Amazon Bid Optimization: Manual vs AI Tools in 2026

A keyword with 20 clicks at $1.50 and one sale is running a 75% ACOS. Most sellers either pause it or hand it to a tool and hope. Drop that same bid to $0.50 and the same 20 clicks cost $10 for the same sale, which is a 25% ACOS. That one change is what Amazon bid optimization comes down to: setting each bid so the click is worth what you pay for it. Software can do this faster than you. It can’t do it better than you on thin data, and it can’t fix a listing nobody wants to click. This guide covers the manual method, what AI bid tools really do, and when each one makes sense.

What Amazon bid optimization actually means

Your bid is the most you’re willing to pay for one click on one keyword or target. Amazon runs a second-price style auction, so you usually pay a bit less than your max, but the bid still decides two things: whether you show up at all, and where you show up when you do.

Bid optimization is the habit of moving those numbers up or down based on what each keyword earns. It isn’t a one-time setting. Conversion rates shift, competitors come and go, your price changes, and a bid that was right in March can be losing money by June.

Our course puts it plainly: bid management is the single biggest lever on campaign performance. Even with an average conversion rate, the right bid can turn a losing campaign into a profitable one. We’ve seen that hold across the stores we manage, which do $200K a month for paying clients today.

There are four levers you can pull, and every tool, manual or AI, is just pulling some mix of them:

1. The keyword or target bid. The base number you set for each keyword, product target, or auto targeting group.
2. Placement adjustments. A percentage on top of your base bid for Top of Search or Product Pages. Amazon allows up to 900%.
3. The bidding strategy. Fixed bids, dynamic bids down only, or dynamic bids up and down. This controls how much Amazon is allowed to move your bid in real time.
4. The budget. Not a bid, but it caps how many of those bids get to fire each day.

The “manual vs AI” question is really about who decides on lever 1 and lever 2, how often, and using how much data.

Why bids matter more than most sellers think

Here’s the example from our bid management lesson that changed how a lot of our students think about placement. A product bids high to win Top of Search and pays $5.00 a click. Even with a great 35% conversion rate, the click cost eats the margin and ACOS lands around 95%.

Same product, but now targeting competitor product detail pages. Less competition there, so the click costs about $1.50. Same number of sales, same conversion rate, and ACOS drops from 95% to roughly 28%. Nothing changed about the product or the listing. Only where the bid was pointed and how much it was.

That’s why bids matter. Small numbers per click become big numbers per month. If you run 3,000 clicks a month, 30 cents of overbidding per click is $900 gone, and that money doesn’t show up anywhere obvious. It just quietly makes your ACOS look worse than your product deserves.

A few things bids decide that people forget:

• Whether your ad shows at all. Bid too low and you sit on page 3 or 4, where nobody clicks. You’ll see impressions with no clicks.
• Which placement you win. Top of Search costs the most and gets the most traffic. Rest of Search and Product Pages are cheaper with less volume.
• How fast you burn the budget. High bids on broad keywords can drain a $10 daily budget before lunch in the US, and then your ad is off for the rest of the day.
• What data you get. If you only ever bid on expensive placements, you never find out whether the cheaper ones convert just as well.

One honest caveat, because it saves people a lot of time. Bids are downstream of your listing. On a sports and fitness launch we ran, the main image showed a four-item bundle at a 20% price premium. Month-one ACOS was 25% with a 13% TACOS, because shoppers clicked more and bought more. Sellers trying to win by matching or undercutting price tend to run 40% to 60% ACOS in month one, because every listing converts about the same. No bid strategy fixes that. It’s a product and image problem first.

How manual bid optimization works, step by step

Manual bidding sounds slow, but the method is simple. You’re doing the same math the tools do, just less often and with your own judgment on top. Here’s the process we teach.

Step 1: Work out your target bid

Every bid should come from three numbers you already have:

Target bid = conversion rate × average selling price × target ACOS

Say a keyword converts at 10%, your product sells for $30, and your target ACOS is 30%. That’s 0.10 × $30 × 0.30 = $0.90. At $0.90 a click, 10 clicks cost $9 and earn one $30 sale, which is a 30% ACOS. Bid more than $0.90 and that keyword runs above target. Bid less and it runs below.

Your target ACOS shouldn’t be a guess. Your break-even ACOS is your margin before ad spend, and our Amazon FBA profit margins guide shows how to work that out line by line. Set the target below break-even for profit, or at break-even for a keyword you’re pushing for rank.

Step 2: Start low and let data come in

At launch, start at the bottom of Amazon’s suggested bid range. If it shows $0.80 to $1.34, bid $0.80. Use dynamic bids down only, so Amazon can lower your bid when a sale looks unlikely but never charge above your max.

Then leave it alone for seven days. Amazon’s sales attribution lags, and a keyword that looks dead on day 2 often shows sales by day 6. Bid changes made on two days of data are mostly reactions to noise.

The one exception: if a keyword gets 1,000 or more impressions and zero clicks in the first 24 to 48 hours, your bid is too low and you’re buried on later pages. Raise it 10% to 20%, wait 24 hours, and repeat until clicks start. Don’t jump to the top of the suggested range in one go.

Step 3: Sort every keyword into a bucket

Once a week, pull the search term report and put each keyword into one of these:

1. Winners. Selling, ACOS below target. Raise the bid 10% to 20% to win more impressions, and move the search term into its own Exact match campaign if it isn’t already.
2. Selling but expensive. Getting sales, ACOS above target. Don’t pause it. Lower the bid toward the target bid from Step 1.
3. Spending, no sales yet. Watch from 12 clicks. At 20 clicks with zero sales, it’s a proven bleeder: pause it or add it as a negative.
4. Irrelevant. A “cordless” search on a wired product. Add it as a negative right away, at any click count.
5. Not enough data. Under about 12 clicks. Leave it.

Step 4: Adjust placements separately

Open the campaign, go to the Bid adjustments tab, and check the placement report. If Top of Search converts well at a healthy ACOS, add a percentage to it. A $1.00 base bid with a 50% Top of Search adjustment means you’ll pay up to $1.50 for that spot. If Product Pages convert better for you, put the adjustment there instead.

Keep it gradual. Change placements by 10 to 25 points at a time and give each change a week.

Step 5: Repeat weekly, harvest monthly

Weekly: the bucket sort and bid changes. Every 30 to 60 days: mine your auto and Broad campaigns for new converting search terms and competitor ASINs, and give them their own campaigns where you control the bid.

That’s the whole manual system. On a small account with 3 to 5 campaigns, it takes about an hour a week.

What AI bid tools actually do (and what they don’t)

“AI” covers a lot in this space, and most of it isn’t magic. Under the hood, nearly every bid tool runs some version of the same loop you just read. It looks at clicks, spend, and sales for each target, estimates the conversion rate, calculates a bid to hit a goal, and pushes the new bid through the Amazon Ads API. The difference is speed and scale: it can do this daily or several times a day, across thousands of keywords, without getting tired.

The tools sit roughly in three groups.

Amazon’s own built-in options (free). Dynamic bidding strategies let Amazon raise or lower your bid per auction based on how likely a sale is. With up and down, Amazon can raise bids by up to 100% for Top of Search and up to 50% for other placements. Amazon also offers rule-based bidding on some Sponsored Products campaigns, where you set a target ROAS and Amazon adjusts bids toward it. Eligibility depends on the campaign having enough conversion history, so check what your console shows. Budget rules can raise budgets on high-traffic days like Prime Day.

Rule-based third-party tools. You write the rules, the tool runs them. “If a keyword has 20 clicks and zero orders, pause it.” “If ACOS is under 20% over 14 days, raise the bid 10%.” This is your manual process, automated. It’s predictable because you know exactly why every change happened.

Algorithmic or “AI” platforms. Tools like Helium 10’s Adtomic, Teikametrics, and Perpetua set bids with their own models toward a goal you choose, such as a target ACOS or a daily spend. Some add dayparting (different bids by hour), forecasting, and cross-campaign budget moves. Pricing varies a lot between them and changes often. Some charge a flat monthly fee, some a percentage of ad spend, so check the current pricing page before you commit.

What these tools do well:

• Volume. Ten SKUs with 400 keywords each is 4,000 bids. Nobody updates that by hand every week without mistakes.
• Frequency. Daily or hourly changes catch shifts a weekly human review misses, like a competitor going out of stock.
• Discipline. A tool doesn’t panic-pause a keyword after a bad Tuesday.
• Dayparting. If your sales cluster between 7pm and 11pm Eastern, a tool can lower bids overnight. Doing that by hand is miserable.

What they don’t do, and this is where most of the disappointment comes from:

• They can’t create data. A model needs clicks and orders to estimate conversion rate. A new keyword with 6 clicks and no sales gives the model almost nothing, so it either guesses or swings bids around.
• They don’t know your strategy. If you’re deliberately overspending on one keyword to rank, the tool sees a high ACOS and cuts it, unless you tell it otherwise.
• They can’t fix the listing. If CTR is weak because the main image is weak, every bid tool will just keep lowering bids until you’re invisible.
• They don’t see your margin unless you give it. Tools that only know ACOS will optimize toward ACOS, even when your real number is TACOS.

The real difference: data, not intelligence

The honest way to frame manual vs AI tools is this. Both use the same formula. A tool applies it faster and more often. You apply it with context the tool doesn’t have. Which one wins depends on how much data you have and how many decisions you need to make each week.

A rough way to think about it:

The “under $2,000” line isn’t a hard rule. It’s where the math usually tips. If a tool costs a flat fee in the low hundreds a month and your whole ad spend is $1,500, the tool has to save you a meaningful slice of that spend just to break even on its own fee. On a big account the same fee is a rounding error.

There’s a bigger point under the table. On the home and kitchen launch we’ve talked about in other posts, month-one ACOS was 15%, and that came from tight niche selection before a single bid was set. Many beginner launches start between 50% and 80% ACOS and grind it down for months. The gap between those two results wasn’t software. It was the product, the listing, and patient manual bidding while data came in.

Mistakes that wreck bid optimization (manual or AI)

Most of the money we see lost on bids comes from the same handful of habits. None of them are fixed by switching tools.

Changing bids on two days of data. Attribution on Sponsored Products takes days to settle. If you cut a bid on Wednesday because Monday and Tuesday showed no sales, you’re often reacting to orders that just hadn’t been credited yet. Give every change at least a week before judging it.

Pausing keywords that sell. This is the most common one. A keyword at 60% ACOS that’s getting orders is telling you the bid is too high, not that the keyword is bad. Lower the bid toward your target bid. Remember the 20 clicks example: $1.50 to $0.50 took the same sale from 75% ACOS to 25%.

Turning on dynamic bids up and down at launch. Amazon can then double your cost per click on Top of Search when it thinks a sale is likely. With no conversion history, that’s an expensive way to learn. Start with down only. Move a campaign to up and down later, once it’s proven and you want more Top of Search share.

Letting a tool and a human fight. If you set a tool to manage bids and then also edit them by hand, the tool sees its changes undone and responds. You end up with bids bouncing every day and no clean data on what worked. Pick who owns each campaign.

Optimizing to ACOS when the business runs on TACOS. A ranking campaign at 40% ACOS can be the one feeding your organic sales. A tool told to hit 25% ACOS will cut it. Our ACOS vs TACOS guide explains how to read the two together before you set any goal in a tool.

Ignoring returns. On a vegetable chopper launch, ACOS sat around 20% and TACOS between 10% and 15%. The dashboard looked like a winner. But that whole niche had top sellers rated under 4.5 stars, which is a sign of a high-return category, and returns ate the margin. Bid tools never see returns. If your category returns a lot, your real break-even ACOS is lower, so your target bids should be too.

Expecting bids to fix a generic product. We managed a wood-therapy massage tool account where every new tactic, including PPC optimization, ASIN targeting, and listing refreshes, gave a short lift and then sales slid back. The product looked like every other one on the page. Better bids got a few good weeks, not a business.

Manual vs AI tools side by side

Here’s the direct comparison, so you can decide what fits where your account is right now. Don’t worry if you land on “manual” for now. Most sellers in their first year should.

A setup that works for a lot of growing accounts is a mix. Use dynamic bids down only on everything. Keep your top 10 to 20 Exact match keywords under manual control, because those carry your rank and you want to make those calls yourself. Let rules or a tool handle the long tail of Broad, Phrase, and auto targets, where there are too many small decisions for one person.

Whatever you choose, two things stay the same. You still need to know your margin, because every tool needs a target. And you still need to read the search term report yourself every few weeks, because a tool will happily keep bidding on a search term that’s relevant on paper and useless in real life.

Learning bid optimization yourself

The good news is that none of this needs to be outsourced or bought. Manual bid optimization is a skill, and once you’ve done the weekly bucket sort a few times, it gets fast. You’ll also be much better at judging any tool later, because you’ll know what it should be doing.

If you want the structured version, our $27 Amazon course has full PPC lessons, including bid management, placement adjustments, auto and manual campaign setup, search term reports, and negative keywords. For how bids fit into the bigger picture at each budget level, the pillar guide on Amazon PPC strategy by ad spend walks through spend from the first few hundred dollars up to $10,000 a month.

A sensible order to learn it in:

1. Calculate your break-even ACOS and your target bid for your top 5 keywords.
2. Run the weekly bucket sort by hand for at least 8 weeks.
3. Add placement adjustments once you have a month of placement data.
4. Only then look at automation, starting with Amazon’s free options before any paid tool.

Frequently Asked Questions

What is Amazon bid optimization?

Amazon bid optimization is adjusting your cost-per-click bids on each keyword, target, and placement so each click is worth what you pay. The core formula is conversion rate × selling price × target ACOS. You raise bids on keywords that sell below target, lower them on keywords that sell above target, and cut the ones that spend with no sales.

How often should I change my Amazon PPC bids?

Weekly is right for most sellers. Amazon’s sales attribution takes several days to settle, so daily changes on a small account mostly react to noise. At launch, wait seven days before your first bid changes, except for keywords with 1,000+ impressions and no clicks, which need a small bid raise.

Are AI bid tools worth it for a new Amazon seller?

Usually not in the first few months. Bid algorithms need click and order data to work, and a new listing doesn’t have much. The fee can also be a large share of a small ad budget. Start manual with dynamic bids down only, and revisit tools once you have several products or hundreds of active keywords.

Should I use dynamic bids up and down or down only?

Start with down only. It lets Amazon lower your bid when a sale looks unlikely but never charge more than your max. Up and down lets Amazon raise bids by up to 100% for Top of Search, which can be useful on a proven campaign you want to push, but it’s expensive on keywords with no history.

How much should I bid on Amazon PPC?

Use target bid = conversion rate × average selling price × target ACOS. At a 10% conversion rate, a $30 price, and a 30% target ACOS, that’s $0.90. At launch, before you know your conversion rate, start at the low end of Amazon’s suggested range and adjust from real data after a week.

Should I pause a keyword with a high ACOS?

Not if it’s making sales. A high ACOS on a selling keyword means the bid is too high for that conversion rate. Lower the bid instead. Pause or negate a keyword when it reaches about 20 clicks with zero sales, or right away if the search is irrelevant to your product.

The Bottom Line

Amazon bid optimization isn’t a contest between you and software. Both use the same math: what a click is worth based on conversion rate, price, and the ACOS you can afford. A tool does it faster and across more keywords. You do it with context the tool doesn’t have, like a keyword you’re pushing for rank, a high-return category, or a main image that needs fixing before any bid will help.

For a launch or a small account, manual bidding with dynamic bids down only is the right call. It’s free, it works on thin data, and it teaches you what good looks like. As your catalog grows into hundreds of keywords, rules or an algorithmic tool start earning their fee, ideally with your top keywords still under your own control. Learn the manual method first. Then any tool you add later is something you can check, not something you have to trust blindly.