Do AMS ads come back to life?
There seem to be two points of view about how to handle AMS ads that are coming to a natural end. Some argue that all ads should remain on the books in case they suddenly revive. Others contend that it is best to kill them off (i.e. ‘pause’ them) and start new ones. Such ads include those with notably reducing impressions as well as those which never had many (or any) impressions to begin with.
Here is a brief background to my own case. I have written three books on very different topics, so there is little overlap in their audiences (and therefore little read-through). The first is about the people who provide hospice care (Life in a Hospice); I sell roughly 30 copies each month (via AMS ads and organically). The second tells the stories of young people with HIV and AIDS in the early 1990s (Wise Before their Time); I sell about 15-20 copies a month. The third is about what it is like to be a grandmother (Celebrating Grandmothers), which comes a poor third, selling 3-5 copies per month. All three sell both as e-books and paperbacks.
I have been doing AMS ads for just over a year and have had reasonable success in both the UK and the US. By late-August 2019, I had sold 109 books in the UK via these ads, with an ACOS of 13%. In the US, I had sold 228 books via these ads, with an ACOS of 26%.
The trial
With respect to pausing ads, I have always taken the second position – quickly pausing any ad that is losing impressions or never took off. But in discussing the issue, Laurence O’Bryan at BooksGoSocial suggested that they should remain on the books. I proposed that I undertake an experiment whereby I revived some previously paused ads to see how they fared. I would do this for two months. I called them my Lazarus ads.
At the same time, as a sort of control group, I would chart my on-going ads. This would be done both in the US and UK.
In mid-August 2019, I revived 18 ads in the UK (6 for each book) and 24 ads (9, 8 and 7 respectively) in the US. All were selected as most likely to succeed (ie they were not randomly selected). At the same time, I had 10 on-going ads in the UK (5, 2 and 3) and 9 in the US (3 each). Over the next two months, I carefully documented the impressions, spend, income and sales of all these ads on excel spread sheets. Eight weeks later, in mid-October 2019, I calculated the results.
As a measure of success, counted the number of sales during the trial period. In addition, I chose a crude indicator of total profit or loss, based on the total income gained minus the total spent on clicks. This, of course, does not take into account the costs necessarily removed from my royalties, namely the proportion taken by Amazon and the cost of printing. As I could not tell which sales were which, and as I was trying to limit the demands on my own time from this exercise, this seemed a reasonable decision.
Outcomes
The result of this experiment was quite similar across both the US and the UK. All of the Lazarus ads revived to some extent, i.e. some impressions were added over the period, but these were generally small in number. With increasing impressions came some new clicks and some sales.
In the UK, I sold only 12 books via Lazarus ads during the trial period, all a single book (on hospice care), gaining me £75.07 after expenditure (cost of clicks) is deducted. The other two did not revive much, but sufficiently to garner some clicks which cost me £3.20 and £0.47 respectively. This gives an overall profit of £71.40.
In the control group, in contrast, I sold 25 books over the period (21, 2 and 2). The overall profit from these sales was £103.09 (£82.34, £2.12 and £18.63). The low profit on the second book can be accounted for by frequent clicks on ads, which would have been paused in normal circumstances.
In the US, I did even worse, selling only 7 books via Lazarus ads during the trial period (4, 3 and 0). One sale was ‘lost’, i.e. one week there was one less order than the previous week and the sales figure was accordingly reduced. I don’t know why unless it was a return (or an error). Several ads had so few impressions that there were no clicks during the whole trial. The overall profit from these sales was $22.97 ($16.27, $11.56 and -$4.86), including the one reduction in books sold.
Finally, in the US control group, I sold 17 books during the trial period (13, 3 and 1). One ad was particularly crucial here, representing 9 sales. This ad had done well prior to the trial, but not notably and I have no idea why it suddenly morphed into a ‘golden ad’. The overall profit from these sales was $181.83 ($113.37, $63.72 and $4.74).
Is it worth reviving old ads?
It transpires that it was mildly worthwhile in this case to revive old ads, as I made £71 and $23 respectively from the experiment (although, as explained, the actual figures would be lower.) But these are not at all large numbers. The other income (from the control group) I would have made in any case. In fact, because I did not pause any ads during the period in order to keep the experiment going, I lost some income from clicks on ads that would otherwise have been paused.
The real question, as in many trials of this kind, is how to recognise which ads to revive. Had I revived only those which were successful, and quickly paused any that were not, the numbers would have been more favourable. I don’t have the answer, aside from the fact that, in general, the earlier the ad, the weaker the revival, presumably because ads in the autumn 2018 could be based on lower costs per keyword.
My own reaction is to be somewhat less hasty in pausing ads, but it still does not seem worthwhile to spend too much time on such ads overall.
This was a small experiment and it may be that others have done much more sophisticated and extensive exercises. I would be delighted to hear from them.
Ann Richardson
http://www.annrichardson.co.uk
