Keyword dumping?



I have just been reading online (not on this forum) about keyword ‘dumping’ in Amazon PPC campaigns and I’ve found a lot of different advice online suggesting you should keep the number of keywords in a single ad-group down to 100 or fewer. I understand the idea is to optimise by removing low-performing and potentially increasing bids of high-performing, but I dont understand the disadvantage in having lots of keywords. Some people were suggesting Amazon would randomly pick batches of 100 to show each week, but I didnt see any evidence of that. Also, lots of people suggest separating exact, broad and phrase into different ad groups but again i dont fully understand the reasons given i can filter by match type and adjust bids accordingly.

I have a manual product keyword campaign with 1000 keywords. This is because i have attempted to capture every variation/permutation of the search terms i expect. Is this bad? Would I be better just using phrase/broad match to basically do that on my behalf?

For a theoretical listing of a tshirt, available in different sizes and different colours, I have taken the approach to add keywords like:

  1. tshirt blue large
  2. tshirt blue medium
  3. tshirt blue small
  4. tshirt red large
  5. tshirt red medium
  6. tshirt red small

And then I have added each permutation 3 times: once as exact (highest bid), once as phrase (medium bid), and once as broad (lowest bid). It didn’t take me long to hit 1000 keywords!

Any advice would be greatly appreciated. I guess if noone has any strong, evidence-based suggestions I can still just trial what I’m reading online, but I prefer not to follow things blindly and I haven’t yet read anything which fully explains why keyword ‘dumping’ is bad. I’m not using anywhere nearly my budget so it isn’t true for me that keyword dumping is exhausting my budget. I think i have probably 70% of keywords never getting any impressions, but whilst that may be unnecessary, surely thats not a disadvantage???



Personally, I don’t think it’s productive to have every variation/permutation on there.

How many people searched for ‘t-shirt blue large’ ‘t-shirt blue medium’ in the last 30 days?

This is my keyword research week and it’s more technical than that, boring? definitely.

There are some softwares you can use (free trial even) and you need to sort by search volume, KD (keyword difficulty), competitors, keyword trends etc.

I used to flood my manual campaigns like in your method few years ago and the only thing that went up was my advertising invoice. Do amazon pick 100 keywords at random? I don’t know.

Things might have changed since I tried your method so I might be wrong of course. Maybe you should try the 1000-keyword idea but with a very small budget just to test the waters. Let us know, if possible please.


Thanks a lot for taking the time to reply. To confirm, my tshirt example was made up to demonstrate my predicament.

How many people searched for ‘t-shirt blue large’ ‘t-shirt blue medium’ in the last 30 days?

…even if the answer was none, my question is, is there a disadvantage to still having the keyword (assuming its not taking the position of a better keyword)?

I used to flood my manual campaigns like in your method few years ago and the only thing that went up was my advertising invoice.

I still don’t understand how it can cause a higher invoice, assuming all my keywords are still relevant to my product, and each has as good as chance as any other of leading to a sale.

I’ve attached a screenshot of my ~1000 keyword ad group (994 to be specific). It’s been running since 17th january, although ive been tweaking it a bit along the way. I can see that 881 of my keywords have never had a single impression, and all orders have come from 32 keywords (26 exact, 0 phrase, 6 broad). I don’t really know what to read from this to be honest. I could increase bids on the 32, yeh. But should I remove the 881? Why would I? Thats what I’m trying to understand. Cheers


Let’s just say that amazon will randomly choose 100 keywords even though we are not sure.

Also let us say that the number of searches for ‘t-shirt blue medium’ in the last 30 days is FIVE.

Those 5 clicks will probably not generate you any sales unless if your listing has a STR of 100%.

Why would you want to put a keyword with 5 monthly views on your keyword list?

If you have low ranking keywords, less people will see it and those that click on it will probably not buy it. Two things can happen;

  • Less clicks, No sales, less advertising bill
  • Less clicks, less sales and more advertising bill

None of those result in high sales which is what we all want.

Why not research the best relevant targeted keywords with reasonable to high search volumes amongst other things and choose the best 100 to start with, run the ad then compare the ACOS to the 17.51% you currently have.

One thing though, I’m not big on ads unless for new products. You can improve your back-end keyword, post on social media etc.


Thanks again for the detailed reply.

Truthfully I still don’t fully understand but I’ve decided to just experiment by changing it. So I’ve archived my 1000 keyword campaign and replaced it with 3 ad groups: 1 for exact with 20 targets, 1 for broad with 10 targets, and 1 for phrase with 10 targets. I figure the broad ad group will basically capture all my permutations. Now I’ll give it a few weeks and see!

FYI: it is a fairly new product (started Nov 21) and it’s the kind of product you would only buy once or twice a year, and you would only buy if you needed it at that time, so i am seeing a lot better ROAS from Amazon PPC than my social media ads.


How did it go? I believe 1 week past from your last post.

I am trying to learn PPC as well and this topic was useful for me.

Thanks to you for raising and @ethical_capitalist for commenting.


It’s hard to know what to read from the data so far but I’ll highlight some stats and you can make up your own mind! Here’s a screenshot of my new campaign since I switched, so date range is 10th Feb-17th Feb. Bear in mine the screenshot above was for a longer period 17th jan-9th feb compared with only one week for this one. My new campaign has 15 ad groups and each one only has exact or broad or phrase, and the biggest one has 67 targets in it, most have 14 targets.

I also summarised these key stats before and after i made the structure change:

I’ve also just had a terrible few days of sales (Tuesday/Wed/Thurs) which may skew the figures. I’ve seen another thread on this forum with others saying the same about bad sales, although what’s a bit weird is my clicks are still basically the same, but the conversion is way down (4 orders from 28 clicks = 14% since tues).

Dont know if this helps. I’ll leave my campaign as it is for a few weeks longer and can report back.


Keep going, I believe your ACOS is still good (much better than mine but I am really a newbie so I just play with it for now).

As you said sales are low this week for me too. Lets see what will happen in long run (at least a month).

I wish you good luck and I lots of sales… Keep us posted.

closed #9

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