There's no single best Kaspi bot "for everyone". The best one is the bot that, on your products, reacts to competitors faster, never goes below your minimum price, can raise prices, and doesn't put your store at risk. You can check all of that yourself during a one-week trial, as long as you know in advance what to measure.
In short
- A website's feature list tells you almost nothing. Watch how the bot behaves on your product cards: how fast it responds to a competitor, whether it holds your minimum price, whether it rises back up after them.
- A 30-product store and an 800-product store have different priorities, so it's worth tuning the criteria weights to your own.
- A test is only fair if some products stay without the bot. Compare net profit, not the number of price changes.
- Before connecting, read what Kaspi's rules say about cabinet access, and ask the service what access it needs.
- What a dumping (repricing) bot is and how it works is covered in "What Is a Kaspi Bot". This article is only about choosing one.
Must-have features: without them, the bot won't do
Some features are so basic that without them there's no point comparing further.
- A minimum price for each product. Not one for the whole store and not a percentage of the current price, but a specific amount in tenge for each SKU. This is your protection against selling at a loss. We showed how to calculate it in "Price Dumping on Kaspi".
- Stopping at the minimum price. If a competitor is below your floor, the bot should hold at the floor rather than follow them down.
- Raising the price. When a competitor goes up or leaves the product card, the bot should raise your price to just below the next seller. A bot that can only lower prices automates a price war.
- A change log. For every product you can see when the price changed, by how much and why. Without a log you can't verify either speed or logic.
Comparison criteria
Reaction speed
Speed isn't "how often the bot checks prices" but how much time passes from a competitor's price change until your new price is visible to shoppers. That time includes both the check and the price being applied on Kaspi's side. Ask the service for both figures and verify them in a test (protocol below).
Why this matters is easy to calculate. While your price is higher than a competitor's, orders in the card go to them. The more often competitors change prices, the more of these "windows" there are per day and the more each minute of delay costs. On a quiet card where the price changes once a week, the speed difference between bots is barely noticeable. On a card with several automated shops, it decides who gets the orders. So assess speed on your most competitive products, not on average ones.
One measurement at lunchtime on a Tuesday means little. Repeat it in the evening, at night, on Saturday: you want both the usual reaction time and the longest one.
Rule flexibility
- A price-cut step in tenge, 1 ₸ by default.
- A maximum price: a ceiling above which the bot won't raise.
- Excluding specific competitors, for example your own second store.
- Different rules for different product groups.
- Switching the bot on and off for an individual product in a couple of clicks.
Stock, warehouses and cities
Kaspi's rules prohibit listing products that are out of stock, except for pre-orders (Kaspi.kz Store rules, clause 1.10). If you sell from several warehouses or in several cities, check how the bot handles stock and pre-orders and whether it disrupts them when changing prices.
Marketplaces
Do you also sell on Halyk Market? Then ask whether the service works there and whether you can set a separate minimum price for the different commission.
Access and security
The Kaspi.kz Store rules prohibit sharing access to the seller cabinet with third parties (clause 5.1). Kaspi doesn't publish a separate position on price automation services. So before connecting any external service, read the rules yourself and ask the service directly: what access does it request, where is the data stored, and can you disconnect it and change passwords at any time? Get the answers in writing.
Pricing transparency
Compare the cost per product that actually runs with the bot, not "up to N products". We explained how to calculate payback in "How Much a Kaspi Bot Costs".
A weighted selection matrix
Distribute 100 points across the criteria according to your store, then score each service on a scale from 0 to 5. Total = sum of (weight × score) ÷ 5.
Below is an example for an imaginary 150-SKU electronics store whose product cards see competitors change prices many times a day:
| Criterion | Weight | Service A | Service B | Service C |
|---|---|---|---|---|
| Minimum price per SKU and stopping at it | 25 | 5 | 5 | 3 |
| Raising the price after a competitor | 20 | 5 | 2 | 4 |
| Reaction speed in the test | 20 | 4 | 5 | 3 |
| Rule flexibility | 10 | 4 | 3 | 5 |
| Stock, warehouses, pre-orders | 10 | 4 | 4 | 2 |
| Change log | 5 | 5 | 3 | 4 |
| Clear access and disconnection | 10 | 4 | 4 | 3 |
| Total out of 100 | 88 | 76 | 68 |
The services here are made up, and so are the scores. But notice that the fastest one, Service B, didn't come first: it's poor at raising the price after a competitor. And in a market where prices go back and forth all day, that's money left on the table literally every day.
For a 20-SKU store where a competitor looks into the card once a week, the picture is different. Speed decides almost nothing there, while ease of setup and the subscription price come to the fore.
Weights for your type of store
The same matrix gives different answers for different stores. Guidelines for distributing the weights (hypothetical; adjust them to your situation):
| Criterion | Small store, up to 50 SKUs, few competitors | Mid-size, 100–500 SKUs, active dumping | Large, 800+ SKUs, several warehouses |
|---|---|---|---|
| Minimum price and stopping | 30 | 25 | 20 |
| Raising the price | 15 | 20 | 20 |
| Reaction speed | 10 | 20 | 15 |
| Rule flexibility | 10 | 10 | 15 |
| Stock, warehouses, pre-orders | 5 | 10 | 20 |
| Change log | 10 | 5 | 5 |
| Access and disconnection | 10 | 10 | 5 |
| Ease of setup and price | 10 | 0 | 0 |
Here's why. In a small store, every sale at a loss shows up in the monthly totals, and competitors rarely move, so the main thing is not to make mistakes and not to overpay for the plan. A mid-size store in a hot niche makes its money on speed and on how the bot raises prices. For a large store, everything comes down to the link with stock and warehouses: if availability drifts across hundreds of items, cancellations will hit quality metrics far harder than a couple of lost minutes.
A 7-day test protocol
A one-week trial is enough if you test to a plan, not "switch it on and watch the orders".
Day 1. Preparation
- Pick 20–40 products that have competitors and split them into two groups that are similar in sales and price.
- Group A runs with the bot; group B stays as before, without the bot.
- For each product in group A, set a minimum price using the full formula.
- Record baseline data for both groups: orders and net profit for the previous 7 days.
Days 2–3. Speed and logic
Speed is easiest to measure on 3–5 products where you know the competitor well. They changed their price at 21:14, your new price appeared in the card at 21:16: that's a two-minute reaction, no website promises involved.
Then find a card where a competitor has gone below your minimum price. The bot should stop at your floor and go no further. And third: wait for the competitor to raise their price or sell out. If your price just stays down there, money is leaking out every day.
Days 4–5. Edge cases
Separately, look at the log for the night from Saturday to Sunday: did the bot work while you were asleep? Check a product that has run out of stock: it mustn't be listed for sale. And a card where three competitors sit at the same price: it's worth seeing where the bot lands in that situation.
Days 6–7. Results
Compare groups A and B using this formula:
Bot effect = (A's profit during the test − A's profit in the base period) − (B's profit during the test − B's profit in the base period)
Group B is there to separate the bot from everything else: sales, holidays, payday week. If there's a gain left and it's bigger than the subscription, the bot works. If not, don't rush to blame it; open the log. It often turns out the bot was hitting minimum prices that were set too high.
What to record
| Metric | How to calculate |
|---|---|
| Reaction time | Time your price appeared − time the competitor's price changed |
| Share of time at the top | Hours in first place ÷ hours observed |
| Average selling price | Revenue ÷ number of orders in the group |
| Net profit | Using the formula from the article on selling at a loss |
| Errors | Sales below the minimum price, listing without stock |
Even one sale below the minimum price caused by the bot is a reason to stop, no matter how many orders it brought in.
How to switch from one bot to another
Switching bots makes for a risky week. The two most expensive mistakes here are a product with both bots running on it at once, and a product left with no minimum price at all during the move.
- Export the old bot's settings: minimum and maximum prices, step, exclusions. If there's no export, copy down by hand at least the minimum prices for your fifty main products.
- Recalculate minimum prices. Since you're moving everything anyway, check them against the full formula: Kaspi's rates changed in 2026.
- Move products in batches of 20–40. Switch off in the old bot, switch on in the new one, check the log, take the next batch.
- Not a single SKU under two bots. They'll start responding to each other, and the price will jump around for no reason at all.
- After a week, compare net profit and time at the top for the moved products with the previous week.
- Shut the old service down completely and change passwords if you gave it access.
Questions to ask the service before paying
- How exactly is the minimum price set, and what does the bot do if a competitor is below it?
- Does the bot raise the price, and up to what level?
- On average, how much time passes from a competitor's price change to my price being applied?
- What access to my store do you request, and how do I revoke it?
- Where is the change log stored, and can I export it?
- How does the bot handle stock, warehouses and pre-orders?
- What happens in case of a failure: does the bot stop or keep going?
- How is the subscription priced: for all products or only those running with the bot?
Red flags
- No clear minimum price for each product. Only a percentage or a "smart mode with no settings".
- The bot can only lower prices. Raising prices is promised "in future versions".
- No change log. You can't verify a single promise.
- Big numbers without a source. Sales-growth percentages and "market shares" without a methodology can't be verified. Judge by your own test.
- A promise of permanent first place. Under Kaspi's rules, position in the card also depends on bonuses, rating, reviews and delivery speed (Kaspi Guide). A bot only controls price.
- A vague answer about access and disconnection.
Common mistakes
- Choosing by the website, not by a test. On paper the features are the same; on real cards the behavior differs.
- Testing without a control group. A rise in orders during the test week may be seasonal, and you'll credit it to the bot.
- Connecting the whole assortment at once. A minimum-price error across 500 products costs more than across 20.
- Counting price changes instead of profit. A thousand changes a day say nothing about money.
- Forgetting about raising prices. A bot that only lowers prices drags both you and the market down.
- Running two bots on one product during a switch. The price starts jumping, and the log can no longer explain who changed it and why.
- Not asking about access. Your store's security matters more than a difference in subscription price.
Selection checklist
- A minimum price is set per SKU in tenge.
- The bot holds at the minimum price if a competitor is lower.
- The bot raises the price after a competitor goes up or leaves.
- There's a change log with export.
- Reaction time has been measured on your own cards.
- Handling of stock and pre-orders has been checked.
- It's clear what access is needed and how to revoke it.
- The test was run with a control group.
- The profit effect covers the subscription price.
- Not a single sale below the minimum price during the test.
How to automate it
If you're considering SaleScout, the protocol above works for it too. The SaleScout Kaspi bot checks prices every 20 seconds, never goes below the minimum price you set, and raises the price when a competitor goes up or leaves. Halyk Market is supported. You can try the bot free for 7 days, no card required, which is exactly the time you need for the test plan in this article.
FAQ
Which Kaspi bot is the best?
The one that, on your products, reacts to competitors faster, never goes below your minimum price and can raise prices. There's no universal best: compare two or three services using the same test protocol with a control group and choose by net profit.
How do I test a Kaspi bot during the trial period?
Take 20–40 products and split them into two similar groups, with the bot on only one. Over a week, time how fast it responds to a competitor, and check that it stops at the minimum price and rises back up after them. At the end, compare how much profit grew in each group.
Is it safe to connect a Kaspi bot?
Kaspi doesn't publish a separate position on price automation services, and the Store rules prohibit sharing access to the seller cabinet with third parties. Before connecting, read the rules, ask the service what access it requests and how to revoke it, and get the answer in writing.
How fast should a dumping bot react?
The more actively competitors change prices, the more speed matters. Measure not how often prices are checked but the time from a competitor's price change to your new price appearing in the card. Compare services on that figure using the same products.
Does a bot need to be able to raise prices?
Yes. When a competitor raises their price or leaves the card, the bot should raise your price to just below the next seller. Without that, you keep selling cheaper than you could on every order.
Can I use two Kaspi bots at the same time?
On different products, yes, if you're comparing services. On the same product, no: the bots will start reacting to each other's changes, the price will jump around, and the log won't tell you who changed it.
How many products should I connect to a bot at first?
Not necessarily. For a test, 20–40 products with active competitors plus the same number in a control group are enough. Once you're sure the minimum prices are right and the bot behaves as it should, connect the rest in batches; that way a settings error affects one batch, not the whole store.
Sources
All links checked Oct 15, 2026. The sources are in Russian; original titles are given with English translations.
- Kaspi Guide for partners. «Какие правила работы с Магазином на Kaspi.kz?» ("What are the rules for working with the Kaspi.kz Store?", clauses 1.10, 5.1) — https://guide.kaspi.kz/partner/ru/shop/conditions/q1342 (updated Sep 30, 2026)
- Kaspi Guide for partners. «Как располагаются продавцы в карточке товара?» ("How are sellers ranked in a product card?") — https://guide.kaspi.kz/partner/ru/shop/goods/general/q3190 (updated Jun 15, 2026)
Bota, SaleScout analyst
SaleScout is our company's product. Kaspi and Kaspi.kz are trademarks of Kaspi.kz JSC; SaleScout is not affiliated with Kaspi.
