Why SellerDragon exists

It started with a spreadsheet subscription that made me angry.

I run a seven-figure Amazon account. My P&L lives in Google Sheets, like every serious operator I know. So I paid for a connector that pulls Amazon data into sheets. It was slow. Then the vendor changed the sync cadence on the base plan – data I'd been getting fresh now arrived late, unless I paid more.

Delaying data you already have, to sell it back faster – that offended me as an engineer and as an operator. I've been building software since before the web existed. So I built my own pipeline: faster, into my own sheets, with a rule I've kept ever since.

Data is never delayed, on any plan. As we get it from Amazon, you see it.

Then the dashboard happened.

The pipeline solved the numbers. But I was still paying for a research suite whose market figures I'd stopped trusting, and doing the daily PPC work across eleven browser tabs. The data was already flowing through my own warehouse – so I built the cockpit on top of it. Daily overview, PPC workbench, dayparting, listing work. A dashboard where seeing the problem and fixing it happen on the same screen.

Then AI ate my Tuesday. In a good way.

I connected my account to Claude through MCP, mostly as an experiment. Managing campaigns by talking to an AI that sees your real numbers – ranks, spend, stranded inventory – and stages the changes for your approval is the biggest shift in how I work since the spreadsheet itself. The experiment became the third door, with a hand brake welded on: your AI proposes, you approve, then it ships.

That's the story.

No pivot deck, no vision statement. A tool that got built because the existing ones were slow, untrustworthy, or didn't exist yet – by someone who runs a real account and flips between all three doors before lunch.