Copying a web table works until you notice the Next button. Your selection contains the current page, while the report's total includes many more records. Repeating the copy-and-paste process makes it easy to miss a page or paste one twice.
Before collecting the rows yourself, look for an export provided by the website. If you need data from the view already open in your browser, Sidelight AI Assistant can try to collect a table across multiple pages and save the rows it finds as CSV. A useful result includes a clear account of the pages reached, followed by checks against the source.
Define what “all pages” includes
Filters determine which records belong in your file. A report for one week is different from the same table without a date filter, even if the columns match. Set the period and status filters first, then note the total displayed by the website, if one is available.
Pagination also differs from infinite scrolling. A paginated table uses controls to move between sets of results. Microsoft's web-extraction documentation describes configuring pagination as part of collecting structured data. With either mechanism, a short initial run helps reveal navigation and column problems before you request a larger collection.
Collect and check the rows
-
Stabilize the view. Apply the filters and choose an appropriate sort order. If records are changing while you work, narrower date or status filters may make the collection easier to check. Do not change the search or sorting halfway through a run: records can move between pages and create gaps or overlaps.
-
Choose a record identifier. Include an order number, product code, or item URL where available. Names alone are often ambiguous. Also name the columns you need and their order. Keep currency separate from amount, and distinguish creation date from due date if both appear in the report. These choices make later duplicate checks much more reliable.
To test the navigation before repeating manual copies, try Sidelight AI Assistant with a small section of your table. Use the first request to check the fields as well as the transition between pages.
-
Ask for a bounded run. Start on the first results page and explicitly include it in the requested total. “Three pages including this one” avoids confusion between three pages overall and three additional pages. For an order report, a practical request is:
Extract this orders table from the first three results pages, including the current page. Include order number, creation date, amount, currency, and status. Combine the rows into one CSV. Report which pages you reached and how many records you obtained. If navigation fails, explain where you stopped rather than presenting the file as complete.
-
Inspect the boundaries between pages. Find the final record on the first page and the first record on the next page in the file. Check that headers have not been included as records. Compare a few identifiers and amounts with the website, including a record with a missing field. If the sample is wrong, correct it before extending the run.
-
Expand the scope and import the CSV. Once the sample matches, request the number of pages needed while preserving the column definitions and filters. The Sidelight AI Assistant export guide covers scope and field selection. In Excel, import the downloaded file using the text/CSV data options and review its structure before loading it. Microsoft's CSV import guidance explains the available approaches.
Recognize an incomplete export
Compare the row count with the website's filtered total. Do not assume every page contains the same number of records: the final page can be shorter, and some sites repeat featured items. If there is no total, the pages reached and the boundary checks provide useful evidence, but they do not prove that every available record was captured.
If rows are missing, ask where the run stopped. An unrecognized navigation control, unfinished loading, or an expired session may explain the difference. Resume from a known point only after checking the overlap with the existing file.
Frequently asked questions
Should I remove every repeated name? No. Different orders can belong to the same customer, and products can share names. Compare stable identifiers and the relevant details before deleting a record.
Will the export be an XLSX workbook? This workflow produces CSV. After importing and checking it, you can save an Excel workbook with the formatting and formulas you need.
Does the file stay connected to the website? No. It represents the records collected during that run. It does not keep following pagination when the report changes.
Start with a few pages, verify their boundaries, and expand using the same definitions. Try Sidelight AI Assistant on your paginated table and request the columns you need in one CSV.