272,889 ad orders · 186,461 new to the brand · 32 weeks, Mar 2 to Oct 11, 2026 · five brands we manage
The supplement brand's numbers cover its orders through keywords and targets: 56.5% of its ad-credited orders and 72.3% of its new-to-brand ones. The rest came through ads with no keyword attached (section 10).
01How we split the orders
When Amazon credits an order to an ad, Amazon Marketing Cloud (AMC) also says whether the shopper was new to the brand: no purchase of any of the brand's products on Amazon in the past 12 months. Our app pulls this every week, for every keyword and target, with its match type and ad type. That lets us see what share of each keyword's orders came from new customers, and what each new customer cost.
The sample is five brands we manage: supplements, insect repellent, home fragrance and snack bars in the US, and skincare in Germany. The data runs from March 2 to October 11, 2026, 32 weeks. The snack bar brand's data ends on August 9.
We sorted every keyword and target into types. Search keywords come in three kinds. Brand-name keywords contain the brand's own name, a misspelling of it, or a product name only that brand uses. Other brands' names are their own type. Generic keywords contain no brand name at all, and we split them into 1 or 2 words and 3 or more. On the targeting side we kept the brand's own product pages apart from other brands' pages, and added category targets, auto targeting and Sponsored Display remarketing.
Two things to keep in mind. AMC leaves out keywords with too few shoppers behind them, so small keywords are under-counted. The keyword rows still hold 91 to 97% of each brand's ad spend. And Amazon credits each order to the last ad the shopper clicked (or, for some ad types, saw) in the 14 days before buying. That matters in section 04. Section 10 lists what this data can't tell.
02Among search keywords, brand names bring the lowest share of new buyers
On brand-name keywords, 57% of the supplement brand's ad orders came from shoppers new to the brand (84,166 orders through its keywords and targets). For the insect repellent brand it was 77% (7,092 orders), for the skincare brand 70% (11,937) and for the snack bar brand 33% (14,759). The home fragrance brand gets almost no brand-name searches, 18 orders in 32 weeks, so it's left out of every brand-name comparison.
Generic keywords bring a higher share for every brand: 77%, 93%, 88% and 63% for the same four, and 91% for home fragrance. The gap is 16 to 30 points, and it holds week after week, as section 08 shows.
A dot needs at least 30 new-to-brand orders, so some rows have fewer than five. Hover a dot for its brand and counts.
Show the numbers
| Keyword type | Supplements US | Insect repellent US | Skincare DE | Snack bars US | Home fragrance US |
|---|---|---|---|---|---|
| Your brand name | 57% of 84,166 | 77% of 7,092 | 70% of 11,937 | 33% of 14,759 | too few |
| Other brands' names | 79% of 2,072 | 95% of 1,318 | 89% of 358 | too few | 98% of 98 |
| Generic, 1 or 2 words | 74% of 20,113 | 93% of 10,132 | 88% of 9,733 | 63% of 4,756 | 90% of 7,294 |
| Generic, 3 or more words | 82% of 13,073 | 93% of 8,751 | 88% of 305 | 65% of 1,055 | 95% of 2,440 |
| Your own product pages | 58% of 7,976 | 82% of 4,255 | 68% of 3,668 | 31% of 227 | 92% of 446 |
| Other brands' product pages | 70% of 12,362 | 94% of 2,767 | 89% of 3,117 | 61% of 873 | 94% of 1,345 |
| Category targets | 72% of 6,091 | 92% of 218 | 72% of 136 | 59% of 1,600 | too few |
| Auto targeting | 61% of 8,067 | 91% of 2,420 | 83% of 7,492 | 59% of 5,742 | 92% of 545 |
| Remarketing (Sponsored Display) | 69% of 934 | 70% of 229 | 51% of 164 | 9% of 1,104 | too few |
Other brands' names bring the highest share of the search keywords, 78% to 98%. Someone searching for a competitor rarely owns your product yet. On the targeting side, remarketing has the lowest median share of all, and section 06 comes back to it.
"New to the brand" is narrower than it sounds. A shopper who buys the brand on its own website, or in a shop, counts as new the first time they buy it on Amazon. So does someone who last bought it more than a year ago. The supplement brand sells mostly through its own website, so our best guess is that part of its new brand-name buyers already knew it well. Amazon's data can't show that.
03Generic keywords cost 1.8 to 9.2 times as much per new customer
Brand-name clicks are cheap, and in three of four brands most of their orders still came from new customers. That makes the brand name the cheapest search keyword for new customers in every brand that has one. Only ads on the brand's own product pages came close, and for two brands they were cheaper still. A new-to-brand order from a brand-name keyword cost 0.23 to 0.62 times the brand's own average cost per new customer. From generic keywords, in the same four brands, it cost 1.1 to 2.1 times the average.
We show cost as a multiple of each brand's own average, because the brands sell very different products, in two currencies. Cost here is ad spend divided by new-to-brand orders.
Hover a dot for its brand, its new-to-brand orders and its share of the brand's spend.
Show the numbers
| Keyword type | Supplements US | Insect repellent US | Skincare DE | Snack bars US | Home fragrance US |
|---|---|---|---|---|---|
| Your brand name | 0.23× | 0.62× | 0.49× | 0.25× | too few |
| Other brands' names | 2.1× | 1.2× | 1.4× | too few | 0.68× |
| Generic, 1 or 2 words | 2.1× | 1.1× | 1.3× | 1.5× | 1.0× |
| Generic, 3 or more words | 2.3× | 1.1× | 2.5× | 1.4× | 1.0× |
| Your own product pages | 0.24× | 0.55× | 0.34× | 1.5× | 0.73× |
| Other brands' product pages | 1.1× | 1.4× | 0.79× | 1.7× | 0.87× |
| Category targets | 1.7× | 2.2× | 3.3× | 1.1× | too few |
| Auto targeting | 1.9× | 0.91× | 1.5× | 1.3× | 1.4× |
| Remarketing (Sponsored Display) | 1.7× | 1.9× | 4.1× | 4.8× | too few |
Brand by brand, a new customer from generic keywords cost 9.2 times one from the brand name for the supplement brand. For the snack bars it was 5.8 times, for skincare 2.7 times and for the insect repellent 1.8 times.
Month by month (months with at least three weeks of data) the ratio moves: 7.0 to 12.0 times for the supplement brand, 1.6 to 3.6 for skincare, 5.3 to 6.5 for the snack bars and 1.2 to 2.0 for the insect repellent.
Show the numbers
| Brand | Generic ÷ brand | 95% interval | Lowest month | Highest month | Full name only |
|---|---|---|---|---|---|
| Supplements · US | 9.2× | 7.4 to 10.7× | 7.0× | 12.0× | 7.1× |
| Insect repellent · US | 1.8× | 1.4 to 2.1× | 1.2× | 2.0× | 2.0× |
| Skincare · DE | 2.7× | 1.9 to 4.1× | 1.6× | 3.6× | 2.6× |
| Snack bars · US | 5.8× | 4.7 to 6.7× | 5.3× | 6.5× | 5.8× |
The size of the gap depends on where you draw the line around a brand search. If only the full brand name counts, the supplement brand's ratio drops to 7.1 times. If a few searches for the product type the brand is known for count as brand searches too, it's 8.2. For the skincare brand the same two choices give 2.6 and 2.4. The gap stays wide either way.
04The brand name gets the credit for a big share of new customers
In the four brands with brand-name searches, brand-name keywords took 9 to 14% of the keyword spend, and Amazon credited them with 17 to 48% of the new-to-brand orders in the keyword data. For the supplement brand that's 11% of the spend and 48% of the new customers through its keywords and targets. Counting its ads with no keyword attached too, the brand name has 35% of its ad-credited new customers. The single biggest brand keyword, usually just the brand name, brought 15 to 36% of them.
This is where the credit rule matters. Amazon gives the order to the last ad before the purchase. A shopper who first clicked a generic ad, or saw the brand on Instagram, and then searched the brand name a few days later is credited to the brand keyword. AMC can trace those earlier ads, but it takes a separate query that we haven't run yet. So the cheap new customers on brand keywords are an upper bound on what brand keywords do on their own.
It doesn't make them worthless. Somebody has to show up when a shopper types your name, and our branded CPC study showed why those clicks cost so little. It does mean the brand name's numbers flatter it, and any generic keyword that sent one of those shoppers looks worse than it is.
05Other brands' names bring new buyers at about generic prices
Bidding on another brand's name brought the highest share of new buyers of any search keyword. It was 78% for the supplement brand (1,626 new-to-brand orders), 95% for the insect repellent (1,249), 89% for skincare (319) and 98% for home fragrance (96). The snack bar brand had too few to count (28).
Those new customers weren't cheap. For the supplement brand one cost 2.1 times its average, about the same as its generic keywords (2.1). For the insect repellent it was 1.2 times (generic 1.1), for skincare 1.4 (generic 1.3). Only the home fragrance brand's were cheaper, 0.7 times, on 96 new-to-brand orders. These keywords were a small part of the spend, 0.6 to 4.6%.
06Remarketing has the lowest share of new buyers and the highest cost per new one
Sponsored Display remarketing shows ads to people who viewed or bought the product before. Viewing a product isn't buying from the brand, so it still reaches plenty of new buyers. At the insect repellent brand 70% of its orders were new to the brand, at the supplement brand 69%, at skincare 51% and at the snack bar brand 9%. That's 986 new-to-brand orders across the four brands with enough data.
Next to everything else it does worst. It had the lowest share of new buyers of any keyword or target type at three of the four brands, and the lowest median of all (60%, against 63% for brand names). Each new customer cost 1.7 to 4.8 times the brand's average, the highest median of any type. Remarketing was 1 to 3% of the spend.
Ads on the brand's own product pages work like brand-name keywords, because the shopper is already looking at one of the brand's products. The new-buyer share was 58% for supplements, 82% for insect repellent, 92% for home fragrance, 68% for skincare and 31% for snack bars. For four of the five brands it was a cheap source of new customers, 0.24 to 0.73 times the average. Ads on other brands' product pages brought 61% to 94% new buyers.
07Match type and ad type barely change who you reach
Within brand-name keywords, exact, phrase and broad match were 4 to 8 points apart in every brand. Within generic keywords they were 0 to 10 points apart. The widest gap is the skincare brand's broad match on generic keywords: 80% new against 87% for exact and 90% for phrase.
A mark needs at least 30 new-to-brand orders. The snack bar brand had no broad-match orders on its brand name.
Show the numbers
| Brand and keyword type | Exact | Phrase | Broad |
|---|---|---|---|
| Supplements · US, brand name | 59% (17,039) | 56% (20,888) | 55% (3,269) |
| Supplements · US, generic | 76% (13,466) | 79% (11,111) | 72% (1,045) |
| Insect repellent · US, brand name | 76% (2,601) | 73% (644) | 81% (2,236) |
| Insect repellent · US, generic | 93% (8,610) | 93% (4,074) | 93% (4,841) |
| Skincare · DE, brand name | 68% (3,310) | 71% (4,652) | 67% (351) |
| Skincare · DE, generic | 87% (4,819) | 90% (3,504) | 80% (499) |
| Snack bars · US, brand name | 34% (4,340) | 30% (576) | too few |
| Snack bars · US, generic | 64% (2,443) | 63% (845) | 62% (383) |
| Home fragrance · US, generic | 90% (5,782) | 94% (1,178) | 95% (1,913) |
Ad type looks the same. Sponsored Brands and Sponsored Products on the same kind of keyword were 0 to 9 points apart. Who you reach depends on what the shopper searched for. How the keyword is matched, or which ad format shows up, changes little.
08Deal weeks didn't bring a bigger share of new buyers
Prime Day ran from June 23 to 26, 2026, and Amazon's October event from October 6 to 8. We compared each event week with normal weeks: the four weeks before and after Prime Day, skipping the week right next to it on each side, and the four weeks before the October event. The weeks after it are still filling in.
In 8 of 9 cases the share of new buyers was lower in the event week, by 0.6 to 4.9 points. The home fragrance brand on Prime Day was the exception, at +0.5 points. Each drop is small next to how much the share moves anyway. In normal weeks it differs from the weeks around it by 2.1 to 3.4 points (one standard deviation), and every event week falls inside the range of normal weeks.
The two events still look different. On Prime Day the share fell at 4 of 5 brands, and 31% of normal weeks show the same. In the October event week it fell at 4 of 4 brands, by 1.6 to 4.9 points, and none of the 22 normal weeks did that. Within brand-name keywords it fell at 3 of 3 brands in October, which none of the normal weeks matched either. So Prime Day brought no change in the share, and around the October event it dipped a little. The October comparison only has the weeks before it, so other things that changed in early October may play a part.
A week needs at least 20 orders on a keyword type to be drawn. The home fragrance brand has no brand-name line, and the snack bar brand's data ends on August 9.
Show the numbers
| Brand | Event week | New to brand | Normal weeks | Difference | Normal spread | Generic only | Cost per new customer |
|---|---|---|---|---|---|---|---|
| Supplements · US | Prime Day | 61% | 65% (8 weeks around it) | -4.1 pts | ±2.6 pts | -3.0 pts | -18% |
| Supplements · US | October event | 60% | 65% (4 weeks before) | -4.9 pts | ±2.8 pts | -2.9 pts | -23% |
| Insect repellent · US | Prime Day | 84% | 89% (8 weeks around it) | -4.8 pts | ±3.4 pts | -2.1 pts | +15% |
| Insect repellent · US | October event | 89% | 90% (4 weeks before) | -1.6 pts | ±2.7 pts | +0.3 pts | -20% |
| Skincare · DE | Prime Day | 76% | 79% (8 weeks around it) | -3.1 pts | ±3.4 pts | -3.7 pts | +41% |
| Skincare · DE | October event | 78% | 82% (4 weeks before) | -3.6 pts | ±3.4 pts | -5.6 pts | +11% |
| Snack bars · US | Prime Day | 43% | 44% (8 weeks around it) | -0.6 pts | ±2.9 pts | +1.0 pts | -19% |
| Home fragrance · US | Prime Day | 95% | 94% (8 weeks around it) | +0.5 pts | ±2.1 pts | -0.5 pts | -22% |
| Home fragrance · US | October event | 87% | 89% (4 weeks before) | -2.8 pts | ±3.0 pts | -1.3 pts | -7% |
The cost per new customer moved both ways. On Prime Day it fell 18 to 22% for three brands and rose 15% and 41% for the other two. Our best guess is that deal weeks bring more of every kind of shopper, including the ones who already know the brand and wait for the deal, so the share of new ones stays about the same.
09A pricier click doesn't buy a newer customer
Across all keywords, the ones with a higher cost per click brought a higher share of new buyers. The rank correlation is +0.31 to +0.59 in the four brands with brand-name keywords. Most of that comes from the brand name itself, with its cheap clicks and fewer new buyers.
Within generic keywords the link mostly goes away. It's +0.54 for the supplement brand and between −0.12 and +0.20 for the other four, weak links in both directions. Paying more for a generic click doesn't reliably reach more new customers.
ρ is the rank correlation within generic keywords. Hover a dot for its numbers.
Show the numbers
| Brand | All keywords and targets | Generic only | Brand name only |
|---|---|---|---|
| Supplements · US | +0.59 (n=391) | +0.54 (n=182) | -0.02 (n=74) |
| Insect repellent · US | +0.47 (n=187) | +0.20 (n=102) | +0.51 (n=23) |
| Skincare · DE | +0.37 (n=134) | -0.12 (n=51) | -0.37 (n=43) |
| Snack bars · US | +0.31 (n=95) | -0.10 (n=41) | +0.07 (n=13) |
| Home fragrance · US | -0.08 (n=63) | -0.11 (n=37) | too few |
10What we can't tell from this data
- Whether the ad caused the order. AMC shows who bought after an ad, not who would have bought without it. That matters most for the brand name. Many of those shoppers were looking for the brand already, and without the ad some of them would have clicked the organic listing instead. Nothing here measures that.
- Which ad came first. Credit goes to the last ad before the purchase, so earlier ads on the way get nothing. AMC can trace the whole path, but that's a separate query we haven't run.
- Clicks and views together. AMC credits an order to the last ad the shopper clicked or, for some ad types, saw. Our query adds both kinds of credit together, so a display ad that was only seen can get the same credit as a click.
- New on Amazon only. A shopper who buys the brand on its own website or in a shop counts as new the first time they buy it on Amazon, and so does anyone who last bought more than a year ago.
- The supplement brand's scope. Its keyword and target rows hold 56.5% of its ad-credited orders and 72.3% of its new-to-brand ones. The other 43.5% came through ads with no keyword or target attached, most likely display ads bought outside Sponsored Ads, and only 32% of those orders were new to the brand. They're left out of every number on this page.
- Small keywords. AMC drops rows with too few shoppers behind them, so small and long keywords are under-counted.
- What a new customer is worth. We measured what a new customer cost, not what they spend later. A cheap new customer who never comes back can be worth less than an expensive one who orders every month.
- Five brands. The supplement brand is 57% of all orders, so every result is shown per brand. The home fragrance brand has almost no brand-name searches, and the snack bar brand's data ends on August 9.
11What we take from it
- 01
Keep the brand-name keywords, and read their numbers with care. They brought new customers at a fraction of the cost. They also get the credit for shoppers that other ads and channels sent their way.
- 02
Price generic keywords as customer acquisition. Between 63 and 93% of their buyers were new to the brand, and each one cost 1.8 to 9.2 times a brand-name customer. Set their targets on what a new customer is worth to you.
- 03
Remarketing brings new buyers too, at the highest price. At two brands 70% and 69% of its orders were new to the brand, at a third 51%. Each new customer cost 1.7 to 4.8 times the brand's average, the highest median of any type.
- 04
Pick match types for reach and control. The share of new buyers barely changed with exact, phrase or broad match, or with Sponsored Brands against Sponsored Products.
- 05
Don't count on deal weeks for a bigger share of new customers. Prime Day and the October event brought more orders of every kind. The share that was new didn't rise, and around the October event it dipped a little at every brand.
- 06
Run the split on your own account. If you have Amazon Marketing Cloud, new-to-brand orders per keyword are one query away. The numbers here come from five brands, and yours will differ.
12How we measured it
- Data. Amazon Marketing Cloud, through the query our app runs every week: impressions, clicks and spend per keyword or target, match type and ad type, joined with the orders Amazon credits to them. Credit is last touch with a 14-day window, and each order counts in the week of the ad click or view.
- Window. Weeks starting March 2 to October 5, 2026, Monday to Sunday. Credit keeps arriving for 14 days after a click, so we only used weeks whose data was pulled at least 13 days after the week ended. The snack bar brand's data ends with the week of August 3.
- New to the brand. Amazon's flag: no purchase of the brand on Amazon in the past 12 months. Shares are new-to-brand orders divided by all ad-credited orders of the advertised products. AMC's sales figure also counts the brand's other products bought after the ad, so we don't report shares of sales.
- Keyword types. We built a list of each brand's own name, its misspellings and its product names, and a list of other brands' names from the keywords the five brands bid on. Generic means neither appears. We split generic keywords by word count. In German one word often does the job of two or three, so that split is rough for the skincare brand. Amazon's own keyword groups for brand, peer brand and category terms go to the matching type.
- Cost. Ad spend divided by new-to-brand orders, as a multiple of the same brand's average over all its keywords and targets. Currencies are never mixed. A brand and type needs at least 30 new-to-brand orders to be shown.
- Coverage. AMC drops rows with too few shoppers. The keyword rows hold 91 to 97% of each brand's ad spend. For the supplement brand's ads with no keyword attached, see section 10.
- Intervals. The ratio interval resamples keywords, 2,000 times. The monthly ratios use months with at least three weeks of data.
- Deal weeks. Prime Day and the October event come from the deal dates in our records. To judge the event differences, we ran the same comparison for every normal week and looked at where the event weeks fall in that spread.
- Re-check. A second analysis, written separately and run on the raw tables, reproduced the numbers on this page in two rounds: the original findings first, then the numbers added after them.