Route Optimization for Sales Reps: The Math That Actually Matters
A rep in Karachi told me last year that his "optimized" route had him driving past 4 outlets he was supposed to visit, just to come back to them 90 minutes later. The software was technically right. The shortest total distance. But it ignored the fact that two of those outlets only open after 11am, one owner takes lunch from 1 to 3, and the kiryana on the corner won't talk to you if you show up before he's done with his morning chai customers.
That's the gap. That's the whole post in one paragraph, honestly.
Most route optimization software treats field sales like a courier problem. It isn't. A DHL driver drops a package and leaves. A field rep has to sell, collect, merchandise, sometimes argue about a returned SKU from three weeks ago, and get a signature. The math is completely different. And I got this wrong at first when we were building Zivni — we shipped a TSP-based algorithm in our first version that looked beautiful on paper and made reps furious in practice.
Let me walk through what actually matters.
The four variables most tools ignore
Classic route optimization solves the Traveling Salesman Problem. Find the shortest path that hits every point and returns to start. Great for pizza delivery. Useless for a rep covering 32 outlets in Dubai's Deira district.
Here's what real sales route optimization needs to factor in:
Outlet open windows. A grocery in Riyadh might shut for Asr prayer. A pharmacy in Manchester won't accept deliveries after 4pm. A supermarket in Lahore wants reps before 10am or they're too busy with walk-in customers. If your route optimization app doesn't know this, it's guessing.
Visit duration variance. A quick order pickup at a small kiryana takes 7 minutes. A modern trade audit at Carrefour in Sharjah takes 45. Treating them as equal is why reps finish day 1 at 6pm and day 2 at 2pm with three outlets skipped.
Frequency rules. Outlet A needs visiting twice a week. Outlet B every 10 days. Outlet C only when stock drops below a threshold. Sales territory routing has to schedule across the week, not just the day.
Rep skill and relationship. Some reps own certain accounts. The shop owner literally won't deal with anyone else. You can't optimize that away with an algorithm. The system has to respect it.
The math that handles all this isn't TSP. It's closer to vehicle routing with time windows (VRPTW) layered with constraint satisfaction. And even then, you need a human override layer, because the algorithm doesn't know that Mr. Hassan's son got married last week and the rep needs an extra 10 minutes for chai.
What good optimization actually looks like
When we rebuilt this at Zivni, we threw out the "shortest path" obsession. Here's what we optimize for instead, roughly in order:
- Outlets visited per day (not km saved)
- Orders captured per visit
- Time-window compliance (did we hit outlets when they're actually receptive)
- Distance and fuel, as a tiebreaker, not the primary metric
One distributor we work with in Muscat was running 18 outlets per rep per day. After three weeks on the new logic, they were hitting 23. Same reps, same vehicles, same territory. The route was technically 11% longer in kilometers. But fuel cost went up 6% while revenue per rep went up 27%. That's the trade nobody talks about.
And here's the thing — reps trusted the new routes within about 10 days. The old "optimized" ones, they were constantly overriding. Which made all the route data useless for analysis anyway.
The merchandiser problem nobody solves
Merchandisers are different from order-booking reps and most software lumps them together. A merchandiser in a Tesco in Birmingham might spend 90 minutes in one store doing planogram resets. They don't need 30 outlets a day. They need 4 to 6, and the route should account for parking, loading times, and which stores have a back-of-house entrance.
We added a separate routing mode for merchandisers after watching one of them in Jeddah spend 22 minutes looking for a parking spot at a hypermarket because the route assumed he'd just pull up curbside like at a small store. Look, optimization software has to know what kind of visit it's planning, or the math is fiction.
What to actually ask a vendor
If you're evaluating route optimization tools — Zivni, FieldAssist, BeatRoute, Repsly, doesn't matter — ask these specific questions:
- Can I set different visit durations per outlet, or just per outlet type?
- Can the route respect prayer times, lunch breaks, and owner-specific time windows?
- Does the rep see the reason the algorithm chose this sequence, or just the dots on a map?
- Can a sales manager lock certain visits to certain times without breaking the rest of the plan?
- What happens when a rep skips an outlet — does tomorrow's route adjust?
If the answer to any of these is "we're working on it," you're buying a courier-routing engine with a sales sticker on it.
The math that matters in sales route optimization isn't about minimizing distance. It's about maximizing the number of productive conversations a rep can have between 9am and 6pm. Everything else is secondary. Distance saved is nice. Fuel saved is nice. But if your rep visited 19 outlets instead of 24 and missed two of his biggest accounts because the system thought a 3pm visit was "efficient" — what did you actually optimize?
That's the question I'd start with.