Google Dataset Search

A single query box over every dataset page on the public web that carries schema.org Dataset structured markup - government portals, university archives, Kaggle uploads and commercial aggregators among thousands of repositories. Ask it for 'electronics retail sales', as this research session did, and roughly twenty results come back: a Kaggle contributor's twelve months of electronic store sales in twelve CSV files, a Statista ranking of global electronics retailers, and a Federal Reserve series on US electronics and appliance stores.

Each hit is a metadata record, not a raw table - title, provider, description snippet, format hint, subject tags and an APA-style citation, pointing at the host repository rather than hosting files. That makes it the fastest way to find out what electronics-retail data exists at all before deciding what to license, buy or model against.

Datadory turns that discovery layer into a deliverable: the index filtered down to genuine datasets, each record normalized into the field dictionary below and the underlying files verified before they reach you - delivered daily, weekly, or hourly. [Get a sample of this dataset](#request) and judge the rows, not the promise.

Three real records pulled during research for this page, exactly as the index describes them:

name     : Electronic Store Sales Data
provider : Kaggle              format : zip (4,996,940 bytes)
tags     : analysis, data cleaning | technique, data visualization
span     : 12 months of single-store sales, 12 CSV files

citation : Dhaundiyal (2023). Electronic Store Sales Data [Dataset]

name     : Global Electronics Retailers
provider : Statista             type   : topic / statistics page

name     : Electronics & appliance stores retail series
provider : FRED               type   : official statistical series

One consumer upload, one commercial statistics page, one official series - three different grades of evidence in three lines of scanning. The citation string is the detail worth noticing: every result ships with ready-made attribution, so provenance survives from our pipeline into your deck or paper without retyping.

Seven fields describe every record, verified against live responses during research. Attributes that appear only when a publisher's markup declares them are held under additional fields on request rather than promised in every row.

Geographic - global by construction: any publicly reachable dataset page with schema.org Dataset markup is eligible. Hosts observed during research for this industry include Kaggle and FRED in the United States plus Statista's international aggregator pages, and the same mechanism reaches national statistical offices and university repositories worldwide.

Temporal - set by whatever each indexed dataset covers, not by the index. The electronics-retail sample observed here spans a single store's calendar-year-2023 sales file through publisher-maintained statistical series with decades of history, so temporal fit is evaluated record by record.

Granularity - one record per indexed dataset landing page. Granularity inside any given dataset depends entirely on its publisher - twelve monthly CSVs in the store-sales case, monthly national series in the FRED case - which is precisely what the metadata lets you screen before committing.

API, files, or your warehouse. Daily, weekly, or hourly.

• Assortment and competitor mapping - retailer rosters such as the Global Electronics Retailers listing give strategy teams a starting universe of chains to profile before deeper firmographic buys.
• Forecast prototyping - the twelve-file electronic store sales archive is a realistic sandbox for testing demand models on real checkout rhythms before committing to enterprise transaction feeds.
• Macroeconomic anchoring - official series for electronics and appliance stores tie a company's performance to the category-level benchmark economists and investors already trust.
• Citable sourcing - because every record carries an APA-style citation string, analysts and academics move provenance straight into reports without reconstructing references.
• Teaching and benchmarking corpora - subject tags such as 'technique, exploratory data analysis' make the index filterable for instructional datasets, a shortcut for course builders and bootcamps.

Market researchers and consultants use it as the sweep-the-board step: one query reveals whether an electronics-retail question has public data behind it at all, before anyone scopes primary research. Data scientists and students get small, citable, real-world files - the kind that make honest practice sets and reproducible demos. Investors and quants anchor category models on official store-series benchmarks surfaced alongside the crowd uploads. Journalists and academics lean on the built-in citation strings, which keep attribution intact from index to publication. Developers building data products get uniform metadata - name, provider, tags, format note, citation - so one parser handles every host identically.