Every shop writes prices, sizes and stock in its own way. Scrapewise group rules turn them into one clean format: the price as a number, the pack size in grams or millilitres, price per kg or per litre, prices in EUR, and one row per product page. Then the rows are ready to match and compare.
"1,29 €", "EUR 1.29" and "1.29 EUR/tk" are the same price written three ways. A spreadsheet sorts them as text, and a chart cannot read them at all.
A 500 g bag at EUR 1.29 is cheaper per kilo than a 1 kg bag at EUR 2.79. Without price per kg the comparison rewards the smallest pack.
A Polish, Swedish or Swiss shop shows its own currency. Until every price is in EUR, the cheapest offer is a guess.
The same product page shows up twice in a link list, and titles carry "-20%!" or "NEW" that break matching. Both need cleaning before any comparison.
Rule types: number from text, price per unit, currency, map values, clean text
Conditions per rule, so one group can hold groceries, drinks and multipacks
Free requests on every new account, no card needed
Cleaning is step 2 of 4. It runs on every scraped row, so the matcher and your price report work on the same clean columns from every shop.
Collect the shops you compare against with a ready data API or an AI scraper for any site. Use a category, a link list, a sitemap or an uploaded file, on demand or on a schedule.
Group rules run on the rows of every scraper in the group: take the number out of a text, compute price per kg, litre or piece, convert prices to EUR, map stock words to yes or no, and trim noise from titles.
The clean rows are linked to your own product list: same EAN or ID first, then text columns such as title, brand and size, then product pictures. Unsure pairs wait for review.
Your product next to each competitor, with the same price per unit and the same currency. Export as CSV or Excel, read it by REST API, or let an AI agent read it over MCP.
Group rules write new clean columns next to the scraped ones. The scraped value is never overwritten, so you can always check where a clean value came from.
Reads the number written right before a unit and converts it: "Oat flakes 0.5 kg" gives 500 grams, "Juice 1 l" gives 1000 ml. Ready unit lists cover grams, kilos, ounces and pounds, millilitres to litres, and pieces.
Divides the price by the pack size and scales it to the unit you compare on: per kg, per 100 g, per 50 g, per litre or per piece. You pick the reference number, so one rule fits any product type.
Converts prices in other currencies into one currency, such as EUR, with current exchange rates, so offers from shops in different countries compare on one scale.
Turns shop-specific words into one value: "Laos", "In stock" and "Saadaval" become true, "Out of stock" becomes false. Useful for stock, condition and category names.
Keeps only the part of a text you need, for example the title without "-20%!" or the model code from a long product name, so titles are ready for matching.
A wrong unit price is worse than an empty cell, because it looks real in every report. The rules are built to leave a cell empty rather than guess.
Multipacks like "3 x 100 g", ranges like "50-100 g" and unclear numbers like "1.000 g" give no value instead of a wrong one. An empty cell is easy to filter for; a wrong number is not.
Each group rule can have up to 3 conditions, such as "product type is yarn" or "pack size is not empty". So yarn is compared per 50 g and coffee per kg in the same group.
If the same product page appears twice in one run, it is kept once and the dropped address is reported, so a price is never counted twice. The same product in different shops is linked by matching.
Any scraped or uploaded rows go in. Clean, typed columns come out, next to the original values.
Rows from any scraper in the group, or a CSV / Excel file you upload as a source.
KAERAHELBED 500G -20%!, 1,29 €, LaosThe reference unit for the price per unit column.
Prices in other currencies are converted to it.
EUROne clean row per product page, 8 columns each
CSV or Excel export, REST API, or MCP for AI agents.
Example rows in the style of a grocery price comparison (illustrative data, not a customer's). The left column is what the shop shows; the right is what the group rules write.
| Column | As scraped (before) | After cleaning | Rule |
|---|---|---|---|
| Price | 1,29 € | 1.29 | Take a number from text |
| Pack size | KAERAHELBED 500G -20%! | 500 (grams) | Take a number from text, grams |
| Price per kg | not shown by the shop | 2.58 | Price per unit: 1.29 ÷ 500 × 1000 |
| Price in EUR | PLN 5,49 (shop in Poland) | EUR value at the current rate | Currency conversion |
| In stock | Laos | true | Map values |
| Title | KAERAHELBED 500G -20%! | KAERAHELBED 500G | Clean text with a pattern |
| Multipack | 3 x 100 g | left empty, not guessed | No guessing |
| Same page twice | 2 rows, same address | 1 row, the drop is reported | One row per product page |
Price per kg, per litre or per piece across every pack size, so the cheapest offer is really the cheapest.
Offers from shops in different countries in one currency, ready for one chart or one report.
Clean titles, sizes and numbers make matching to your product list more reliable, and one row per product page keeps counts honest.
Write the cleaning once as group rules instead of fixing cells by hand. In Scrapewise a rule takes the number out of a price text, reads the pack size, computes price per unit, converts the currency and trims title noise. The same rules run on every shop and every new run.
Put the shops in one group and add rules that write the same clean columns for all of them: price, pack size, price per unit, currency and stock. Each scraped value stays next to its clean copy, so every shop ends up in one format you can compare and check.
Compare price per unit, not the shelf price. First take the pack size out of the title, for example 500 g, then divide the price by it and scale to per kg, per litre or per piece. Unclear sizes such as multipacks are left empty instead of guessed.
There are two kinds. The same product page twice in one run is kept once, and the dropped address is reported. The same product sold by different shops is not a duplicate to delete: product matching links those rows to one product in your own list.
Scrape one shop, add a price per kg rule, and see the clean column next to the raw one. Every new account gets 5 free requests; after that you pay as you go from a prepaid balance. bebo.ee, a grocery price comparison, runs on the same scrape, clean per kg and match flow. bebo.ee compares 12 grocery e-shops: more than 130,000 products, and about 36,000 of them are matched across shops so their prices can be compared.
What the cleaning rules do, what they refuse to do, and how they fit with matching.
Product data cleaning turns the text a shop shows into values you can compute with: a price as a number, a pack size in grams or millilitres, a price per kg or per litre, one currency and one word for stock. In Scrapewise you set it up once as group rules, and every scraper in the group follows them on every run.