ECB’s Daily Price Dataset: millions of web-scraped prices show why ‘effective price’ beats sticker price

Central banks are increasingly using web-scraped online prices because they are high-frequency, timely, and closer to the “real” shopping experience than monthly snapshots.

Between 2024 and 2025, the Eurosystem (via the ECB/PRISMA network) published multiple pieces showing how daily online price data can measure price dynamics and even help nowcast food inflation.

What the research built (DPD / PRISMA)

Across Eurosystem publications, a consistent story appears:

• The ECB collected a new daily dataset via web scraping of online supermarket prices to study how price changes behave under high inflation (evidence covering April 2022 to January 2024).
• Banque de France describes a “Daily Price Dataset (DPD)” that collects thousands of prices every day, accumulating several million prices for close to 100,000 food products from supermarkets in large euro-area economies.
• Other central banks (e.g., Malta) highlight that web-scraped supermarket prices can support real-time monitoring / nowcasting of food inflation.

Why this matters for competitive price monitoring

The big lesson is simple: the “price” that matters is rarely a single number on the PDP.

High-frequency datasets work because they track the full offer reality:

• Availability changes (in-stock vs out-of-stock)
• Promo mechanics (temporary discounts, badge changes, coupons)
• Landed price (shipping thresholds, fees, delivery promises)
• Assortment churn (products disappear, variants change)

If central banks need daily granularity to understand inflation, you will need context + cadence to understand competitors.

What to copy for Trackabl (actionable)

1. Track landed price
   Store item price + shipping + free-shipping threshold + delivery ETA.

2. Track availability
   Log in-stock status and “only X left” scarcity claims.

3. Separate event types
   “Real markdown” vs “promo framing” vs “shipping threshold change” vs “availability change”.

4. Use sampling + cooldowns
   Daily scraping is powerful, but alerts should be throttled to avoid noise.

A practical benchmark

If you can answer these with data, your pricing intelligence is already closer to how researchers do it:

• Did the competitor really cut price, or did shipping/fees change?
• Did the offer change because the product disappeared/reappeared?
• Was the “discount” just a badge change?

Takeaway: high-frequency research confirms what merchants feel: effective price beats sticker price.