Very volatile ROAS



I am new to selling on Amazon (started beginning of November 2021) and therefore also new to Amazon PPC. I have a Sponsored Product campaign which I spent most of the first month playing with to optimise it (as best I could with limited data!) and trying to understand it. The last few weeks I’ve pretty much left it alone so it’s more stable, and I haven’t changed my listing or pricing much in that time. But I’m noticing my ROAS is incredibly volatile and wondered if this was normal. For example, the last few days of orders on my campaign is 4, 8, 10, 4, 3, 15, 2, 0 (I understand the last few days may not be up-to-date yet). As I said, this is without changing much (if anything). For confirmation, this is using my entire £100 budget each day. I understand it is kind of luck who clicks on the link and if they buy it, but it seems a bit volatile. Does anyone have any advice or insight? I know I would be better off having weeks/months worth of data rather than days but for now this is what I’m working with.

Thanks in advance,


Probably all sales are volatile on Amazon. They can go up and down, almost literally according to the weather, and a whole range of other factors, some of which are unidentifiable. There are regularly threads on here asking “Why have my sales dropped?” and the answer is usually “Who knows?”


That’s what I figured. Makes it pretty hard to really optimise a campaign then. I thought I’d nailed it the day I got 15, then the next day I only got 2. Oh well. Thanks for the reply :slight_smile:

  1. What types of campaigns do you use? What is your advertising structure?
  2. Have you recorded your sales keywords and ASINs? Is there anything to notice?
  3. Is your product seasonal?
  4. You know, it’s Christmas time after all.


Its a manual keyword targeting sponsored product campaign. I have down-only bidding. It’s the only one I’ve really tried for now. I’ve spent a lot of time adding a lot of negative keywords to remove noise clicks. I’ve tried to generally increase good performing targets (keywords) and reduce bids for poor performance. But it does seem a bit random so thats hard because one day one keyword performs well and the next it doesn’t! I know I should use longer time frames than day-by-day but for now this is what I’ve got as I’ve only been doing it a few weeks.

It’s not a seasonal product.

Yeh I figured Christmas will affect things but can’t quite work out in my head how I would expect it to affect things. If people are searching for my product because they need it, then I don’t see why it being Christmas would affect whether or not they end up buying it. But I am new to all this so there is a lot I don’t know yet!


I suggest you try the campaign structured approach: Auto - Exact - Broad - Product, for example. All under one budget, of course. Auto campaigns sometimes perform miracles. And sometimes Exact KW is doing well, but the same KW as a Broad would be nothing but a waste of money.
I would also suggest trying Sponsored Display campaign with pay for thousand impression model. If set correctly they are cheap and are a great tool to reach a broad audience on and off Amazon. But if I remember that correctly, you should have a registered Brand.
Yes, the day is too short as a time frame. Amazon suggests 2 week period as the sale could be attributed to the Ad two weeks later. I, though, would suggest 1 week, because if a sale occurs 3 weeks later than Ad, you won’t see it in metrics anyway.
Well, people like to watch products and imagine things. Have you tried coupons/vouchers for example? The number of clicks is always higher than those of redeems.


Thank you so much for taking the time to respond in so much detail! I haven’t tried an auto campaign yet because I (perhaps naively) figured I was better placed (than Amazon AI) to adjust bids according to results and what i thought people would search for etc., because the product is quite niche. Even with my own manual targeting, I have broad, exact and phrase and I price them all separately. I’m actually doing quite well from broad & phrase because 1) I have added so many to the negative keywords that I dont get much wacky stuff any more and 2) the bids are much cheaper so I can afford more lost clicks. When you say ‘Auto campaigns sometimes perform miracles’ -> is this based on any particular feature? Or is it more anecdotal and its perhaps just worth a go for a few weeks?

Any thoughts on dynamic bids - down vs up & down? I read you should keep it down only until you have a few weeks of data then perhaps switch it to up & down but I havent made the switch yet.

I have tried Sponsored Brand but not Sponsored Display so I’ll give that a go - thanks!

I had a coupon around Black Friday time for 10% off and as you said, I had many more collections than redeems. But I ended the coupon because I thought it was ridiculous for Amazon to charge me for redemptions.

Thanks again!


Amazon AI is a strange thing. Sometimes it performs excellently and sometimes the opposite. I read once in an article regarding PPC ads: “<…>Everyone hates Auto campaigns. And everyone runs Auto campaigns. The market is always changing, so are the keywords<…>”
Before such structure in different campaigns, I also liked Phrase keywords, but after separation - not so much. But that’s our product and my experience.
No, sometimes Auto campaigns just guess wonderfully precise.

I prefer dynamic bids down only as ACoS is one of my top concerns.
Coupons greatly attract views thanks to green coloration “Save 10%”, but if the product price is low, redemption cost is comparatively too high.


Thanks a lot. I’ll have a play with auto campaigns and VCPM. Cheers

closed #10

This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.