A blank cell beats a wrong one: an FF&E schedule demo
A demo on fictional vendor pages: paste a product link into an FF&E schedule and the row fills, with blanks, check notes and statuses kept visible.
If a schedule row could fill itself from a product link, would you trust what lands in it? That’s the question behind this demo, which I built in October 2026, as a solution concept for interior design firms: how does the designer stay in control of the schedule when automation is doing the typing?
Quick disclaimer up front: the vendor pages are fictional, built for testing. This is a demo, not client work and not a case study.
What the demo does
A designer pastes a product URL into an FF&E schedule in Google Sheets, and the row fills in the image, name, brand, SKU, finish, colour, dimensions, price and description. It runs when the link is pasted, one row at a time.
It fills only empty cells, so anything the designer has already typed in never gets overwritten. The product image is copied to Google Drive and shown in the cell, so the schedule doesn’t depend on the vendor’s site to display it.
Four sources, in a fixed order
The demo looks in four places, in order. Each step fills only what the earlier ones left blank:
- The store’s structured product data, where the site provides it, including the selected variant.
- Product markup embedded in the page.
- The sharing tags a page publishes for link previews.
- An AI read of the page text.
The AI step gets the strictest rule. Any value it returns that doesn’t appear in the page text gets dropped. Cells the AI fills carry a “check” note, so a person knows to confirm them.
On sites without the store data, the sharing tags needed two extra checks (yes, the IT geek in me enjoyed this part). A store name tacked onto the end of the shared title gets removed when it matches the site name or brand, and a shared image that looks like a logo or a default share image gets ignored.
My design rule: a blank cell beats a wrong one. A blank is easy to spot. A plausible wrong value isn’t.
Every row ends with a status
When a row finishes, it shows one of five statuses: FILLED; PARTIAL, with a note listing what’s missing; NO DATA, when the page is built by JavaScript; BLOCKED, when the page needs a login; or ERROR, when the page isn’t found. A row that came back incomplete says so.
What the testing shows, and what it doesn’t
As of October 2026, all testing used fictional vendor pages built for the purpose. A local test suite, with a stand-in AI that deliberately invents values, produced no wrong fills. Disabling five of the safeguards one at a time (the page-text check on AI values, variant handling, taking the brand from the site name, title clean-up and the logo filter) made those tests fail. With a real AI model on 12 test URLs, there were no wrong fills, and no value that the test pages contained was missed. Pasting a URL into the demo sheet filled the row.
To be honest about the limits: these pages were built for the test, so the results show the logic works, not how often real vendor sites expose clean data. No real vendor sites or client data were used, and I haven’t measured any time saving.
How it differs from a batch workflow
My home-furnishings work includes a content workflow for a brand that also pulls a product title and description from product links, but in a different shape. It runs in batches from a menu. A person confirms each product link before a row is processed, and the reviewed pins are exported as a CSV for scheduling. Scheduling and publishing stay human steps. This demo fills one row at a time, on paste, with nothing downstream. Both follow the same rule: never overwrite existing values.
Why the two take different shapes is my design view, not a tested finding. A batch with a confirmation step suits work that feeds another system, where a wrong link would travel further before anyone sees it. Filling on paste suits a schedule built one item at a time, where review happens right in the sheet. I’m not claiming one is better than the other.
Where I’d use it in a design practice
These are proposed uses, not tested with any firm:
- A designer pastes vendor links to get first-pass schedule rows. Blanks and PARTIAL notes show what still needs a manual spec check.
- Because images are copied, the schedule doesn’t depend on the vendor keeping an image online.
- The “check” notes and row statuses together become a review queue before the schedule goes to anyone else.
How this would fit your firm’s schedule template, vendor mix or approval process is untested. If you’re wrestling with FF&E schedules, I’d love to hear what slows yours down.