The product went live, so the ads went on. A few days later the spend was real and the search report was a junk drawer. I could not say which words belonged to the product, because I had never written the question. I had bought traffic. I had not bought an observation.
Early ads are a way to learn, with a stop
A new listing has almost no history. Ads can put it in front of people who already search for the job it does. That is useful. It is not a permanent ad system, and it is not proof the product “has demand” after forty-eight hours.
I pick a dollar amount I am willing to spend to learn, and I write it down. When it is gone, I stop and read. Refilling it because the campaign “needs more time” is how a test becomes a habit.
Write the question first. Example: which of these searches are actually people looking for this lid?
Start with words you already saw on the shelf
Use the searches from the listing review, the ones that brought up true competitors. Keep the set small enough that you will finish the report. A huge broad campaign on day one produces a document you will not read.
Example only. Your numbers will be different.
Skip words that attract a different product. Those clicks look like a conversion problem. They are a targeting problem. I would rather have a short list of honest searches than a busy account full of strangers.
Change one thing, then wait
Look daily only for waste you can name, such as a search for a product you do not sell. Save the real read for the week. If you change the bid, the title, and the price on the same morning, the next report is unreadable. I have done that. I learned nothing and felt busy.
Each day
- Do this
- Cut a search that is clearly the wrong product
- Leave this alone
- A new title and new bids together
Once a week
- Do this
- Answer the question you wrote
- Leave this alone
- A stack of new campaigns
People click and do not buy
- Do this
- Check the photo, the price, and the stock
- Leave this alone
- A higher bid
| When | Do this | Leave this alone |
|---|---|---|
| Each day | Cut a search that is clearly the wrong product | A new title and new bids together |
| Once a week | Answer the question you wrote | A stack of new campaigns |
| People click and do not buy | Check the photo, the price, and the stock | A higher bid |
Ads send people to the page you actually have. If that page fails the path you planned, more spend will not repair it. Keep a search that earns orders at a cost your sheet can carry. Cut a search that brings the wrong shopper, and write why, so you do not invite it back next month.
This is the same path we run in the Amazon Selling System. You approve the product, the factory, and the launch. We do the work.