Custom data collection · one public source
Get the rows you need, with their sources attached.
Tell me which public website, which fields and what you will use the data for. I return a structured CSV and Excel workbook, a short field guide and checks you can inspect.
Request a USD 35 starter →Starter scope
USD 35
One publicly accessible source, up to 1,000 available records and five agreed fields, within three business days after scope is confirmed.
If the source cannot supply a field or is blocked, I report that before accepting the order. No invented values or guaranteed record count.
What you receive
- A UTF-8 CSV and an Excel workbook, with one record per row and a source URL for each row.
- Selected fields cleaned and consistently named; missing values left explicit.
- A short data dictionary: field meanings, source, collection time and limits.
- Counts of records, unique source URLs and incomplete rows.
- One clarification round on the delivered file.
What I need from you
- The public source URL and the exact fields you need.
- A clear filter, such as category or location, and your intended record count.
- Your preferred column names or a sample output row if you have one.
The starter covers a single collection. Recurring monitoring, private accounts, personal contact lists, paid data sources and bypassing access restrictions need separate review and pricing.
Independent demonstration · 1 October 2026
50 public pages → 1,000 checked rows
I collected 50 catalogue pages from Books to Scrape, a practice website that says it welcomes scraping. The run returned 1,000 complete records and 1,000 distinct source links, in 33 seconds on 1 October 2026. Both exports were re-read and compared with the collected records. Its prices and ratings are fictional and have no market meaning.
| Title | Price | Availability | Source |
|---|---|---|---|
| A Light in the Attic | £51.77 | In stock | Product page |
| Tipping the Velvet | £53.74 | In stock | Product page |
Download all 1,000 rows (CSV) → · Excel → · Run checks →
The script rejects HTTP failures, missing fields, duplicate source links, empty pages and pagination loops. Eight tests exercise normal results and these refusal cases without sending requests. A real order gets its own checks based on the agreed source and fields.
Personal report example · calculated 1 October 2026
Find matching prices. Identify new products.
A useful catalogue report separates two questions: which current prices match your rule, and which product identifiers appeared after the previous complete collection.
102 matching prices
Of the 1,000 saved training records above, 102 have a last minor-unit digit of 1: for example £26.41. The rule uses exact decimal values. These fictional prices are a demonstration, with no claim of a real discount.
Download the 102 matching rows →Zero new products on an identical rerun
The first complete snapshot establishes a reference. Reading that same snapshot again adds nothing. A new identifier is checked in a separate test fixture. This example analyzes one saved snapshot; it does not compare two live days.
Five executable checks cover decimal prices, invalid prices, the first read and identical rerun, a new identifier, and refusal of empty or duplicate snapshots. A failed collection must keep the previous reference instead of making a catalogue look empty.
Dated calculation → Source snapshot →
Recurring reports are a separate scoped job. I first verify your source, currency, current-price field, product identifiers and expected frequency, then agree on the complete collection, hosting and handover. No customer retailer or daily scheduler was tested in this example.
Can you collect from any website?
No. I inspect the source and fields first. Availability, access rules and site changes affect what can be delivered. I confirm the exact scope before payment.
Can this run every day?
Yes, as a separate scoped job when the source supports it. We would agree on schedule, hosting, failure alerts, ongoing cost and handover before I build it.