What 'Assisted' Import Means, and Why It Matters That the Software Asks
Fully automatic import sounds better than assisted import until it silently converts a unit, groups the wrong rows or fills a blank. What 'assisted' means in practice and the questions you want asked.

"Fully automatic" is the phrase every vendor wants on the slide. For a supplier file full of ambiguities, it's the phrase to read most carefully. Automation isn't the problem; well-built automation applies approved rules, rejects ambiguous rows or asks for review. The problem is unchecked assumptions: every ambiguity a tool resolves silently is a decision made without you. Here is what assisted import means, question by question, and why the questions are the point.
The decisions hiding in a supplier file
Open a typical furniture price list and count the things a tool would have to decide.
Which row is the header, when rows 1 to 3 are a title block. Whether "W", "d" and "ht" are dimensions, and in what unit, when the only unit declaration is a note on row 2. Whether "920" and "920mm" in the same column are the same unit. Whether "Lilac", "LIL" and "lilac" are the same color. Which of "Trade" and "RRP (inc VAT)" is the selling price. Whether seven rows with a shared code stem are one product with seven variants, or seven products. Whether a two-seater and a three-seater are two products or one on a size axis. What to do with a blank depth. Whether "SEATING" on row 5 is a product.
A tool that makes unchecked assumptions may answer these, right or wrong, and move on. Controlled automation answers the ones it can from rules you've approved and asks you the rest.
What Furniture Connect asks, and when
At preview. The import shows the sheets, the headers it found and sample rows before anything happens, so a header on row 4 under a title block is visible and correctable [1].
At mapping. The assistant suggests where each column should go and you review the suggestions. For a layout it recognizes, such as a standard Shopify or WooCommerce export or a supplier template with title rows above the header, manual mapping may not be needed; review the detected mapping and settings anyway [1].
At units and grouping. Where something is unclear, such as whether dimensions are in centimeters or inches or how colorways are grouped, it asks rather than deciding silently [1]. File-wide questions can often be settled once; a column with mixed units or exceptions can still need a row-level answer.
At variants. For a range with variants, the assistant asks how the rows should be grouped when the structure is unclear, then builds the variants from the confirmed answer [1]. "Are these two rows the same material?" is the kind of question to settle before one interpretation is applied across 40 products.
At the gaps. Autofill columns write from the fields you let them read, show a preview, and run on ten sample cells before continuing [2]. Point them at copy, not at facts the supplier hasn't supplied, and the sample run is where you confirm they leave a missing dimension or material alone.
At the end. The assistant reports how many products were added and updated. New products arrive as drafts; nothing goes live because a file was imported [1].
Why asking is faster than it sounds
Because file-wide decisions can often be settled together, and approved mappings can be reused. When you save a supplier mapping, you can reuse it for a matching layout; the assistant offers this after the first import [1]. Saved mappings don't necessarily preserve every conversational decision, so expect the occasional repeat question. On re-import, use stable SKUs, confirm the assistant's proposed matches and include only the columns you intend to change. Match on stable SKUs rather than product names [3].
Compare that with an automatic tool that gets the unit wrong: 40 products listed at ten times their width, discovered by a customer, corrected one by one. "Automatic" moved the work from before import to after it, and multiplied it.
Where assisted still means your judgment
PDFs, product photos and website addresses go through the same assistant, but there the assistant is reading and extracting rather than mapping columns, which means more questions and a closer check of the result [1].
And the hardest questions remain yours. Is a different seat height a different product? Does this retailer want the brand in the title? The software can surface the question. Only your merchandiser can answer it.
The wider journey is in getting a new furniture range online.
What to look for in any import tool
Ask three things in the demo. Does it show me the file before it does anything? When a unit or a grouping is ambiguous, does it apply a rule I approved, ask me, or decide silently? What does it do with a blank? If the answers are "no", "decides silently" and "fills it in", you've found the tool that makes the work faster to get wrong.
For what the questions look like end to end on a real file, see what happens when you import a supplier spreadsheet. For the broader principle of what a catalog should and shouldn't decide, see what a catalog should and shouldn't decide for you.




