Amazon sells advertising by the click, so the bill arrives on traffic rather than on orders. CPC stands for cost per click, and the auction sets that price week by week. That price is half the decision, and the other half is evidence about the term, scored against the unit margin.
The short version
- The click price is a snapshot. It reports what the auction charged last week, and says nothing about the market sitting behind the term.
- A bid is a research decision. We put 90+ data points behind a market before entering it, and the same evidence decides which terms earn budget.
- Six rows, weighted, with a pass mark you set. Score what you know, weight each row by what it costs when it is wrong, then read the total against a bar you set first.
- An unknown scores zero. Every blank row is a fact your clicks are buying while research hands it over for nothing.
- Re-score on the four signals. Rating trend, return rate, conversion rate, and cost of customer acquisition trajectory decide the term across 60 to 90 days.
Score the bid, not the click price
Most sellers judge a bid against what the same term cost them last week. That comparison holds no information about the product the click lands on. Score six things instead, and weight each one by what it costs you when it is missing.
Work one term at a time, starting with whatever spent the most last month. A row you can evidence takes its full weight. A row you are guessing at takes zero, because a guess and a blank cost the same money.
| Row | What a full score looks like | Weight |
|---|---|---|
| Unit margin, current | You know what the unit keeps after landed cost, referral fee, and fulfillment, priced off the latest freight quote | 30 |
| Order evidence on the exact term | The term carries its own order history instead of hiding inside a campaign average | 20 |
| Market direction | Demand behind the term grows year over year rather than merely looking busy today | 15 |
| Return rate | Under 8% on the product the click lands on, so returns are not eating the orders | 15 |
| Rating position | The listing sits 0.2 stars or more above the niche average | 10 |
| A costed alternative | You have priced the same order through another route, a promotion or a creator campaign | 10 |
Flapen figures as of September 2026. The 8% and the 0.2 stars are our thresholds. The weights and the pass mark are yours.
Write your pass mark down before you score anything, so the bar cannot drift toward the bid you already wanted. Seventy out of 100 is a defensible place to put it.
A term below your mark does not get a bigger bid. It gets its missing row filled in.
The evidence behind one bid
Every launch decision here runs on 90+ data points, and bid decisions inherit the same set. Market size, growth trajectory, return rate, segment dynamics, and the rating gap all sit inside it.
A review count is nowhere near the top of that list. It is only the easiest thing to count.
By the 2026-08-26 capture, 193,753 niches had been scored through that set, and 4.8% of them cleared. The rejection rate is the whole argument. Most markets look healthy in a snapshot of today and come apart on the trajectory.
Fifty operators here run about 70 brands by hand across all 23 Amazon marketplaces. The same six rows decide a bid raise on any of them.
Blank rows fill from sources that cost nothing. Margin comes from the supplier invoice, the freight quote, and the fee schedule you are already paying.
Order evidence comes from your own search term data, read at the term level rather than the campaign level. Return rate comes from your returns report, per product and never per account.
Rating position comes from the negative reviews under what already sells, which is where the rating gap is measured. The market tells you where to innovate. You do not guess.
When the auction moves and the score does not
The auction reprices weekly. Your six rows barely move in that time, because margin, returns, and rating do not change on a weekly cycle.
So a rising click price is not a reason to re-score. It is a reason to check whether the term still clears your mark at the new price.
Re-score on a window instead of on a price change. Four signals decide it: rating trend, return rate, conversion rate, and cost of customer acquisition trajectory, read across 60 to 90 days.
A term that misses your mark twice inside that window is not underbid. The product behind it cannot pay for the traffic, and no bid repairs that.
What an agency will not tell you about CPC ads
Score the person raising the bids on the same shape, and write the pass mark down before the first row. Weight each row by what it costs you when it is missing.
| Row | What a full score looks like | Weight |
|---|---|---|
| Evidence behind a bid change | They name the term, its orders, and the margin on the destination product, in writing | 30 |
| Fee shape | A flat fee that does not rise when your spend rises | 25 |
| What the report leads with | Profit per product beside ad cost per product, ahead of impressions and ad-attributed sales | 20 |
| A stop rule | They can say what would make them tell you to spend less | 15 |
| Exit terms | Month to month, and you keep the account, the campaigns, and the creative | 10 |
Flapen figures as of September 2026. Score every provider against it, this one included.
Here is our own scoring, so you can mark it. The fee is flat, from $800 a month for one product to $2,400 for five, with all 50+ services included. No commission sits on your ad spend.
The contract runs month to month on 30 days of notice. On the day you leave you keep the Seller Central account, the campaigns, the creative, and a written handover. If we land under your pass mark, hire somebody else.
Related answers
- Sponsored display
- Amazon display ad
- Rank agencies by Amazon DSP case studies
- Top ecommerce audit firms for marketplace scale-up
- Amazon account measurement and audits: the complete guide
One free thing to do this week. If you run between one and five products, list the five search terms that spent the most last month. Score each one against the six rows above.
Count the rows you cannot fill. Those blanks are the research your clicks are currently paying for at auction prices.
Get those terms and the products under them audited in writing, at no charge, inside 48 hours, from Flapen.
