To join data across sheets and tables without VLOOKUP, connect the files to tableArth.ai and ask your question in plain English. It reads every tab, sheet, and table you connect, works out how they relate to each other, and joins them on shared keys automatically — the same way tables join inside a database — so nothing needs a lookup formula or a flattened, combined copy first.
Why joining data across sheets is so painful
Most spreadsheet work starts in one tab and ends up needing another. A deal only makes sense next to the rep who owns it; an order only makes sense next to the customer who placed it. The moment a question spans two tabs, two files, or a sheet and a database, the standard toolkit is VLOOKUP, INDEX-MATCH, or — if those feel too fragile — copying everything into one master sheet by hand.
VLOOKUP was never built for this. It looks up a single value, in one direction, and returns only the first match it finds. Insert a column and the reference breaks. Add a second matching row and the formula silently ignores it. Chain a few lookups together to combine several tabs, and the sheet becomes something only its author can safely edit.
The fallback is worse: export everything into one flat sheet and rebuild it by hand whenever the source changes. That works once. It does not scale to a workbook with a dozen tabs or data that updates daily — and every re-copy is a fresh chance to paste the wrong range or miss a row.
The shift from formulas to asking a question in plain English looks like this:
| Task | VLOOKUP / manual | Ask tableArth.ai |
|---|---|---|
| Combining two tabs | Formulas you build and maintain | Ask a question that spans both |
| Combining two files | Export, paste, align columns by hand | Connect both to one Workbook |
| A sheet plus a database | Export the database table first | Query the database directly, in place |
| When source data changes | Re-copy, re-paste, re-check formulas | Workbook stays synced automatically |
| Multiple matching rows | First match only, one direction | Matches many rows, either direction |
| An unclear relationship | You guess, or dig through columns | It asks a clarifying question |
What a join actually is
A join is how two related tables get combined based on a column they share — a customer ID that appears in both a Customers tab and an Orders tab, or an email address that shows up in both a leads list and a billing table. The shared column is the key. Once two tables share a key, you can ask a question that draws from both without ever merging them into a single physical sheet.
That is a different thing from a lookup formula. VLOOKUP finds one
value and pulls a single related field back — a formula bolted onto a
static range. A real join is relational: it can match many rows on
either side, work in either direction, and hold at real scale. It is
exactly what happens when two tables in a database are joined on a
foreign key — orders.customer_id = customers.id — except
that logic normally never reaches a spreadsheet.
tableArth.ai brings that database-style join to sheets, Excel files, and databases together. Ask a question that spans two tabs, two files, or a sheet and a database, and it finds the shared key and joins on it — instead of asking you to build the relationship yourself first.
Joining tabs within one file
The simplest case is a workbook with more than one tab — a Google Sheet with Leads, Deals, and Reps, say, or an Excel file with Inventory and Sales. Connect the file, and tableArth.ai reads each tab as its own table, keeping its own columns and types rather than flattening the whole file into one undifferentiated block.
Ask a question that only one tab can answer — total leads this month — and it reads the Leads tab. Ask a question that needs two — "which reps closed the most deals from leads that came through the paid campaign?" — and it recognizes that a lead ID in Deals matches a lead ID in Leads, and a rep ID in Deals matches a rep ID in Reps, joins the tabs on those shared columns, and answers directly. No helper column, no VLOOKUP chain, no tab built purely to hold a lookup formula.
This works whether the workbook has two tabs or a dozen — each additional tab is just another table to draw into a join, not another formula to maintain.
Joining across files and sources
The same idea extends past a single file. A Workbook can hold more than one source at a time — a "Sales pipeline" Google Sheet and a separate "Support tickets" Excel file, for instance — connected together rather than merged into one. Ask a question that needs both, and tableArth.ai joins them on whatever column identifies the same thing in each: an account name, a customer ID, an email.
It also reaches into a database. Add a Google Sheet of sales targets and a MySQL, PostgreSQL, MongoDB, or Druid table of booked orders to the same Workbook, and a single question — "how are we tracking against target by region this quarter?" — pulls from the sheet and queries the database, joining the two the same way it would join two tabs. The database can run inside your own private environment; nothing has to be exported or copied out to be part of the join.
Because a Workbook stays synced with every source it holds, this doesn't need to be rebuilt each time the underlying data changes. Update the sheet, add rows to the database, and the next question still reaches across all of it — live, without a re-upload. More on this in live data vs. one-time uploads.
How tableArth.ai infers the relationship
Joining automatically only works if the system understands what it's looking at first. tableArth.ai reads the structure of every tab and table you connect — column names, types, and sample values — before a single question is asked.
From there, it looks for identifiers that appear in more than one
place: an ID, an email address, an order number, a SKU. A column
named customer_id in one table and id in
another, both holding the same kind of value, reads as a relationship
the same way a foreign key does in a database. When a question needs
both tables, that shared identifier is what it joins on.
From there it's the same engine behind every natural-language question in tableArth.ai: it writes the query — including the join — runs it against your connected sources, and streams back an answer with a chart, typically in under five seconds.
Not every relationship is obvious, and tableArth.ai doesn't pretend it is. When more than one column could plausibly be the join key, it asks a clarifying question instead of guessing — the same way it suggests relevant questions for a newly connected dataset. That gets you to a correct join faster than debugging a wrong VLOOKUP after the fact.
Do I still need VLOOKUP or INDEX-MATCH?
No. Once your sheets, files, or databases are connected, you ask your question in plain English and tableArth.ai performs the join for you — reading the relevant tabs or tables, matching them on a shared key, and returning the combined answer. You don't write a lookup formula or build a helper column.
Can it join data across different files, not just tabs in one file?
Yes. tableArth.ai understands files with multiple tables and tabs across Google Sheets, Excel, and databases, and a Workbook can hold more than one file or source at once. Ask a question that spans two files and it joins related data the same way it would join two tabs in the same workbook.
Can it join a spreadsheet with a database?
Yes. A Workbook can combine Google Sheets or Excel files with databases such as MySQL, PostgreSQL, MongoDB, and Druid. Ask a question that needs both, and tableArth.ai pulls from the sheet and queries the database — including databases running in your own private environment — to return one unified answer.
How does it know which columns to join on?
It reads the structure of every connected tab or table — column names, types, and sample values — and looks for shared identifiers, such as an ID, email, or order number, that link two tables the same way a foreign key does in a database. If more than one relationship is plausible, tableArth.ai asks a clarifying question instead of guessing.