what happens after the dashboard turns red.

Most area sales managers in Pakistan, Nigeria, Kenya, Bangladesh — they came up through the same system. They know which reps fake visits because they used to do it themselves. Asking them to crack down on their team feels like betrayal. So the data sits there, ignored, until someone at HQ finally notices.

The brands that actually solve ghost visits in field sales do three things together:

1. They make the data visible to the rep too, not just the manager. When a rep can see their own attendance pattern flagged in real-time, behavior shifts before any conversation happens.

2. They redesign the beat plan around what's physically achievable. The best route to market FMCG strategies I've seen assume 28-32 quality visits per day, not 50 garbage ones.

3. They tie a chunk of incentive to verified visits — ones with a photo of the shelf, an order placed, or a scanned SKU — not to logged visits.

A brand in Nairobi I spoke with switched their incentive structure last year. Verified visits only. Ghost visits dropped from 26% to under 4% in eleven weeks. They didn't fire anyone. They just stopped paying for fake work.

Where this goes next

The next frontier isn't catching ghost visits. It's predicting them before they happen. Some of the more interesting startup business in Pakistan circles are working on systems that look at rep fatigue indicators, beat plan stress scores, and historical patterns to flag which reps are likely to start faking visits next week — and intervene with a coaching call or a beat plan adjustment first.

Is that surveillance? Maybe. But anyone who's run a field force of 300 people across three provinces knows the alternative is worse: a slow rot you only notice when your market share has already been eaten.

The distributors getting ahead of this aren't the ones with the biggest tech budgets. They're the ones who stopped treating their reps like a black box and started treating field execution like a real engineering problem with real signals.

What would your numbers look like if you pulled last week's GPS logs against your reported visits right now?