Do Amazon ads ever actually become profitable?

@rewrite_rachel OP

1mo ago

I've been running Amazon ads for two months and my ACOS is sitting around 80 percent, which means I'm clearly losing money on every sale. Before I quit entirely I want to know: do these ever turn profitable for normal authors, or is it a money pit dressed up as marketing? What am I missing?

11 comments

@margins_mike1mo ago

The number that matters isn't ACOS in isolation, it's ACOS against your break-even, which is your royalty divided by your price. On a $4.99 ebook at 70 percent you keep about $3.49, so your break-even ACOS is roughly 70 percent. At 80 percent you're only slightly underwater on that ad, not doubling your money away. The fix is bids and targeting, not quitting.

@rewrite_rachelOP1mo ago

Okay this reframes the whole thing. I was treating 80 percent like a disaster when my break-even is apparently around 70. So I'm leaking a little, not hemorrhaging. That's a tuning problem, not a 'burn it down' problem.

@margins_mike1mo ago

Exactly. And that's before read-through, which is the part that makes the whole thing work. If you have a series, an ad on book one can run at 100 percent ACOS or worse on purpose, because the profit shows up in books two through five that you paid nothing to advertise. Judge the series, not the single ad.

@ananke_writes1mo ago

Read-through is the entire game for me. Book one ads run at about 90 percent ACOS, a deliberate loss. But my read-through to book four means each book-one buyer is worth roughly $11 in royalties across the series, against a $4 ad cost to acquire them. The standalone metric screams 'unprofitable' while the series math is comfortably green.

@stetson_dan1mo ago

How do you actually calculate read-through to put a real number on a book-one reader? I've never trusted my own math here.

@ananke_writes1mo ago

Take book two's sales divided by book one's sales over the same window, that's your one-to-two read-through percentage, then chain it down the series. If 40 percent go one to two and 80 percent of those continue each step after, you can estimate total royalty per book-one buyer. Mine lands near $11. Once you know that number you know exactly how much you can pay to acquire a reader.

@penname_pending1mo ago

Negative keywords are where most wasted spend hides. After a month I pulled my search term report and found I was paying for clicks on author names and titles nothing like my book. I added 60-odd negatives and my ACOS dropped from 75 to about 55 in two weeks with zero other changes. Mine that report weekly, it's free money left on the table.

@deadline_dodger1mo ago

Worth a hard truth: ads amplify, they don't fix. If your cover, blurb, or also-boughts are weak, ads just buy you clicks that don't convert and your ACOS stays ugly no matter how you bid. Before blaming the ads, check your product page conversion. A great page at a mediocre bid beats a great bid on a page that doesn't sell.

@rewrite_rachelOP1mo ago

Reporting back after a week of this thread's advice: pulled the search term report, added negatives, dropped bids on click-only keywords, and stopped panicking about the single-ad ACOS since book one feeds a trilogy. ACOS down to 62 and trending. The 'judge the series not the ad' point was the unlock. Thank you all.

@scrivener_refugee1mo ago

Bidding discipline matters too. I start every keyword low, around the suggested bid or under, and only raise bids on terms that have actually produced sales, never clicks alone. Auto campaigns to discover terms, then graduate the winners into exact-match manual campaigns where I control the bid tightly. Letting Amazon spend on broad match unsupervised is how the money pit forms.

@query_trenches1mo ago

Standalone books are genuinely harder to make pay on Amazon ads because there's no read-through to subsidize acquisition. With a single title your ad has to profit on that one sale, so your break-even ACOS is a hard ceiling. If you only write standalones, ads can work but expect thin margins and lean harder on the list and organic instead.