Before increasing book ad spend, I want to know what the existing spend bought: visits, downloads, purchases, or enough earnings to cover its cost. Those are different answers, and a cheap click doesn’t settle any of them. If I can’t connect the spend to the goal, increasing the budget gives me a larger bill for the same unanswered question.
This is a worked example of the checks I’d make, using invented figures. It isn’t a client result or a benchmark for what your book should achieve. The author in the example has a series starter to sell, a small test budget, and a perfectly reasonable question: should I put more money behind this?
Start with the question the ad was meant to answer
Suppose the author spends $60 sending readers to a page for Book 1. The ad platform reports 120 outbound clicks. That’s $0.50 per click. It sounds manageable until I ask what happened after the click, because the money was meant to help sell the book, and nobody is paid royalties for a reader opening a browser tab.
I’d write down the dates, country, advertised format, actual destination and intended action before looking at the totals. An ad for a paperback delivered to readers in a country where shipping costs more than the book is a different test from an ebook ad sent to a local retailer. If several ads, countries and offers are combined in one total, separate them before deciding which one deserves more money.
My guide to choosing book marketing metrics covers the reporting set. Here the job is narrower: follow one paid test far enough to make its next spending decision.
Keep the stages separate
In this example the site’s analytics records 94 visits associated with the campaign, 24 clicks to a retailer, and 16 sample downloads. Those numbers aren’t a single shrinking queue of identified people. One reader might download the sample and click to the store; another might return later. The ad platform and website can also count different actions. I’d check definitions, date ranges and the actual link before explaining the gap.
Google documents outbound click measurement in GA4 as measurement of clicks leaving your site. That retailer click tells me someone followed the link. It doesn’t tell me they completed a purchase on the retailer’s site. A sample download likewise doesn’t show that the file was opened, read or enjoyed.
The first useful inspection is wonderfully unglamorous: tap the ad’s destination on a phone. Is the book the one the ad promised? Can a new reader find its place in the series? Is the format available? Does the sample open? Do the buy buttons go to the right book and store? Record what actually fails. A working link won’t prove the offer is persuasive, but a broken one gives you something concrete to repair.
Do the earnings math before calling it a win
Now suppose the author can reliably connect eight purchases to this test, and each produces $2.40 in earnings after the relevant retailer deductions. That’s $19.20 in earnings against $60 in ad spend, a $40.80 shortfall before any other business costs. The ad cost per purchase is $7.50. Those are illustrative inputs, not a royalty-rate claim.
If the author also sees twelve purchases in the overall sales dashboard, I wouldn’t automatically credit all twelve to the ad. Some may have come from a newsletter, an existing reader or a search. Keep the eight attributed purchases and twelve total purchases as separate observations. Attribution assigns credit under a reporting method; it doesn’t prove the ad created a sale that would otherwise never have happened.
Amazon explains why Ads and KDP reports differ: Ads reports attributed activity, while KDP includes activity regardless of its source. Ad sales value and author earnings are also different amounts. I compare the cost with the earnings available to pay it, using the correct format and market, rather than treating the sales column as money I keep.
If you need to establish those underlying numbers first, start with reading your KDP dashboard. I would also keep free downloads, paid copies and subscription page-read earnings on separate lines. Adding them together produces a larger number with no useful meaning.
What about the rest of the series?
Book 1 might reasonably cost more to acquire than it earns on its own if readers continue to later books. But a series existing isn’t evidence that these particular readers continued. I wouldn’t fill the shortfall with hoped-for Book 2 income.
Use your own series read-through evidence to decide whether you have a defensible estimate, and label it as an estimate. Give readers time to finish; don’t compare a new group of Book 1 buyers with years of accumulated Book 2 sales and call the ratio this campaign’s result. If you can’t connect the later earnings with enough confidence, calculate the first-book result and keep the possible later return separate.
The decision I would make here
With the example as it stands, I wouldn’t raise the spend. I’d pause this test, check the purchase route and decide whether there is a specific change I can afford to test. Eight purchases show that some people bought. They don’t establish whether targeting, the ad promise, price, the page or some combination caused the shortfall.
If the link is broken, fix it before spending again. If the ad promises a complete standalone and the page reveals the middle of a continuing series, repair that mismatch. If everything works but the economics still don’t support the cost, a lower acquisition cost or a different offer may be necessary. Changing the cover, audience, price and description together might improve the total, but it would leave me unable to tell which change helped.
Before restarting I’d write a bounded test: the one change, the maximum I can afford to lose, the result I’m measuring and the date I’ll review it after allowing for the platform’s reporting delay. When the cap is reached, pause and assess. There isn’t a universal number of days or clicks that makes every book’s result conclusive.
Frequently asked questions
Does a low cost per click mean my book ad is working?
It means the reported clicks were inexpensive under that platform’s definition. Whether the ad is working depends on its goal, the actions after the click and the cost relative to the value you can reasonably establish.
Can I keep advertising while I investigate?
Only if you deliberately accept the cost and have a reason to continue. I’d stop traffic to a broken destination immediately and pause a test that has reached its spending cap. You don’t have to buy more data before checking whether the existing data can answer the question.
Check the destination before the next test
The Amazon Listing Field Guide helps you inspect the page readers reach. If the harder part is deciding which result deserves another test, my strategy work starts with your own costs, readers and capacity. See How I Can Help