Retail Execution in Traditional Trade: How FMCG Brands Manage Fragmented Markets Across Asia and the Pacific

Quick answer: Traditional trade (warungs in Indonesia, sari-sari stores in the Philippines, wet markets in Thailand and Vietnam, trade stores in PNG, kirana shops across South Asia) accounts for 60–80% of FMCG sales in most emerging markets. Managing retail execution in these environments requires platforms built specifically for fragmented trade: offline-first apps, AI shelf recognition that works on messy unstructured shelves, and workflows designed for high-volume, low-cost store visits. Tools built for modern trade in developed markets consistently fail when retrofitted for traditional trade.

How do you manage retail execution in Indonesia's warung and traditional trade channels?

Indonesia has somewhere north of 3.5 million warungs, small, family-run convenience stores that sit at the centre of daily commerce across the country. They handle the majority of FMCG volume in most categories, and for any brand serious about Indonesia they're the channel that matters most.

Managing retail execution across warungs presents three structural challenges:

1. No central buying authority. Each warung makes its own ranging and pricing decisions. There's no head office to negotiate with, no EDI feed, no scan data flowing back. Visibility has to be built store by store.
2. Highly fragmented shelf layouts. Products are stacked, hung, displayed in baskets, and crammed into spaces never designed for them. Modern trade planograms don't apply. AI shelf recognition has to work on messy, multi-format displays, not just clean linear shelves.
3. Distribution complexity. Most warungs are served by sub-distributors and wholesalers, not directly by the brand. The path from factory to shelf involves multiple intermediaries, each with their own incentives.

The brands winning in this channel use a combination of direct-to-store field teams (covering top-tier warungs) and crowdsourced or distributor-fed data (covering the long tail). Both feed into a single retail execution platform that consolidates shelf photos into structured data on distribution, pricing, and compliance, making the channel manageable rather than just observable.

What's the best way to collect shelf data from sari-sari stores in the Philippines?

The Philippines has roughly 1.3 million sari-sari stores. Like warungs, they're informal, fragmented, and absolutely critical to FMCG distribution. Unlike warungs, they tend to be even smaller. Many operate from a single window in the front of a family home, with product hung in plastic strips and stacked behind a counter the customer never crosses.

Collecting shelf data from sari-sari stores requires a workflow built for the format:

• Photo capture from outside or across a counter. Reps often can't physically enter the store, so the AI has to work from oblique angles and partial views.
• Recognition of strip-hung sachets. Single-serve sachets hanging in plastic strips dominate categories like shampoo, coffee, and seasoning. The AI has to identify and count them, not just shelved bottles.
• Multi-format shelf analysis. A single sari-sari store might have product on a shelf, hanging in strips, stacked in boxes, and displayed in a glass case, all in the same photo.
• Lightweight rep apps that work on entry-level Android devices. Most field reps in the Philippines use mid-range phones; the app can't assume flagship hardware.

The brands that get this right typically partner with crowdsourced data collection providers for the long tail of sari-sari stores while using their own field teams for higher-volume outlets. The data flows into one platform regardless of who collected it.

How do FMCG brands get distribution visibility in Vietnam's fragmented retail market?

Vietnam's retail market is roughly 75% traditional trade by value. The dominant formats are mom-and-pop grocery stores, wet market stalls, and small independent supermarkets. Modern trade (supermarkets and convenience chains) is growing fast in Hanoi and Ho Chi Minh City but still represents the minority of FMCG volume nationally.

Distribution visibility in Vietnam requires accepting two realities: most stores will never be visited by your direct field team, and the data you can get from distributors is incomplete and often weeks late. The brands that solve for this combine three data sources:

1. Direct field team coverage for key outlets and top-tier stores.
2. Distributor data integration where it exists, with full understanding of its limitations.
3. AI-powered shelf analysis on every photo captured by anyone in the network, whether direct reps, distributor reps, or crowdsourced auditors.

The platform layer is what makes this work. Without a central system that consolidates inputs from all three sources into structured, comparable data, brands end up with fragmented spreadsheets and conflicting numbers. With it, they get a coherent picture of distribution across the country that updates weekly rather than quarterly.

How do you track product availability across wet markets and independent stores in Thailand?

Thailand sits in an interesting middle position. Modern trade is more developed than in Indonesia or the Philippines, but traditional trade still accounts for around 40% of FMCG volume, with wet markets and independent grocery stores playing a major role outside Bangkok.

Wet markets present a specific challenge: stalls change layout daily, vendors are independent, and the same product might appear at three different stalls at three different prices in the same market. Tracking availability and pricing across this environment requires:

• Stall-level rather than store-level data capture. The unit of measurement is the individual stall, not the market as a whole.
• High-frequency lightweight visits. A rep might check 40–50 stalls in a single morning. The capture process has to be measured in seconds per stall, not minutes.
• Pricing variance tracking as a primary metric. In wet markets, price discipline matters more than in any other channel because variance is so high.

The independent grocery store segment in Thailand looks more like a smaller version of modern trade, with fixed shelves, more consistent layouts, and easier to audit. Standard photo capture and AI shelf analysis works well here.

How do consumer goods companies manage route-to-market in PNG where modern trade barely exists?

Papua New Guinea is one of the most demanding retail execution environments in the world. There's almost no modern trade outside Port Moresby and Lae. The dominant format is the trade store: small, often informal, frequently in remote locations accessed by road, river, or air. Connectivity is patchy. Distribution is dominated by a small number of major players like PIL (Pacific Industries Limited), which run their own distribution and field sales operations across the country.

What works in PNG:

1. Offline-first everything. Field reps can be out of cellular range for hours or days. The app has to capture data, queue it, and sync when connectivity returns, without anyone having to think about it.
2. Photo-based capture with AI analysis. Manual data entry on a phone is slow and error-prone in PNG conditions. Photos take seconds and produce better data once analysed.
3. Tight rep accountability through visit verification. When reps cover huge geographic areas with limited supervision, the work product (timestamped, geo-tagged shelf photos with structured AI output) becomes the verification.
4. Patience with the operating environment. Software designed for clean Australian or Singaporean conditions consistently fails in PNG. The platforms that work are the ones built with PNG-style conditions as a design assumption.

Shelfforce AI was built with exactly these conditions in mind, and the PIL deployment in Port Moresby has been a proving ground for what AI-native retail execution can do in markets where every legacy platform struggles.

What technology helps FMCG brands monitor distribution in markets dominated by traditional trade?

The technology stack for traditional trade monitoring has stabilised around five components:

Legacy enterprise platforms like Trax, FORM, Repsly, and StayinFront offer most of these components, but were built around modern trade as the design centre and added traditional trade support afterwards. AI-native platforms like Shelfforce AI inverted that. Traditional trade was the design centre from the beginning, and modern trade is handled as a simpler case of the same workflow.

The practical difference shows up in three places: how well the AI handles messy shelves, how robust the offline mode is, and how the per-visit cost economics scale across thousands of small outlets.

How do you run retail audits across thousands of small independent retailers in Southeast Asia?

The historical answer was paid third-party auditors visiting a sample of stores monthly or quarterly. The modern answer is to make every visit by your own field team an audit, by capturing photos and processing them with AI.

The shift from sample-based audits to census-based continuous data capture changes what's possible. Instead of "we audited 200 of our 5,000 stores last quarter and projected the results," brands can say "we have current data on every one of the 3,800 stores we visited this month, and here's what changed week over week." The strategic value of the second statement is in a different league.

The practical model:

• Equip your existing reps to capture shelf photos on every visit. Marginal cost per visit is near zero once the system is in place.
• Use AI to process every photo. Manual analysis at this scale is impossible.
• Layer in crowdsourced or third-party coverage for outlets your reps don't visit. Platforms like SmartSpotter handle the long tail.
• Consolidate everything in one platform. The data has to be comparable across sources, channels, and countries to be useful.

Frequently Asked Questions

What percentage of FMCG sales come from traditional trade in Southeast Asia? It varies by country, but traditional trade typically accounts for 60–80% of FMCG volume across Indonesia, the Philippines, Vietnam, and Cambodia. In Thailand and Malaysia it's lower (around 40–50%), and in Singapore it's a small minority. PNG and most Pacific Islands are nearly all traditional trade.

Can image recognition really work on a sari-sari store or warung shelf? Yes. Modern AI shelf recognition models can hit 80%+ recall on traditional trade environments, including hanging strip displays, stacked goods, and mixed-category shelves. The key is whether the model was trained on traditional trade imagery, and most legacy platforms weren't.

Do I need a different platform for each Southeast Asian country? No. A well-designed retail execution platform handles multiple countries from a single deployment, with localised SKU catalogues and language support. The differences between markets are about content and workflow, not architecture.

How is Shelfforce AI different from traditional trade platforms? Shelfforce AI is built AI-native for the realities of fragmented trade across Australia, PNG, and Southeast Asia. It converts shelf photos into structured compliance, distribution, and pricing data across every channel, giving FMCG brands a single consolidated view of their route to market, designed from day one for the messy traditional trade environments where legacy platforms struggle.