Most marketplace brands in India run on a morning ritual. Someone opens each seller portal, downloads yesterday's sales, the warehouse and marketplace stock, and what's still in transit, then types it into a set of spreadsheets the whole business plans from. For the eyewear and accessories brand whose orders we already automate, that meant eight channel folders and around twenty workbooks — Myntra, Nykaa, Nykaa Fashion, Ajio, Blinkit, Flipkart and two Amazon accounts, plus warehouse stock from Unicommerce. It took hours, every day, and the sheets were too important to get wrong.
Rule 1: fill their sheets, don't build new ones
The tempting move is a shiny new dashboard. We didn't. The team's workbooks have years of formulas, pivots and habits built on top of them, so the automation writes into the exact files, tabs, columns and date rows they already use. Nobody had to learn anything new, and every downstream formula kept working on day one.
Rule 2: only ever fill blank cells
The job never overwrites a number a person typed. For every cell it would write, it checks what's there: blank gets filled; an identical value is marked SAME; a different value is marked CONFLICT and left alone. The conflicts turned out to be the most useful output. They showed exactly where the team's numbers and the portal's numbers disagreed — usually returns and cancellations that a portal removes from its figures a few days later.
Rule 3: run on copies until the numbers match
For the first few days every run wrote into test copies in a separate folder, never the real files. We compared the copies cell by cell with what the team entered by hand for the same days. When a day's verification came back 100 out of 100, and Flipkart's figures matched the team's exactly, the client switched it to real mode. Even now, every real run saves a backup of each original file before writing a single cell.
Rule 4: every source lags and fails differently
Flipkart's sales report runs about two days behind, and asking it for a date range that someone already requested — the team or us — returns a server error rather than the data. The fix was to look up the report that already exists for that range and download it. Myntra's report jobs can take over an hour on a busy morning, so the job waits instead of giving up. Amazon's vendor analytics report revenue without GST, so we applied the team's own rule (the GST rate by product category) and checked it against several past days they had filled by hand before trusting it.
Rule 5: read spreadsheets by their headers, not by landmarks
Our one real bug came from the in-transit sheet. The code found the status column by looking below a row labelled 'Total'. Only one of the four tabs actually had that label; the others had an unlabelled total row, so the status was never read and shipments already marked Completed were counted as still in transit. The client spotted it the next morning. The fix was to locate columns by their header names wherever they are, and correct the affected values. If your automation reads a human-maintained sheet, assume the layout differs slightly on every tab.
Rule 6: send one plain summary
The job runs on a server at a time the team chose, never on anyone's laptop, and sends one email afterwards: what was written per channel, what was skipped as already filled, every conflict, and anything that failed with the reason. On a normal day that's tens of thousands of cells the team no longer types.
What it changed
The morning ritual became reading one email. The team still owns the sheets and can type into any blank cell before the job runs — it will simply respect what they entered. And the conflict list gives them something they never had before: a daily check that their numbers and the marketplaces' numbers agree.
Thinking of automating your own marketplace reports
Start with the blank-only rule and a few days on copies; it makes the whole project safe to try. At HowAutomate we build this alongside order processing and price tracking for Indian marketplace sellers — happy to look at your sheets and tell you what can be filled automatically.
Frequently asked questions
Can marketplace sales and stock data be filled into Google Sheets automatically?
Yes. A scheduled job can download sales, stock and in-transit data from Myntra, Nykaa, Ajio, Blinkit, Flipkart, Amazon and Unicommerce and write it into the team's existing workbooks, in their exact tabs and columns.
Will automation overwrite numbers my team typed?
It shouldn't. Our rule is blank-only: the job fills empty cells, marks matching values as SAME and reports different values as CONFLICT without changing them. Every real run also backs up the original file first.
Why does Flipkart's sales report return an error for some dates?
Flipkart refuses a new request for a date range that has already been requested. The fix is to find the existing report for that range and download it. Flipkart's sales data also runs about two days behind.
How long does it take to switch from manual to automated reports?
Typically a few days of running on test copies, compared cell by cell with the team's own entries, before switching to the real files.
Amit has 6+ years of experience building data pipelines, AI agents, and automation systems for businesses across India and globally. He founded HowAutomate to make enterprise-grade automation accessible to growing businesses.