The first campaign is where most PPC habits, good and bad, get set. Build it well and the account grows from a clean base. Build it in a hurry and the mess compounds for a year. This walkthrough is for a seller setting up Sponsored Products for the first time, on one product, with the goal of learning what works before spending to scale it.
Before the console
Check the listing. A campaign sends shoppers to the product page, and the page decides whether they buy. Main image clear at thumbnail size, title leading with what the product is, bullets that answer the obvious questions, price in line with the category, at least a handful of reviews. If the listing is not ready, the campaign will tell you so expensively.
Know the break-even ACoS. Selling price, minus referral fee, minus fulfillment fee, minus product cost, divided by selling price. That percentage is the ACoS at which ads break even. A first campaign can run above it for a while, because it is buying data and sales velocity, but you need to know where the line is.
Write down ten keywords. The terms a shopper would type to find this product. Brand plus product, product type, the main use case, the main feature. Do not research exhaustively; the automatic campaign is about to do that for you.
Campaign one: automatic
Sponsored Products, automatic targeting. One product. Name it with the product and the word "auto" so you can find it later.
Daily budget: enough to buy 15 to 30 clicks a day at Amazon's suggested bid for the product. If the suggestion is $0.80, a $15 to $25 daily budget is a reasonable start. Too small and the campaign runs out before the day's shoppers arrive and you learn nothing.
Bidding strategy: dynamic bids, down only. This lets Amazon lower your bid when a conversion looks unlikely and never raises it. The conservative default, and right for a campaign whose job is discovery.
Bids on the four match groups: start at the suggested bid for close match and substitutes, a bit lower for loose match and complements. These can be refined once there is data; they do not need to be right on day one.
No negatives yet. You do not know what to exclude. That comes in two weeks.
Campaign two: manual, exact match
Sponsored Products, manual targeting, keyword targeting. Same product. Name it with the product and "exact."
Keywords: the ten you wrote down, exact match only. Exact match on terms you are confident about gives you control over the bids that matter most, and keeps the automatic campaign from being the only thing serving on your best terms.
Bids: the suggested bid for each, or slightly above for the two or three terms that most precisely describe the product. These are the terms most likely to convert, and it is worth being visible on them from the start.
Budget: similar to the automatic campaign. Bidding strategy: dynamic down only, same reasoning.
Two campaigns, one product, both running. That is the whole first setup. Resist adding broad match, product targeting, or a second product. The point is to learn, and two campaigns are enough to learn from. The campaign structure guide covers where the account goes from here.
The first two weeks: do nothing
The hardest step. Amazon keeps attributing orders to clicks for days after the click, so the first few days' ACoS is always worse than it will settle to. Keywords need enough clicks to say anything, and a keyword with eight clicks and no orders is not a failed keyword; it is an unmeasured one.
Check daily for two things only: that the campaigns are spending (if not, bids are too low or the listing is ineligible), and that they are not running out of budget by early afternoon (if so, raise the budget modestly). Everything else waits.
Week three: the first review
Pull the search term report for the automatic campaign. It lists every query that triggered your ad, with clicks, spend and orders for each. Three groups emerge.
Terms that converted. Add them to the exact match campaign as new keywords, with bids based on how well they converted. This is keyword harvesting, and it is the main thing the automatic campaign is for.
Terms that spent and did not convert, and clearly do not fit the product. Add them as negative keywords in both campaigns so they stop spending.
Terms that spent, did not convert, and might fit. Leave them. They may need more clicks to judge.
Then look at the exact match campaign. Keywords converting at an ACoS under your break-even can take a bid increase. Keywords well above it with enough clicks to trust can take a decrease. Keywords with few clicks wait. The ACoS reduction guide covers the next level of this.
What to look at once it is running
Spend by hour. Even a first campaign runs around the clock, and in many categories the overnight hours produce clicks at a fraction of the daytime conversion rate. Once there is a few weeks of data, dayparting is often the first efficiency change worth making, because it changes nothing about keywords or bids. It just stops paying for the hours that do not sell.
Budget behavior. A campaign that runs out by noon on strong days is leaving the afternoon and evening to competitors. Budget rules that raise the cap on specific days, or pacing that holds budget for the better hours, are the next step after dayparting.
Mistakes to skip
Changing bids every day. Reacting to noise, and resetting the learning each time.
Adding every keyword from a research tool. Fifty keywords in a first campaign means fifty bids to manage and a search term report too large to read. Start small; let the auto campaign find the rest.
Judging the campaign on week one. The numbers are not settled and the sample is tiny.
Skipping the negatives. By week three there will be irrelevant terms spending. Every week they go unaddressed is budget gone.
Frequently asked questions
How much should I budget for my first Amazon PPC campaign?
Enough to get clicks every day for two weeks without the campaign running out by noon. For most products that is a daily budget that buys 15 to 30 clicks a day at the suggested bid. The point of the first campaign is data, and a budget too small to produce it teaches nothing.
Should my first campaign be automatic or manual?
Both, as two campaigns. The automatic campaign finds the search terms shoppers actually use for your product. The manual campaign, seeded with a short list of obvious terms, lets you control bids on the ones you are confident about. Running only one of them means either no discovery or no control.
How long before I should change anything?
Two weeks, unless something is clearly broken. Amazon's attribution keeps counting orders for days after a click, and a campaign needs enough clicks per keyword to say anything. Changing bids after three days means reacting to noise.
Once the first campaign has a few weeks of data, Off Hours dayparting and budget rules are the usual next step: the schedule and the cap, running every day without a login. Start a free 14-day trial.