What a Kaspi bot costs depends on the pricing model (per product, a package of "up to N products", or per store) and on how many products you actually hand over to it. A more useful question, though, is how much it brings in above the subscription price. One formula with five terms answers that, and you can test it on your own products in a week, before you pay anything.
In short
- Compare pricing per product that actually runs in the bot. The package's upper limit is irrelevant here.
- A bot's value comes from extra orders, raising your price after competitors, and hours saved. From that you subtract the margin you give away when prices are cut more often.
- Break-even: subscription ÷ profit per order = how many extra orders a month you need for the bot to pay for itself.
- Verify the calculation on a control group during the trial period, not by looking at your store's overall trend.
- How to choose a bot by features is covered in "The Best Kaspi Bot: How to Choose".
Pricing models
There are essentially three payment models on the market. The numbers vary between services and change often, so let's start with how it works.
| Model | How it's charged | Who it suits | What to watch |
|---|---|---|---|
| Per product | A fixed monthly amount for each connected SKU | Few products, testing, seasonal assortment | The price as the number of products grows |
| Package of "up to N products" | A fixed amount for a range | A stable assortment close to the package's upper limit | How many products you'll actually connect |
| Per store or turnover | An amount per store or a percentage | Large stores | How turnover is calculated, what's included |
Price per active product
The main comparison metric:
Price per active product = Monthly subscription ÷ Number of products actually connected to the bot
A hypothetical example: an "up to 100 products" package at 35,000 ₸ costs 875 ₸ per product with 40 products connected, and about 368 ₸ with 95. Same package, different price. There's no point connecting the bot to products without competitors just to "fill the package": the bot changes almost nothing there.
What else costs money
The service's invoice won't include setup time. You'll have to calculate a minimum price for every product; the first hundred SKUs will take several hours, and that's if your landed-cost table is already in place. You also need time to check: one typo in a minimum price can cost more than a month's subscription. And if your stock lives in an inventory system, syncing it with the bot can turn into a separate mini-project.
The payback formula
Monthly bot effect = Extra orders × Profit per order + Price raise × Orders + Hours saved × Cost per hour − Margin loss − Subscription
Let's go through the terms.
Extra orders
Orders you got because you reacted faster to a competitor's price cut. Without a bot, part of the time you sit above the leader and the orders go to them. Count them not by the store's overall growth but by the difference with a control group (details below).
Price raise
When a competitor raises their price or leaves the product card, the bot raises your price to just below the next seller. Done by hand, this is often missed: a price cut in the morning stays low until evening or until next week. It's calculated as the average difference in selling price × number of orders.
Time saved
The hours previously spent tracking prices by hand, multiplied by the cost of a manager's hour or your own. We showed how to calculate manual hours in "What Is a Kaspi Bot": products × checks per day × minutes ÷ 60.
Margin loss
Because the bot reacts faster, on hot cards it will more often sit closer to your minimum price. That's the real price of the top spot in the card, and we subtract it honestly. The good news is that this loss has a floor: the bot won't go below a correctly calculated minimum price.
Worked example
Picture a store (a hypothetical example, not a client case) that hands 80 products with competitors to a bot. On average 1,500 ₸ is left per order; for the calculation, let's take a subscription of 35,000 ₸ a month.
| Term | Assumption | Per month |
|---|---|---|
| Extra orders | +40 orders × 1,500 ₸ | +60,000 ₸ |
| Price raise | +150 ₸ on average × 300 orders | +45,000 ₸ |
| Margin loss | −60 ₸ on average × 300 orders | −18,000 ₸ |
| Time saved | 40 hours × 2,000 ₸ | +80,000 ₸ |
| Subscription | hypothetical | −35,000 ₸ |
| Effect | +132,000 ₸ |
All assumptions are hypothetical. Without the time term, the effect is 52,000 ₸; that version is more honest if the manager is on a fixed salary anyway and their hours aren't freed up for other work.
Break-even point
Extra orders to break even = Subscription ÷ Profit per order
In the example, 35,000 ÷ 1,500 ≈ 24 extra orders a month, not counting price raises and time. That's less than one order a day for the whole store. If your products don't reach even that, the problem is usually not the bot but product selection: on cards without competitors, a bot has nothing to do.
Break-even table
How many extra orders a month you need to cover the subscription at different levels of profit per order. The subscription prices in the table are hypothetical; plug in your own service's price.
| Monthly subscription | Profit 500 ₸ | Profit 1,000 ₸ | Profit 1,500 ₸ | Profit 3,000 ₸ |
|---|---|---|---|---|
| 20,000 ₸ | 40 | 20 | 14 | 7 |
| 35,000 ₸ | 70 | 35 | 24 | 12 |
| 50,000 ₸ | 100 | 50 | 34 | 17 |
| 100,000 ₸ | 200 | 100 | 67 | 34 |
The conclusion from the table is unpleasant but useful: the less you make per order, the more extra orders the bot has to bring in. If 500 ₸ is left per order and the subscription is expensive, automation can wait; first you need to understand where your profit is going. That's what "How to Stop Selling at a Loss on Kaspi" is about.
When a bot won't pay off
- There are no competitors in your cards, or they change prices once a week.
- Your minimum price is almost equal to the market price: the bot has no room to cut.
- Your rating or quality metrics keep you from reaching the top at any price.
- You have few products, and the subscription is a package designed for hundreds of SKUs.
How to estimate the cost of an hour
The "time" term is debatable, so calculate it carefully.
- If a hired manager handles prices, the cost of an hour is their monthly salary and tax costs ÷ working hours. But count the time only if you actually cut those hours or give the manager other work that brings in money.
- If you handle prices yourself, an honest estimate is how much you earn per hour when you work on purchasing, product research or supplier negotiations. Freed-up time is only valuable if you spend it on that kind of work.
- Nights and weekends. Prices usually aren't tracked by hand at all during these hours. Here the bot doesn't save time; it adds working hours that didn't exist before. That shows up in the "extra orders" and "price raise" terms, not in "time".
Manual entry errors
There's a saving you can't calculate in advance: errors when changing prices by hand. An extra zero, a price entered on the wrong product, a discount someone forgot to remove after a promotion. Recall such cases over the past six months, add up what they cost and divide by six: that's roughly what manual entry costs per month. A bot doesn't eliminate errors entirely, since you can make a typo in the minimum price too. But there are far fewer manual operations.
Seasonality and pricing
If your assortment is seasonal, review the subscription by season. Off-season products with almost no competitors and few sales can be disconnected from the bot: in the per-product model this cuts costs immediately, and in the package model it may let you switch to a smaller package. Conversely, before the season and before major Kaspi promotions, connect products in advance, with verified minimum prices, not on the first day of peak demand.
How to verify the calculation during the trial
Don't take your own hopes or the promises on a service's website on faith. It's easier to check.
- Two groups. 20–40 products with competitors, split into groups that are similar in sales and price. Group A runs with the bot, group B without.
- Baseline. Orders, average selling price and net profit for both groups over the 7 days before the test.
- Test. 7 days with the bot on group A.
- Effect:
Effect = (A's profit during the test − A's profit in the baseline) − (B's profit during the test − B's profit in the baseline)
- Scale to a month and to the full assortment. Multiply the weekly effect by the number of weeks and by the ratio of the number of products you'll connect to the number of products in group A. This is a rough estimate: the most competitive products may produce a bigger effect than the rest, so the test result on them can't be mechanically extended to the whole assortment.
- Compare with the subscription.
Separately record how many times the bot raised the price and by how much. Before a test, this term is easy to underestimate: when pricing by hand, price raises simply go unnoticed.
Comparing different services' pricing
| Question | Why it matters |
|---|---|
| What does it cost per product you'll actually connect? | Packages look cheaper than they are |
| Is raising the price included? | Without it, the "price raise" term is zero |
| Is there a trial period, and how long is it? | A week is the minimum for the test plan above |
| Is a card required for the trial? | Risk of forgetting to cancel |
| What does the subscription include besides price changes? | Analytics, advertising, other platforms |
| How does the price change as the assortment grows? | Moving between packages can be a jump |
Common mistakes
- Comparing subscriptions by the package's upper limit. Calculate the price per product that actually runs.
- Measuring the effect by the store's overall growth. Seasons and promotions bring growth without a bot; you need a control group.
- Forgetting margin loss. On an active card, the bot more often sits closer to the minimum price.
- Ignoring price raises. In our example, this term is comparable to extra orders.
- Counting time that isn't freed up. If the manager doesn't take on other work, there's no saving.
- Connecting products without competitors. The price per active product goes up; the effect doesn't.
Checklist
Start with the simplest thing: how many of your products have competitors and actually need a bot. Everything else depends on that number, and it may turn out to be much smaller than your catalog. Then:
- The price per active product is calculated for each pricing option.
- You know the average profit per order for these products.
- The break-even point is calculated: subscription ÷ profit per order.
- The test was run with a control group.
- The effect includes both price raises and margin loss.
- Time saved is counted only if it was genuinely freed up.
- You've checked how the price will change as your assortment grows.
How to automate it
SaleScout's Kaspi bot pricing, according to salescout.me as of Oct 15, 2026:
| Plan | Price per month | Products | Price per product with a full package |
|---|---|---|---|
| «Гибкий» ("Flexible") | 1,000 ₸ per product | from 1 | 1,000 ₸ |
| «Мини» ("Mini") | 35,000 ₸ | up to 100 | 350 ₸ |
| «Стандарт» ("Standard") | 50,000 ₸ | up to 500 | 100 ₸ |
| «Бизнес» ("Business") | 80,000 ₸ | up to 800 | 100 ₸ |
| «Про» ("Pro") | 100,000 ₸ | up to 1,300 | ≈ 77 ₸ |
The prices in the table may go out of date, so check them on salescout.me before paying. Using the formula above, "Flexible" works out cheaper than "Mini" while you have fewer than 35 products in the bot. The bot itself checks prices every 20 seconds, never goes below your minimum price, and raises your price when a competitor goes up or leaves. Kaspi Marketing AI currently comes free with the bot. The trial is 7 days with no card, and the control-group test from this article fits into exactly that week.
FAQ
How much does a Kaspi bot cost?
Every service does it differently: some charge per product, some by a package of "up to N products", some for the whole store. To compare, divide the subscription by the number of products you'll actually connect. Check prices on the service's own website before paying, because they change.
How do I know whether a Kaspi bot will pay off?
A quick estimate is subscription ÷ profit per order: at 35,000 ₸ and 1,500 ₸, that's about 24 extra orders a month. The full formula is longer: extra orders × profit per order + price raise × orders + time saved − margin loss − subscription. If the result is above zero, the bot pays off.
How can I check a bot's effect before paying?
During the trial period, take 20–40 products and split them into two similar groups. The bot runs only on one. After a week, compare how much net profit grew in each; the difference is the bot's contribution, with seasonality and promotions factored out.
Do I need to connect the bot to all my products?
No. A bot makes money where there are competitors and prices change often. On products without competitors it does almost nothing, but it raises the price per active product on a package plan.
Which is better value: paying per product or a package?
Work out at what number of products the package becomes cheaper: package price ÷ price per product. For example, if one product costs 1,000 ₸ a month and a package of up to 100 products costs 35,000 ₸, the package is better value from 35 connected products.
Does a Kaspi bot have a free trial?
Terms vary between services. You can try the SaleScout Kaspi bot free for 7 days, no card required. Use that week not to "look at the interface" but for a control-group test: that way you'll get an estimate of the effect in tenge before your first payment.
Is a bot worth paying for if I have fewer than 20 products?
Look at your cards, not your store size. If competitors on your products change prices several times a day, a bot can pay off even on a few products, especially on a per-product plan where you're not paying for an empty package. If a competitor only shows up once a week, managing by hand is easier and cheaper.
Why might a bot not pay off?
If your cards have few competitors, your minimum price is almost equal to the market price, your rating or quality metrics keep you from reaching the top, or you have too few products for the package you chose. In those cases, solve those problems first, then connect a bot.
Sources
All links checked Oct 15, 2026. The sources are in Russian; original titles are given with English translations.
- SaleScout, pricing on the home page — https://salescout.me/ru (snapshot of Oct 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.
