Guide 04 of 12

How to Extract Tables Using Instant Data Scraper

Extract HTML tables from any public web page with Instant Data Scraper: headers, merged cells, links, multiple tables, multi-page tables and Excel exports.

10 min read · Updated · By the Instant Data Scraper team

How to Extract Tables Using Instant Data Scraper

Tables are the easiest kind of web data to extract and the most useful: rankings, statistics, timetables, price lists, sports results, specifications. But copying a web table by hand often goes wrong — headers break, links disappear, merged cells shift columns, and large tables take ages.

This guide shows you how to extract tables accurately with the Instant Data Scraper tool, how the tool handles difficult tables, what to do when a "table" isn't really a table, and how to collect tables that are split across several pages.

What counts as an HTML table?

A real HTML table is built with dedicated tags:

<table>
  <caption>Largest cities</caption>
  <thead><tr><th>City</th><th>Population</th></tr></thead>
  <tbody><tr><td><a href="/city-a">City A</a></td><td>12,500,000</td></tr></tbody>
</table>
  • <tr> is a row.
  • <th> is a header cell, <td> a data cell.
  • <caption> is the table title.

Tables that only look like tables

Many modern sites draw grids with <div> elements and CSS. They look identical to a table but contain no table tags. Table mode will not find them — Automatic mode or CSS selector mode will. Right-click the grid and choose Inspect to check which kind you have.

Using HTML table mode

Why choose Table mode

Automatic mode already detects tables, but Table mode ignores everything else on the page. That gives a cleaner list of datasets when a page is full of menus, cards and widgets.

What the tool does for you

  • Header detection — reads headers from <thead> or a first row made of <th> cells. Without headers, columns are named Column 1, Column 2…
  • Merged cells — a cell with colspan="2" is repeated across the columns it covers, so rows stay aligned.
  • Links in cells — when a column contains links, an extra column such as City link holds the full URL.
  • Layout tables ignored — tables used only for page layout, or with fewer than two rows, are skipped.
  • Every table separate — each data table becomes its own dataset, ranked by size, with its caption or ID in the tab name.
  • Character encoding — pages in older encodings are converted to UTF-8, so accented characters stay correct.

Pages with many tables

Wiki pages, sports statistics and financial reports often contain dozens of tables. After extraction:

  1. Look at the dataset tabs — each shows the table's caption and row count.
  2. Click through the tabs to find the one you need.
  3. Export each table separately; every export contains only the selected dataset.

Numbers, currencies and dates

Values are exported as displayed on the page. That keeps your data faithful to the source, but means some cleaning in Excel:

Plain numbers

Values like 12500000 or 3.75 become real numeric cells in Excel exports, ready for formulas.

Formatted numbers

Values with separators or symbols (12,500,000, $43.99, 5 kg) stay as text. Convert them with:

=VALUE(SUBSTITUTE(SUBSTITUTE(B2,"$",""),",",""))

Footnotes and references

Cells like 1,234[3] include footnote markers. Use Find & Replace with wildcards, or =LEFT(B2,FIND("[",B2)-1).

Dates

Different sites use different date formats. Use Data → Text to Columns or DATEVALUE to convert them consistently.

Tables split across pages

Large tables are often paginated ("Showing 1–50 of 1,200"). To collect them:

  • Open Extraction options and tick Support pagination.
  • Set the maximum number of pages.
  • Leave the Next-page selector empty for automatic detection, or enter one such as a.next.

Rows from each page are appended to the same dataset because the tables share the same headers. See How to Scrape Multiple Pages Automatically.

When table extraction goes wrong

ProblemCauseFix
Table mode finds nothingGrid built from divs, or loaded by JavaScriptTry Automatic, then JavaScript page mode
Columns shiftedComplex rowspansCheck affected rows in Excel; use CSS selector mode for full control
Totals row in dataFooter row in tableDelete the last row after export
Header missingTable uses td for headersRename columns in Excel
Only first pagePagination not enabledEnable pagination

Common kinds of web tables and how to handle them

Statistics and rankings

Clean header rows and numeric columns. Table mode works perfectly; convert formatted numbers in Excel.

Comparison and specification tables

Often transposed — features as rows and products as columns. After export, use Excel's Paste Special → Transpose to flip the table if you prefer products as rows.

Timetables and schedules

Frequently use merged cells for days or times. The tool repeats merged values across columns; for rowspans (cells merged vertically), fill blanks in Excel with Go To Special → Blanks and =A1 (the cell above), then paste as values.

Financial tables

Negative numbers may appear as (1,234). Convert with =-VALUE(SUBSTITUTE(SUBSTITUTE(SUBSTITUTE(B2,"(",""),")",""),",","")) where the value starts with a bracket.

Sports results

Score cells like 3–1 can be split into two columns with Text to Columns using the dash as the delimiter.

Automatic mode vs. Table mode

AutomaticHTML table
Finds real tablesYesYes
Finds div-based grids and cardsYesNo
Number of datasetsUp to 6 best candidatesEvery data table on the page
Best forUnknown or mixed pagesPages with many tables

Advanced: extracting one part of a table

Sometimes you only need some rows — for example a single section of a long table. Use Custom CSS selector mode:

Row selector:  table.wikitable tbody tr
Fields:
Name: td:nth-child(1)
Value: td:nth-child(3)
Link: td:nth-child(1) a@href
  • td:nth-child(3) picks the third cell in each row.
  • Change table.wikitable to the class or ID of the table you want.
  • Header rows without td cells produce empty rows, which are skipped automatically.

Verifying a table extraction

  • Row count — compare with the number of rows shown on the page (excluding headers).
  • Column alignment — check a row that contains merged or empty cells.
  • Totals — if the page shows a total, sum the column in Excel and compare.
  • Encoding — look at names with accents or symbols.

What people do with extracted tables

  • Build charts of statistics for reports and presentations.
  • Combine tables from several pages into one dataset for analysis.
  • Compare specifications across products.
  • Track changes in rankings by extracting the same table monthly.
  • Feed reference data (such as country codes or units) into other systems.

Walkthrough: building a comparison from three tables

Suppose three public pages each contain a table of country statistics for a different year, and you want one combined sheet.

  1. Extract the first page in HTML table mode and choose the statistics table from the dataset tabs.
  2. Untick link columns you don't need and export to Excel. Add a column called Year and fill it in.
  3. Repeat for the other two pages, keeping exactly the same columns.
  4. Paste all rows into one sheet — or use Power Query From Folder to combine the files automatically.
  5. Insert a PivotTable with Country as rows, Year as columns and the statistic as values.

Within a few minutes you have a year-by-year comparison that would have taken an hour to type by hand.

Table terms glossary

  • thead / tbody / tfoot — the header, body and footer sections of a table.
  • colspan — a cell that spans several columns.
  • rowspan — a cell that spans several rows.
  • caption — the table's title, used to name the dataset tab.
  • Layout table — a table used only to position content, not to hold data.

Why table extraction saves so much time

Copying a table by hand seems quick until you try it on a real page. Browsers copy hidden elements, footnote markers and formatting along with the values. Links disappear entirely, and merged header cells push data into the wrong columns. On a table with a few hundred rows, fixing those problems can take longer than typing the data from scratch.

A structured extraction avoids all of that. Each cell becomes exactly one value, header names are preserved, links get their own columns and the result is identical every time you run it. That repeatability matters: if you extract the same table next month, the columns will line up perfectly with last month's file, so you can compare the two without any manual re-arranging. For anyone who regularly works with published statistics, price lists or rankings, table extraction quickly becomes one of the most useful tools in their kit.

Who extracts web tables

Journalists and researchers pull statistics from public reports and encyclopaedia pages to check facts and build charts. Sports fans and analysts collect league tables and results. Finance professionals gather published figures such as exchange rates or index constituents. Teachers prepare datasets for classroom exercises. In each case the same three steps — Table mode, choose the dataset, export — deliver clean data in seconds.

A note on accessible tables

Well-built tables use header cells, captions and scope attributes so that screen readers can announce them correctly. Those same features help scrapers: headers become column names and captions become dataset names. If you publish tables yourself, building them accessibly also makes them easier for others to reuse — and for you to extract later.

Key takeaways

  • Real tables use table, tr, th and td tags — check with Inspect.
  • Table mode ignores everything else on busy pages.
  • Merged cells are repeated so rows stay aligned; link columns are added automatically.
  • Convert formatted numbers in Excel with VALUE and SUBSTITUTE.
  • Enable pagination for tables that span several pages.

Conclusion

HTML tables are the easiest and most rewarding web data to extract. With Table mode you get correct headers, aligned rows, link columns and proper encoding in seconds, and every table on the page is available as its own dataset. Check whether a grid is a real table, choose the right dataset tab, export to Excel or CSV and convert formatted numbers before calculating. For long tables, enable pagination; for partial tables, use CSS selectors. Following this approach, you can turn almost any published table into analysis-ready data far faster and more accurately than copying by hand.

Step-by-step instructions

  1. Open the page with the table

    Make sure the table is public and fully visible.

  2. Paste the URL

    Paste it into the Instant Data Scraper tool.

  3. Select HTML table mode

    Or keep Automatic if you're not sure the grid is a real table.

  4. Start the extraction

    Click Start Extracting and wait for the preview.

  5. Choose the right table

    Use the dataset tabs — each shows the caption and row count.

  6. Remove unneeded columns

    Untick extra link columns you don't need.

  7. Export

    Download Excel for analysis or CSV for importing.

Example

CityCity linkCountryPopulationArea (km²)
City 1https://wiki.example/City_1Country 1259134771026.1
City 2https://wiki.example/City_2Country 241193506015.2

Population and area are exported as numeric Excel cells; the city link is its own column.

Best practices

  • Use Table mode on busy pages to avoid unrelated datasets.
  • Check the first and last rows for headers or totals mixed into the data.
  • Inspect the grid — if it isn't a real table, use Automatic or CSS selector mode.
  • Keep link columns only when you need them.
  • Enable pagination for long, paginated tables.
Try it on a real page

Open the Instant Data Scraper tool and extract your first dataset in seconds.

Start Extracting

Frequently asked questions