Secondary Sales Tracking in Emerging Markets: How AI Turns Messy Data Into Distribution Strategy

A friend who runs sales for a mid-sized biscuit brand in Karachi told me last month that his team spends 11 days closing secondary sales numbers. Eleven days. By the time the report lands on his desk, the month is basically over and any decision he makes is a post-mortem, not a strategy.

This is the quiet crisis nobody in FMCG likes to talk about.

Primary sales — what the company ships to distributors — is clean. It's invoiced, it's in the ERP, everyone agrees on the number. But secondary sales — what the distributor actually sells to retailers — is where the story gets messy. And that's the number that actually matters. Because primary can be inflated with stock pushes, credit terms, and end-of-quarter theatrics. Secondary tells you if product is actually moving off shelves.

So why is it still such a mess in 2024?

The data problem nobody solved for 30 years

Here's the thing about distributor data in emerging markets. It comes in every format imaginable. Some distributors use Tally. Some use a homegrown Excel sheet a nephew built in 2011. Some just handwrite invoices and photograph them at the end of the week. In Bangladesh, Indonesia, Nigeria, Pakistan — you name it — a national FMCG brand might be pulling data from 400 distributors, each with a slightly different SKU code, a different naming convention, a different definition of what a "return" even means.

I used to think this was a solvable problem with better software. Give everyone the same DMS (distributor management system), enforce compliance, done. But I got that wrong. Distributors don't work for you. They work for themselves, and they carry 15 other brands, and they'll adopt your system only when it's easier than what they already have. Which is almost never.

So the data keeps arriving messy. SKU "BSC-CHOC-100" from one distributor is "CHOCO100G" from another and "Choco Biscuit 100gm" from a third. Multiply that by 400 distributors and 200 SKUs and you get why brand managers still fly blind.

Where AI actually earns its keep

The honest truth is that most "AI in FMCG" talk is marketing fluff. Predictive demand forecasting sounds sexy but if your input data is garbage, no model saves you. The place where machine learning genuinely moves the needle is upstream — cleaning, matching, and reconciling the mess before anyone tries to forecast anything.

SKU normalization is a boring but massive win. A decent model can look at "CHOCO100G" across 40 spelling variants and confidently map them to a single master SKU with 97% accuracy. That alone shrinks the monthly closing cycle from 11 days to about 3.

Outlet-level matching is the next unlock — sorry, the next real gain. When distributor A and distributor B both claim to sell to "Rehman General Store," is that the same store? Fuzzy matching on address, GPS if you have it, phone number, owner name. Suddenly you know your true retail universe isn't 180,000 outlets like the sales team claims. It's 112,000. And 38% of them haven't been billed in the last 60 days.

That's a distribution strategy conversation. Not a data conversation.

At Zivni, which is one of the platforms we've written about before at Alif Zero, the team built their field sales product specifically around this problem — cleaning secondary sales data at ingestion and reconciling it with what field reps actually see in stores. When your rep says an outlet is active but the distributor hasn't billed it in two months, that's a red flag the system can surface automatically instead of hiding in a spreadsheet nobody opens.

From cleaned data to actual decisions

Once the data is trustworthy, the questions you can ask change completely.

Instead of "what did we sell last month," you start asking things like: which 200 outlets in Lahore drive 40% of our category volume, and are we servicing them weekly or fortnightly? Which SKUs cannibalize each other in the same beat? When we ran that trade promo in South Punjab, did secondary actually lift or did the distributor just pre-book stock to hit the slab?

Look, these aren't new questions. Good sales directors have been asking them forever. The difference is now you can answer them in an afternoon instead of commissioning a three-month market research study that arrives after the season ends.

And this is where the best route to market for FMCG in emerging markets is getting rewritten. The old model — blanket coverage, hit every outlet, measure primary — is expensive and dumb. The new model is surgical. Cover the outlets that matter at the frequency they deserve, based on their actual secondary throughput, not their theoretical potential.

One consumer goods company I spoke with in Lagos cut their field force by 18% and grew secondary sales 22% in the same year. They didn't hire better reps. They just stopped sending reps to outlets that weren't buying, and doubled visits to outlets that were.

The part everyone gets wrong

Honestly? The biggest mistake I see brands make is treating this as an IT project. They buy a platform, run a pilot in one region, generate a beautiful dashboard, and then nothing changes because the regional sales manager doesn't trust the numbers and keeps running his territory off his own Excel.

Data without cultural adoption is expensive wallpaper.

The brands winning at this are the ones where the national sales head walks into the Monday review and refuses to discuss anything that isn't on the dashboard. Where distributor claims get validated against the system before payment. Where the field rep's incentive is tied to outlets billed, not just volume shipped. It's a discipline problem masquerading as a technology problem.

And when the discipline is there, the AI genuinely does something remarkable. It stops being a reporting tool and becomes a distribution strategy engine — telling you which markets to expand into, which distributors to replace, which SKUs to kill, which routes to redesign.

But only if someone actually reads what it says.

So what's the honest state of secondary sales tracking in emerging markets right now? Better than five years ago. Nowhere near where it needs to be. And the brands that figure it out in the next 24 months are going to eat the lunch of the ones still closing their books on day 11.