Screen recording: source page → field mapping → verified export → scheduled delivery.
See the market every day
Public web data for business — collected, verified and delivered to where your team works.
You name the sources, fields and frequency. We handle collection, cleaning and validation.
Fixed quote before work begins
We build the sample export on your sources
| Source | SKU | Price | Δ 24h | Stock |
|---|---|---|---|---|
| Marketplace A | SKU-40118 | 14 900 | 0,0% | 312 |
| Marketplace B | SKU-40118 | 12 150 | −18,5% | 47 |
| Retailer C | SKU-40118 | 15 290 | +1,9% | 128 |
| Marketplace A | SKU-40119 | 8 900 | 0,0% | 0 |
| Catalog D | SKU-40119 | 9 420 | −3,1% | 76 |
Sample structure. We agree the fields and the source list before collection starts.
We are not selling a platform you have to learn. We deliver a specific set of data that closes the gap between what your team needs to know and what it actually sees.
“We find out about a competitor's promotion three days after it starts”
We check competitor prices and stock as often as hourly, with an alert the moment someone drops below your floor price — not a report at the end of the week. The signal arrives where your team already sits: Telegram, email, your own spreadsheet.
“We have no idea what our dealers actually list, or at what price”
Listing-level monitoring across every platform: assortment, availability, new arrivals, search position. An RRP breach becomes a row in a file rather than a rumor from a sales rep. You also see where the category is empty.
“Our CRM is full of two-year-old contacts”
Company data from open registries and public sources: segmentation, enrichment, database refresh. Your sales team works from a list that is actually alive.
“We need a defensible number for a market we are not in yet”
Regular industry snapshots: price movement, assortment depth, who enters and who leaves. One method in every period — which is why the trend can be trusted.
“We need training data our lawyer will sign off on”
Clean, structured datasets from public sources, with documented provenance and jurisdiction for every source. A written legal position is available per project.
Of the six steps, only three are yours, and none takes more than half an hour. Everything else — feasibility assessment, legal review, scrapers, cleaning, monitoring, repairs — happens on our side.
Write to us: which sites, what data, how often, and what you will do with it. Links are enough — no technical specification needed.
Our lawyers review the jurisdiction of the sources and of your company. In parallel we work out the technical cost: static pages are cheap, pages rendered with JavaScript are not. You get one figure, not a range.
Not a demo dataset from someone else's project. We build scrapers for your actual sources and send you a real export in an agreed field structure.
Open it next to the source and compare. If a field is wrong, missing or awkwardly structured, this is the cheapest moment to change it.
Collection on a schedule into the channel you choose: a file, an API, writing straight into your database, a cloud export. Every batch is validated before it is sent.
Sites change their layout, fields move, pages start rendering in the browser. We track breakages and fix the scrapers within the same service — and we warn you when a change at the source affects what your numbers mean.
| In-house team | Working with us | |
|---|---|---|
| Path to the first data | Hiring, then infrastructure, then the first scraper | A sample on your sources first, then regular delivery |
| Ongoing costs | Salaries, servers, proxy infrastructure, on-call duty | One line per month, fixed at signing |
| When a source changes its layout | A task competing with your product roadmap | Our problem, within the same fee |
| Legal position on collection | Your lawyers, source by source, from scratch | Reviewed before the start, in writing on request |
| When the person who built it leaves | The expertise leaves with them | The contract is with a company, not an individual |
Below is what the result looks like once we are live. Not an abstract diagram, but what a category manager, a department head and an analyst actually open.
| Source | SKU | Price | Δ 24h |
|---|---|---|---|
| Marketplace A | SKU-40118 | 14 900 | 0,0% |
| Marketplace B | SKU-40118 | 12 150 | −18,5% |
| Retailer C | SKU-40118 | 15 290 | +1,9% |
| Catalog D | SKU-40119 | 9 420 | −3,1% |
| Marketplace A | SKU-40120 | 21 300 | 0,0% |
We agree the field structure before collection starts, so nothing has to be reworked at acceptance. Choose the channel by who will use it: an analyst, a developer, a data team or a manager.
source,sku,name,price,currency,delta_24h,stock,checked_at
Marketplace A,SKU-40118,Model X 500 ml,14900,RUB,0.000,312,2026-09-09T06:00:04Z
Marketplace B,SKU-40118,Model X 500 ml,12150,RUB,-0.185,47,2026-09-09T06:00:06Z
Retailer C,SKU-40118,Model X 500 ml,15290,RUB,0.019,128,2026-09-09T06:00:09Z{
"sku": "SKU-40118",
"checked_at": "2026-09-09T06:00:06Z",
"offers": [
{ "source": "Marketplace B", "price": 12150,
"currency": "RUB", "delta_24h": -0.185,
"in_stock": 47, "url": "https://…" }
]
}-- we write, you just query. no intermediate export. INSERT INTO prices_daily (source, sku, price, currency, delta_24h, in_stock, checked_at) VALUES ('Marketplace B','SKU-40118',12150,'RUB',-0.185,47,'2026-09-09 06:00:06');
delivery: cloud: s3://your-bucket/prices/dt=2026-09-09/ bi: power_bi, refresh 06:15 alerts: - rule: competitor_price < your_floor_price channel: telegram, #price-alerts - rule: weekly_summary channel: email, category manager
| Field | Type | What it means |
|---|---|---|
| source | string | Which site or marketplace the row was read from |
| sku | string | Your identifier, matched to the source listing during setup |
| price | number | The price a buyer sees in the region you specify |
| delta_24h | number | Change since the previous export, calculated on our side |
| stock | integer | Availability exactly as the source publishes it |
| checked_at | date and time | When the page was actually read, not when the file was sent |
Small products that live alongside your process instead of adding a new one. Built on the same data feed, as a separate project with a fixed scope.
Fresh data goes straight into Claude, GPT or your company's own agent. An employee asks a question in plain words and gets an answer based on this morning's data, not on whatever was in the model's training set.
A price dashboard, an RRP violation monitor, an assortment tracker. Only the screens your team will actually open, with your access rules and your logic. Nothing extra to pay for or configure.
We connect the data feed to the analytics tools you already use and set up notification rules: a dumping alert in Telegram within minutes, a weekly summary emailed to your manager.
Almost everything we do is built for a specific task. Cobi is the exception: a finished product for teams that sell on marketplaces. It does not replace the category manager. It takes over the part of the job that consists of refreshing browser tabs.
The assistant chat works in Russian. If you need a platform that is not on the list, write to us and we will see what can be done.
Collecting public data is lawful, and on its own that answer is worth little. What matters to your lawyer is something else: who checked, under which jurisdiction, and whether it can be seen in writing. All three are part of the service.
We document the sources and the intended use before anything is built.
The legal regime of the source's country and of your company's country. Checked by a cyber-law firm that supports every one of our projects.
Personal data, platform terms, copyright and database rights — in relation to what you intend to do with the result.
The project goes into work, and the written legal position on it is available to you on request — before the contract is signed, not after.
This is the line that separates a vendor your compliance team will approve from one it will turn down.
If a source can only be collected by crossing one of these lines, we say it cannot be done and suggest an alternative source. That conversation happens before work begins, not after you have paid.
Cost depends on three things: how many sources, what share of pages need a full browser to render, and how often you need the data. We give you a single number once we have worked through your task: no hourly billing and no extra charges when a source breaks.
Collecting data that a website publishes to any visitor without a login is lawful in the jurisdictions where we work. But we do not rely on that general answer: every project is reviewed by a cyber-law firm, covering the legal regime of the source's country, of your company's country, and how you intend to use the result. If the opinion is negative, we say so plainly and propose a different source.
Most of what we deliver — prices, listings, stock, catalog structure — contains no personal data at all. If a project touches company information or contacts from open registries, the legal review covers the legal basis, retention periods and your obligations as the data controller. We will tell you plainly if a request cannot be fulfilled as it is worded.
You set the frequency: hourly, daily, weekly, or a one-off snapshot. We first build a sample on your sources and agree the set of fields, then start regular delivery. Sources that render entirely in the browser take longer to build and cost more per page — we tell you which of yours fall into that category before we name a price.
It happens constantly, and it is the main reason most in-house data collection projects quietly die. We monitor for breakage, fix the scrapers within the monthly fee, and warn you separately when a change alters not where a field sits on the page but what it means.
The result belongs to you: you can use it, store it and sell derived analytics within the limits set out in the project's legal review. We sign NDAs as a legal entity. Your field structures, source lists and SKU mappings are never used for anyone else.
That is exactly the route we recommend. Take a pilot: one source, your real SKUs, a fixed price. You receive a genuine data export that you can check against the live pages, plus a coverage report — what we managed to collect and what we did not. If the quality does not suit you, that is where it ends.
It depends on the number of sources, how complex the pages are to render, and the frequency. These three things determine everything else — which is why we price against your task rather than publish a price list that would be wrong for your case anyway. The estimate comes as a fixed figure that already includes scraper fixes and legal review.
The infrastructure is designed to behave predictably: a reasonable request rate, requests spread out over time, and retries after a failure instead of hammering. We do not pose as an authorized user and we do not create load that would stop a site from serving its own visitors. This is a design constraint, not a courtesy.
The practice of collecting public data, the legal side of it, breakdowns of specific platforms, and how category teams work with the numbers.
A single product listing can show up to six prices: base, strikethrough, discounted, coupon, card, and unit price. We look at which one to use for which decisions, and where mistakes are most common. For category managers and anyone setting up price monitoring.
Read the article →Where the line falls between open and closed data, what the courts have decided on the subject, and what does not become permitted simply because the data is open. For lawyers and executives deciding how to work with web data.
Read the article →What you can see without logging in on Uzbekistan's largest platform and which tasks that data solves: prices, control of brand sellers, new arrivals, reviews. For brands and distributors that sell in Uzbekistan.
Read the article →Two or three links to sources and one sentence about the decision you are trying to make is enough. In return you get a collection plan, a delivery format and a fixed quote.
We work across time zones. If your legal or procurement team needs documents before work begins, let us know and we will send the company details, an NDA template and a sample legal opinion in advance.
