A distributor with 200 shops has thousands of rows of order data every month. Inside them are the answers to the most important sales questions: which shops could actually buy more, which items to offer, and when. The problem is that reading that data shop by shop takes time nobody has. This is where AI is most useful.
The solutionWhich shops could buy more, every nightSee how it worksThe problem: the data exists, but nobody reads it
Almost every distributor already records every order in a system. What's rare is someone who reads it regularly for every shop. So salesmen visit shops out of habit, not opportunity. A shop that starts moving to a competitor only gets noticed after three months without an order.
Manual analysis is possible. At one distributor we work with, analysing a single shop in Excel took one to two hours: pulling order history, comparing it with similar shops, looking for items they stopped buying. With hundreds of shops, that means hundreds of salesman hours every round. In the end it was rarely done.
How AI analyses each shop
The process runs every night, without anyone pressing a button.
1. It reads every shop's order history. AI takes 12 to 18 months of order data per shop: which items, how many, how often.
2. It compares against the expected pattern. Each shop is compared with its own history and with similar shops, for example home appliance shops with similar turnover in the same city.
3. It finds signs of opportunity. For example, a shop that used to buy fans regularly but stopped, a shop that buys refrigerators but never water dispensers even though similar shops buy both, or a shop ordering less and less often. We cover the five most common signs in five signs your customer shops could actually buy more.
4. It puts a rupiah value on each one. Every opportunity gets an estimated value, so salesmen know which to chase first.
5. It puts them in the visit plan. The result doesn't stop at a report. The list of shops and items to offer goes straight into each salesman's visit plan for the next day.
What people still do: the salesman decides and talks to the shop owner. AI only makes sure no opportunity is missed because nobody had time to read the data.
The results in the field
At the same distributor, after the analysis started running automatically:
| Measure | Before | After |
|---|---|---|
| Analysis time | 1–2 hours per shop, in Excel | Every shop in one night |
| Staff hours per round | About 400 salesman hours | Zero |
| Accuracy | Depends on who does the analysis | 2–4 times more accurate, because every shop is compared the same way |
These figures come from the client's own records. What sales managers notice most isn't the hours, but how the weekly meeting changes: from guessing which shops have problems to discussing a list that's already there.
What data you need
The good news is that AI doesn't need new data. What it needs usually already exists:
- order history per shop, per item, per date
- the shop list with each shop's area and salesman
- the item list with categories
- receivables data if available, so shops that are paying late aren't pushed to order more
This data can be pulled from the ERP or accounting system you already use. The system doesn't need replacing.
How salesmen use it
This part is often underestimated. Even the best analysis is useless if salesmen don't open it. So the result needs to appear where they already work every day: in the sales app on their phone, as a short "shops for today" list with one or two items to offer per shop.
After each visit, the salesman records a voice summary. AI turns it into a report and notes whether the offer worked. That makes the next night's analysis more accurate, and shows managers which salesmen need help.
Ready-made software or built for you
Some sales force automation apps already have recommendation features. The question is whether their rules match the way your distributor sells: item categories, seasonal patterns, price tiers, credit rules. If they don't fit, salesmen will ignore the recommendations.
We cover the full trade-off in ERP, ready-made software or a custom app for distributors, and the costs in what it costs to build a custom app.
Frequently asked questions
Does AI replace salesmen? No. AI prepares the list of shops and items to offer. The relationship with the shop owner, the negotiation and the final decision stay with the salesman.
How many shops before this is worth it? Once one salesman handles more than 30 to 40 shops, reading every shop's data by hand each month is usually no longer possible. That's where automatic analysis starts to pay off.
What about new shops with no history? New shops are compared with similar shops in the same area. After two or three months of orders, their own history starts to be used.
Where to start
If you want to try it, start with one area or one sales team for 30 days. Record the sales from recommended shops and compare them with other areas. The general steps are the same as in how to use AI in your company: one job, a clear measure, then expand.
See how this runs in the Sales App, or tell us about your sales team in a free consultation.