Amazon suggested bids are useful in the same way a weather forecast is useful. They tell you something about the auction climate. They do not tell you whether your SKU can afford to go outside without a coat.
Open Campaign Manager, add a keyword, and Amazon shows a low bid, a suggested bid and a high bid. For a self-service brand owner, that can feel like a shortcut. Amazon has the auction data. Amazon knows what competitors are bidding. Amazon wants relevant ads to win. So surely the suggested bid is a sensible starting point?
Sometimes, yes. Often, not without translation.
The named mistake I see is copying Amazon’s auction recommendation into advertising software as if it were a profit recommendation. A suggested bid can help you win impressions. It does not know your landed cost, refund rate, current coupon, next replenishment price, stock cover, organic rank, campaign role or cash-flow pressure. The auction may be right and the business decision can still be wrong.
My stance: treat Amazon suggested bids as auction weather, not bid instructions. Before your ad software turns a suggested bid into a starting bid, bid ceiling or automation rule, it needs profit permission from the SKU, keyword and campaign role.
This guide is for brand owners managing Amazon Ads themselves, usually from around €1.5K monthly spend. At that stage, you have enough budget for suggested bids to move real money, but not enough waste tolerance to “let the algorithm learn” across every keyword. FiveX helps by connecting ad targets to SKU margin, fees, returns, stock and automation rules, so suggested bids become one input in a controlled decision instead of the decision itself.
What current suggested-bid advice gets right
The existing guidance is not bad. Perpetua explains Amazon PPC structure well and often uses Amazon’s suggested bid as a practical starting point for automatic campaigns. Their dynamic bidding advice also makes an important point: the suggested bid does not automatically change when you add dynamic up-and-down bidding or placement multipliers, so the real auction ceiling can become much higher than the number you first saw.
Teikametrics explains the classic bid formula clearly: average order value multiplied by conversion rate multiplied by ACOS target gives a maximum CPC. They also recommend structuring campaigns by margin, which is one of the most underrated PPC basics. If a 40% margin product and a 14% margin product share one target and one bid logic, your reporting is already muddy.
BidX frames modern Amazon PPC as more than bid adjustments: campaign structure, retail readiness, creative execution, budget governance and automation have to work together. Pacvue’s retail media content makes a similar point at enterprise level, where share of voice, dayparting, budget rules, inventory and profitability-aware automation all influence bid decisions.
Helium 10 is useful for launch-oriented sellers because it connects PPC to keyword ranking and search-term discovery. SellerMetrics and Ad Badger are more cautious: they warn against blindly following suggested bids and recommend using account data, actual CPC and break-even ACOS.
The gap is what happens between “do not follow blindly” and “what should my software actually do on Tuesday morning?” That is where self-service teams need a repeatable system.
The missing layer: a suggested-bid permission ledger
A suggested-bid permission ledger is a simple table your advertising software should maintain before bids go live. It does not need to be fancy. It needs to answer six questions:
- What is the SKU’s real contribution margin after marketplace fees, fulfilment, landed cost, discounts, expected returns and handling?
- What conversion rate can this target reasonably expect for this campaign role?
- What is the maximum CPC the SKU can pay before the target breaks even or breaks the learning budget?
- Is the campaign defending existing demand, converting proven non-brand demand, discovering new demand or recovering visibility?
- Is there enough stock, Buy Box stability and listing readiness to justify buying extra clicks today?
- How far is Amazon’s suggested bid from the account’s actual CPC and the SKU’s profit ceiling?
Only after those checks should software decide whether the suggested bid is accepted, clipped, staged, quarantined or ignored.
This matters because Amazon’s suggested bid is not evil. It is just solving a different problem. Amazon asks: “What bid may help this advertiser compete for this auction?” Your business has to ask: “What bid can this SKU afford for this commercial job?” Those are cousins, not twins.
The formula: convert suggested bids into profit-permitted bids
Start with a practical maximum CPC:
Profit-permitted CPC = selling price × allowable ad share × expected conversion rate
The allowable ad share is not always your full margin. If the SKU has a 32% contribution margin after fees and fulfilment, you may only allow 20% for conversion campaigns, 12% for learning campaigns and 6% for branded defence. The rest protects profit, returns, discount noise and operating risk.
Then compare that profit-permitted CPC with Amazon’s suggested bid:
- If suggested bid is below the profit-permitted CPC and operational checks pass, use it as a starting bid or controlled ceiling.
- If suggested bid is 10-40% above the permitted CPC, clip it to the profit ceiling and monitor impression loss.
- If suggested bid is more than 40% above the permitted CPC, quarantine the target unless there is a deliberate ranking or launch reason.
- If actual CPC is far below suggested bid, use account evidence instead of platform recommendation.
In FiveX, this logic belongs next to SKU profitability and campaign automation, not in a spreadsheet that someone opens once a month. The ad rule needs to know the margin. The margin view needs to know which targets are asking for more money. The operator needs one place to see the trade-off.
Scenario 1: the low-price product that cannot afford Amazon’s enthusiasm
Imagine a kitchen brush selling for €18.95. After referral fees, fulfilment, landed cost, expected returns and payment costs, the SKU has €4.10 contribution margin per unit. The listing converts at around 12% on relevant mid-tail terms. Amazon suggests €1.35 for “dish brush with scraper”.
If you let the full €4.10 margin fund the click, the theoretical break-even CPC is €0.49: €4.10 × 12%. But you should not spend the entire margin just to break even. For a normal conversion campaign, the brand allows 70% of margin to ads, so the permitted CPC becomes €0.34. For a learning campaign, it allows 45%, so the permitted CPC becomes €0.22.
Amazon’s suggested €1.35 is not slightly high. It is four to six times above the sensible bid range for this SKU. The correct software action is not “start at suggested bid and optimise later”. The correct action is quarantine or clip: start at €0.25-€0.34, require 15-20 clicks before changing the bid, and stop the target if spend reaches €8 without an add-to-cart or sale signal.
The trade-off is clear. You may lose impressions. Good. Impressions you can only buy at negative contribution are not growth. They are expensive proof that the auction is too hot for this product’s economics.
Scenario 2: the higher-margin bundle that deserves a staged test
Now take a skincare bundle selling for €42.00. After fees, fulfilment, landed cost, expected returns and a normal promo reserve, it keeps €14.70 contribution margin. The target “retinol night cream set” converts at an estimated 9% based on similar campaigns. Amazon suggests €2.10.
The break-even CPC is €1.32: €14.70 × 9%. The brand’s conversion campaign allows 90% of margin because this bundle has strong repeat purchase potential and healthy stock, creating a permitted CPC of €1.19. The learning campaign allows 55%, creating €0.73.
Here the suggested bid is still too high, but not absurd. The better move is staged permission. Launch the target at €0.90 in exact match if listing readiness is strong, raise toward €1.19 only if actual conversion holds above 9% after 30 clicks, and keep broad or phrase learning closer to €0.70 until search-term evidence arrives.
This is where advertising software should be helpful rather than stubborn. A simple “never use suggested bids” rule would underbid every promising target. A blind “match suggested bids” rule would burn margin. The operator move is to let evidence earn the next bid step.
Scenario 3: branded defence does not need to pay the conquest price
Consider a replenishable supplement selling for €29.95. The brand ranks organically at position two for its own brand-plus-category term, has 21 days of stock and wants to hold a 19% TACoS target. Amazon suggests €1.60 for a branded keyword because competitors are bidding aggressively.
This is where many accounts overpay politely. The campaign shows great ROAS because branded shoppers convert well. Everyone relaxes. But the question is not whether branded ads convert. The question is how much incremental protection you bought versus how much demand you would have captured organically anyway.
For defence campaigns, I like a bid floor and a share-of-voice rule rather than matching Amazon’s suggested bid. If the brand already has organic rank one or two, the software might cap branded bids at €0.45, raise to €0.70 only when paid share drops below 65% or a named competitor appears above the fold, and pause expansion when stock cover falls below 14 days.
FiveX can support this by combining advertising data with organic visibility, stock cover and SKU margin. The product hook is not “AI magically bids better”. The useful hook is that the automation sees the commercial context Amazon’s suggestion does not.
How to build the rule set in self-service ad software
1. Split campaign roles before importing bids
Do not feed all suggested bids into one automation rule. Create separate rules for defence, conversion, discovery and ranking support. A €1.20 suggested bid can be acceptable in one role and reckless in another.
2. Store SKU margin as a live input
Your bid ceiling should change when FBA fees move, coupons go live, return rates rise or landed cost changes. If the ad tool cannot see SKU contribution margin, it cannot responsibly approve suggested bids. In FiveX, this is exactly why ad performance sits next to profitability dashboards and product-level economics.
3. Add operational vetoes
No suggested bid should scale when the product has poor listing readiness, missing Buy Box, low review strength for the category or thin stock. A simple stock veto is powerful: below 14 days of cover, freeze bid increases; below 7 days, reduce or pause non-defence spend.
4. Compare suggested bid with actual CPC
If Amazon suggests €1.80 but your account repeatedly wins clicks at €0.92, let actual CPC lead. Suggested bid is market guidance. Actual CPC is paid evidence. Your software should show the ratio between suggested bid, current bid, actual CPC and profit-permitted CPC.
5. Protect learning budget with hard stops
Discovery campaigns need room to learn, but not infinite patience. Set stop rules by spend and click count. For example: if a target spends one expected gross profit amount without a sale, cut 25%; if it spends two without a sale, pause or move to review. That is kinder to your budget than waiting for blended ACOS to complain later.
What to measure after launch
Once the target is live, judge the suggested bid by outcomes, not by whether impressions arrived. Track:
- Suggested bid acceptance rate: what percentage of Amazon recommendations passed profit permission?
- Clip rate: how often did software cap bids below Amazon’s suggestion?
- Actual CPC gap: how far was paid CPC below or above the suggested bid?
- Contribution after ads: did the SKU stay positive after fees, returns and ad spend?
- Stock-risk spend: how much budget went to SKUs below the stock threshold?
- Learning payback: which quarantined or staged targets eventually earned higher bids?
These metrics make the system smarter. Over time, you learn which categories can safely start closer to suggested bid, which match types need bigger discounts, and which products simply cannot compete in certain auctions without a price, conversion or margin fix.
The operator checklist
Before accepting an Amazon suggested bid, ask:
- Can this SKU afford the CPC after real contribution margin, not gross margin?
- Is this a defence, conversion, discovery or ranking-support target?
- What conversion rate assumption is the bid using, and is it based on evidence?
- Would dynamic bidding or placement multipliers lift the real ceiling above the visible bid?
- Does the product have enough stock and listing readiness to deserve more demand?
- Is actual CPC already proving that Amazon’s suggested bid is too high?
If those questions are not answered, the bid is not ready for automation. It is ready for review.
Final thought: suggested bids need supervision, not suspicion
Amazon suggested bids are not your enemy. They are a useful auction signal. The problem starts when self-service teams treat them as a command from a smarter system. Amazon is smart about Amazon’s auction. It is not accountable for your SKU P&L.
The practical move is to put a profit permission layer between Amazon’s recommendation and your ad software. Let the platform suggest. Let your margins, stock, campaign role and actual CPC decide. That is how a brand owner can use automation without handing the budget to auction weather.
And that is the bigger FiveX philosophy: marketplace advertising should not be optimised in a vacuum. The best bid is not the highest bid you can justify in Campaign Manager. It is the bid your business can afford, repeat and scale profitably.