How AI Shelf Analysis Is Transforming Retail Execution for Pakistani FMCG Brands

Last month, I was sitting with the regional sales head of a mid-size FMCG company in Lahore. He pulled out his phone and showed me photos his field reps had taken at outlets across Punjab. Hundreds of them. Shelves, coolers, display racks — all snapped quickly during visits.

"Sufyan, we have all this data," he said. "But nobody actually looks at it. My team takes the photos because it's required. Then nothing happens."

This is the reality for most FMCG companies in Pakistan right now. You've got reps visiting 20-30 outlets a day across Karachi, Islamabad, Faisalabad, Peshawar. They're taking shelf photos as proof of visit. But those images just sit in a folder somewhere. Nobody's analyzing them. Nobody's acting on them.

That's where AI shelf analysis changes everything — and I don't say that lightly.

The real problem isn't data collection, it's insight

Pakistani FMCG brands have gotten pretty good at collecting field data. Most sales teams now use some kind of app. Photos get uploaded. GPS gets logged. But here's what I've seen over and over: the gap between capturing a shelf photo and actually knowing what's happening on that shelf is enormous.

Think about what a national sales manager actually needs to know:

• Is our product on the shelf where it should be?
• Is the competitor taking more space than last month?
• Are we compliant with the planogram we agreed on with the retailer?
• Is there a stockout happening that nobody reported?

Traditionally, this required someone — usually a tired area sales manager — to manually review photos and fill out audit forms. Or you'd hire a third-party retail audit firm that gives you a report... six weeks later. By the time you act, the damage is done.

AI shelf analysis does this in seconds. Your rep takes a photo, the AI identifies products, counts facings, checks placement, and tells you immediately whether that shelf matches what it's supposed to look like. That's not a future thing. It's happening now.

How planogram compliance AI actually works in the field

I'll be honest — when we first started exploring this at Zivni, I was skeptical about how well it would work in Pakistani retail environments. The kiryana store in Saddar, Karachi doesn't look like a Walmart aisle. Shelves are messy. Lighting is terrible. Products are stacked in ways that would give a merchandising team nightmares.

But that's exactly why planogram compliance AI is so valuable here. It works despite the chaos.

Here's the basic flow: your field rep walks into an outlet, opens the app, and takes a photo of the shelf or display. The AI model — trained on thousands of product images specific to your SKUs — processes that image and identifies what's on the shelf. It maps this against your target planogram or display guidelines. Within seconds, you get a compliance score.

Red flags pop up automatically. Missing SKUs. Wrong placement. Competitor encroachment. The rep can fix issues on the spot instead of moving to the next outlet oblivious. And the area manager gets a dashboard view across all outlets without manually checking anything.

What I think makes this particularly powerful for Pakistan's market is the traditional trade problem. We're not talking about 50 modern trade stores with neat shelving. We're talking about hundreds of thousands of general trade outlets across cities and towns where retail execution has historically been a black box.

The numbers that matter

I've been tracking results from early adopters, and honestly, some numbers surprised even me.

One FMCG company we've been working with saw their planogram compliance go from around 40% to over 70% within three months of using AI shelf analysis. That's not because their reps suddenly became better merchandisers. It's because they could finally see the problem in real time and fix it.

Another thing — stockout detection. A mid-size dairy brand in Punjab was losing an estimated 8-12% in sales from out-of-stock situations that nobody was catching fast enough. When the AI started flagging stockouts from shelf photos, their order fill rates improved because reps started placing replenishment orders during the visit itself.

Retail execution technology like this pays for itself fast. When you're spending millions on trade promotions and in-store displays, knowing whether those investments are actually showing up on the shelf isn't optional. It's the whole point.

Why this matters more in emerging markets

Here's something I think about a lot. In markets like the US or Western Europe, brands have sophisticated retail audit systems. Nielsen and IRI give you shelf data. Modern trade retailers share POS data. There's infrastructure for this.

In Pakistan, the UAE, across Southeast Asia and Africa — we don't have that luxury in traditional trade. And traditional trade is where 70-80% of FMCG sales happen. So you're essentially flying blind on the majority of your business.

AI shelf analysis isn't just a nice efficiency gain here. It's giving FMCG brands visibility they've literally never had before. A sales director in Karachi can now see what shelves look like in Multan, Sukkur, and Quetta in real time. That was impossible two years ago without sending someone physically.

I was in Dubai recently talking to a distribution company that handles multiple FMCG brands across the UAE. Same story. They've got merchandisers visiting grocery stores and baqalas across Dubai, Abu Dhabi, Sharjah. Photos get taken. Nobody analyzes them. The moment we showed them what automated shelf recognition could do, they immediately got it.

The competitive advantage for brands who adopt this early is real. If you know your shelf compliance is slipping in a specific area before your competitor does, you act first. You fix it. You protect your space. In FMCG, shelf presence is revenue. It's that direct.

What I'd tell a sales manager considering this

If you're running field sales for an FMCG company and you're thinking about AI shelf analysis, here's my practical advice:

Start small. Pick one region or one product category. Train the AI on your specific SKUs — this matters because generic image recognition won't cut it for Pakistani products with Urdu labeling and local packaging. Get your reps comfortable with the photo process. Then expand.

Don't treat it as a surveillance tool. I've seen companies make this mistake — they use shelf photos to punish reps instead of helping them. The reps push back, take bad photos on purpose, and the whole thing falls apart. Position it as a tool that helps the rep do their job better and close more orders.

And integrate it with your order management. The real magic happens when a rep sees a stockout on the shelf analysis and immediately places a replenishment order in the same app. That's a closed loop. That's where revenue impact shows up.

We're building all of this into Zivni because I genuinely believe retail execution technology is the next big unlock for FMCG growth in markets like ours. The brands that figure this out in the next 12-18 months are going to pull ahead significantly. The ones that keep collecting photos nobody looks at... well, they'll keep wondering why their shelf space is shrinking.