Amazon PPC Product Targeting and Placement Strategy [Alexa, Play That Amazon Ads Podcast]

Stephen Noch: Welcome back everyone to That Amazon Ads Podcast, where today we are doing something unique. We are describing this as a change-my-mind episode. In our Discord community there was discussion around recent search terms appearing in product targeting campaigns, meaning keywords are appearing in the search term results, with a lot of confusion and speculation. I posted my thoughts on LinkedIn, Mike Danford at Adverio responded, and we went back and forth, so we said let's jump on a conversation and let everyone listen in. Mike, why do not you let everyone know who you are.

Mike Danford: Looking forward to this, I just came back from vacation. I am with Adverio, we primarily work on Amazon, Walmart, Target, and we go after larger catalogs, hundreds to tens of thousands of products across the platforms. With that there are a lot of nuances and different tactics we have to take in the ads.

Andrew Bailiff: Excited to dive into this. When you are targeting product targets, you start to see some search terms show up. When we first did an episode on this, we did not have enough data to see how impactful it was. So as a starting point, what do people need to know about product targeting and some changes happening in the Amazon advertising landscape?

Mike Danford: The caveat I go back to is we have large catalogs with hundreds to thousands of campaigns, so we have to be really careful about cross-contamination of targeting. A common issue is how people want to separate branded or unbranded in their targeting segmentation, and they go into the product attribute targets as well. It is really hard to not have your branded terms, because you cannot negate a search term in your product attribute target campaign. So the way around that is to be really aggressive with a high product placement percentage and only put ASINs in there, not use it for keywords. You want to keep that data as clean as possible, we have to because we have such large campaign sets. We have seen impact from negating a keyword at both the ad group and campaign level. Otherwise, if you are not using just a product placement, you are going to have contamination with search terms in those campaigns.

Andrew Bailiff: Let's unpack that. You are telling us that when we target product targets, we have keywords showing up, and we should segment those so our product targeting campaigns go after ASIN targets showing up on product detail pages, and separately have a campaign going after product targets specific to top of search and rest of search. Why should we do both?

Mike Danford: The biggest thing is, when you negate an ASIN, you are not just negating that product page or that ASIN, you are negating every keyword that ASIN target triggers in that campaign. So if you want to block the ASIN because it has 50, 60, 100 clicks with no sales, you cannot just negate the ASIN, you are going to negate all the search terms that may be favorable for it on the SERP. If you had a campaign split and negate the ASIN because it is underperforming, you can still get the rest of search and top of search placements in the other campaign. That is one reason, and the biggest one, that started this conversation.

Stephen Noch: So the question this individual posed was, since product targeting triggers keywords too, do you think they will compete with actual keyword campaigns, and will this mess up performance over time? Yes, they will compete, but the question is what compete means. And that is where Andrew and I are leaning toward no on messing up performance, and Mike is leaning toward yes. My response was, I do not think trying to separate product targets to only product page placements and keywords to only search placements is worth the time and effort. My main point is that 80 to 90 percent of PPC managers are not doing the three basic things, good campaign structure, the right keywords, and the right bids, before they dive into these technical and nuanced conversations. So Mike, do you think that will mess up performance over time, and are you actually trying to accomplish that product targeting has no search terms appearing in the search term reports with that product-page-only strategy?

Mike Danford: I will answer with a question, because I am curious what you both find. In general, when you go after a PDP or an ASIN versus a search term, do you see any delta in CPCs, ROAS, ACoS, any difference?

Andrew Bailiff: I have some data from a client, let me look. Stephen can give his non-data answer first.

Stephen Noch: I like operating off how I am feeling that day. Do I see a big difference between product targeting versus keyword generally? I separate it by whether it is brand or non-brand, I set different target ACoS, but not for keyword target versus product target.

Mike Danford: And in one of your episodes about your audit process, one of the first things you do is clean up the account, which is segmentation. I would consider that basic, which is what you are referring to. This is that level you need to keep the data clean so you do not have these debates, it segments the data from the onset if you are going to do branded versus unbranded versus competitor, but you need to go all the way down to the targeting. Just do it from the beginning. You are saying there is heavy lifting to do that, but it is not, we do it for thousands of campaigns, you take the same structure and put ASINs in instead of keywords, the top of search and rest of search placements are higher on the keyword campaign, and the product pages higher on the other, it is a mirror. And we see the performance benefits.

Andrew Bailiff: A question for you. With Amazon advertising you can only win a certain number of placements on any given search result. So if you have a keyword targeting campaign, a broad match campaign, and a product targeting campaign all going after the same keywords, only one or at max two is ever going to win that placement. So if you are trying to win a placement with keywords, why would you not just keep it in a keyword campaign?

Mike Danford: You do not know what that keyword's following is, what is indexing for a competitor product or how your product will index over time, so it is a bit of a discovery. What you are referring to is also bid inflation. We deal with that, we have hundreds and thousands of campaigns with very synergistic products all in the same category, so we do see bid inflation, you have to work hard for it. And with Cosmo and Rufus and all this consumer behavior, we have come to find that having that setup actually lets different campaigns trigger differently. Our objective belief, we cannot confirm it in the back end, is that certain campaigns serve differently depending on the demographics and consumer signals of the shopper. So if you have one campaign that works well for this target demographic, this other campaign might do better for another. I want objective data behind that, but that is another thing we are looking at, and it is why we keep the data clean, so we can tell when that happens.

Andrew Bailiff: So you are saying by targeting keywords in a keyword campaign and in a product target campaign, you are potentially reaching different audiences, and you would not reach the same people if you just maximized your keyword campaign?

Mike Danford: I do not have an answer on that, that is what we are trying to figure out. I think the placements are different and the cost competitiveness of the different placements are different from a product campaign versus a keyword campaign. I do think a keyword campaign is going to do better on keyword targets than a product attribute campaign on keywords.

Stephen Noch: But why? Last week we set up a competitor ASIN targeting campaign that was plus 900 percent on product pages, and another that was plus 900 percent for top of search and rest of search. I was excited to come in with data, but they got zero impressions despite five or six dollar bids, and I think that is because other campaigns were bidding and picked up all the impressions. It is really difficult to funnel the traffic to one campaign, so we did not capture the data we wanted. A side note, campaigns can have one keyword and still not get many impressions because other campaigns are picking it up, and I do not think that is a bad thing.

Andrew Bailiff: That is my point on the total number of ad placements. If it is all going to funnel into one, I think you are just segmenting the audience, segmenting the clicks and the spend, not actually reaching different people. If you just maximized and fully budgeted one, all that spend would come through your keyword campaigns, and we are not getting delivery on this other one because we are already targeting the placements everywhere else.

Mike Danford: I see where you are going, and the piece I am referring to is when you negate. If you have a single product attribute campaign going after a product page and rest of search, and you negate an ASIN, you are going to lose all of the search traffic it is following around. If an ASIN target is not doing well on its PDP, we want to negate or pause that target and only stop that, we do not want to impact the search that is following around. That is the main purpose of the segmentation.

Stephen Noch: Here is where I got confused. You did a bulk application of negative product targeting, but only picked products that had zero sales, and then the next day the top of search and rest of search sales plummeted across that campaign, even though what you negated had zero sales. If those product targets had top of search sales, it would have been attributed to the product target, right?

Mike Danford: I do not think so, and I think there is a key point, that is an automatic campaign. I am talking about negating in an automatic campaign as opposed to a manual product targeting campaign.

Stephen Noch: That is the source of the confusion. So you went through the search term results of the auto campaign and negated a bunch of those ASINs that were appearing, and then there were other keywords that dropped.

Mike Danford: Yes. When you negate an ASIN that is showing poor performance in the search term report of an auto campaign, you are going to impact the SERP that is following around it as well.

Andrew Bailiff: My thought on that is, why are we leaving so much spend in autos anyway? I am just harvesting out of my autos, I am not trying to keep those running, so I can control it much more easily and forget about this nuanced conversation.

Mike Danford: Andrew, what are your CPCs for your largest catalog, do you have cheap clicks compared to your other accounts, and where do those cheap clicks come from if you are not doing the auto and long-term harvesting?

Andrew Bailiff: Off the top of my head I do not know exactly.

Mike Danford: I suspect it is ASIN or PDP. With large catalogs, some have high AOV and shoppers click a bunch of PDPs and you get really cheap clicks, it compounds but the conversion is longer, could be 30 or 40 clicks before you get it. So you get a lot of thin clicks across a lot of products, but you are pulling those products in, filling up your DSP and PDPs, and building awareness elsewhere. That is part of why we keep some autos going for those cheap clicks, it works. We also have a ton of seasonality, a lot of our brands have six or seven tentpoles a year, and the semantics change so much from one year to the next for a holiday, so we have to have the autos on because they help us pick up on the semantics. We have an RPA for search query performance, but you cannot go through thousands of products' search queries, and the brand-level report only shows the top couple thousand, so the autos help us catch that over time.

Stephen Noch: I do see an interesting point. Product targeting is essentially targeting a basket of keywords with phrase match, whatever keywords that product is indexed for, you can appear on those search pages. So if you negate a competitor's ASIN because you do not want to conquest, but that ASIN was indexed for all these keywords, you would suddenly not appear on any of those search terms, you would negate all those keywords with that product target, which is very interesting.

Andrew Bailiff: But if I have an auto campaign for one product and negate all my brand ASINs, all those brand ASINs index for non-brand keywords, but my auto campaign still shows up on those. So negating an ASIN does not seem to negate all the keywords indexed for that ASIN on both ends. Negating competitor ASINs, now I am losing all my sales on those keywords, but it does not work the same when I negate my branded ASINs.

Mike Danford: I would be curious, let's confirm this, I was trying to confirm it before we came on. I imagine there is another ASIN or keyword in that campaign, outside your branded ASINs, that is triggering. I do not know if you are fully negating every keyword your product triggers for, but if there is another product or trigger you did not negate, a competitor ASIN it follows around, you are going to get that search term pop up.

Stephen Noch: Andrew sent me a message before we hit record saying this is about to get so confusing.

Mike Danford: The takeaway I have is, we are trying to clean up as much data as possible from Amazon because it is so commingled and changes over time. Campaigns get deprecated, targeting comes out, and the platforms do not tell you when you will start showing up for SERP in your product campaigns and vice versa. When we do the segmentation out of the gate, it helps us see that happening and keep the data clean. Managing larger campaigns, we do not have the luxury of single-keyword, single-product campaigns, so we need as much clean segmentation as possible so when we see something we can react. In that example, we negated our underperforming ASINs and it had a ripple effect, because that campaign did not have the segmentation with top of search and product placements split. If it was segmented correctly, which is what we did after the test, now when we negate an ASIN, we are only negating it on that product detail page, not on the SERP.

Stephen Noch: Here is my concluding thought. I do not negate items with one click and no sales, because if you do that a lot you will lose sales. You have a bunch of stuff with one click and one sale, and in these massive catalogs with tens of thousands of low-data queries, you will have stuff that gets a sale every once in a while and everything else with one click and no sales. Even though it looks like you will not lose sales, all those other items with one click were next in line to get that second click and the sale. So you basically just lost a ton of traffic. There is no individual search term being the main driver, it is a bunch of low-data stuff averaging out to a certain amount of sales, and when you negate 80 percent of the traffic that did not have sales, that is all the stuff that was next in line to convert.

Mike Danford: The response is, there are still hundreds or thousands of other keywords and targets in that campaign triggering traffic and sales, so in theory it should hold. But that is the point, you actually did negate those by negating the ASIN, which is Andrew's comment, if I negate all my branded ASINs I lose all my indexing. That is where we are, trying to figure out if that is really what is going on.

Andrew Bailiff: I think we have to go back into the lab and do a little more investigation. You came on to change my mind, you have not changed it yet, but you have changed the fact that I am going to do more investigation this week and set up some different campaigns to test. That is progress, and it has been an insightful discussion, people got to hear some unfiltered conversation between PPC people.

Mike Danford: I am looking forward to what you find out, that is why I brought it. Are you seeing this, is it campaign type, age of campaign, number of products, number of keywords, level of spend.

Stephen Noch: Mike, you are an AdLabs user, would you give a quick thumbs up or down as an agency using it?

Mike Danford: The best feature for us, with large campaign sets and hundreds of thousands of search terms, is the non-paginated search terms, thank you for that. We have crashed it loading search terms, it seems like around a hundred thousand it breaks, but aside from that the response has been great. I do not use it as much on a regular basis, but the team has been very positive on it, and it manages our larger catalog.

Stephen Noch: If people want to get in touch with you, what is the best way?

Mike Danford: Adverio.io, we will have a page for you there, with a toolkit of tools I built that we use for our larger brands and that help other sellers.

Andrew Bailiff: Thanks again, Mike, for coming on, and for everybody tuning in, come back next week for another episode.