At m19, we have witnessed several missteps sellers might make when starting with PPC on Amazon. This article shares six of the most common practices to avoid in order to rapidly improve your advertising campaigns. Those practices are as follows:

  1. Blocking wrong keywords
  2. Abusing phrase and broad match
  3. Stopping costly campaigns
  4. Combining bids adjustments
  5. Trying to get the first position with insane bids
  6. Sponsoring the same ASIN in multiples campaigns

Amazon PPC reminders

Before diving into the subject's core, we wanted to start with a couple of refreshers on some fundamental Amazon PPC concepts.

Second price auction
The first point sellers ought to understand about Amazon PCC is its auction mechanism. Amazon uses a second price auction where the highest bid wins but only pays the second-highest bid. Amazon operates this sophisticated auction as sellers place a bid for displaying ads. For example, if D’s bid is the highest and C’s is the second-highest, then D wins the auction but only pays a CPC of $4 (Figure 1).

Figure 1: second-price auction

Search term and keyword match type
A second dominant concept is match type: the different ways sellers' keywords will capture search terms. When users surf on Amazon, they enter a search term describing their needs. Keywords will then grab the search term according to different matching types. On Amazon, there are three matching types:

  1. Exact match type (the most precise)
  2. Phrase match type
  3. Broad match type (the widest)

Phrase matching type is more precise but encompasses less search than broad matching, whereas it will be less precise but retain more traffic than exact matching. For example, the user enters "soccer shoes" as the search term. Exact matching looks for the precise keywords and matches "soccer shoes" as exact matching keywords (Figure 2). Phrase matching will allow ads to be visible when the keyword contains "soccer shoes" and something else, for example "Adult soccer shoes". Finally, with broad match "soccer" and "shoes" keywords will be eligible to appear separately, such as "Soccer shiny shoes". Overall, the user's search term will be captured with either exact, phrase or broad matching.

Figure 2: Soccer shoes three match types

Negating a keyword
The third concept is negating or blocking keywords. Negating is also linked to match type, but instead of capturing the search term, we stop it. There are only two ways to negate a keyword, either with a negative exact or a negative phrase. Parallel to the matching described before, the negative exact is included in the negative phrase. Following the previous example on soccer shoes, using negative exact will exactly block "soccer shoes" keywords, whereas negative phrases will block keywords that contain "soccer shoes" together. We will later explain how to strategically use the different matching types to capture and stop search terms.

Amazon arbitration
The last fundamental is Amazon arbitration. Earlier, we mentioned that Amazon uses a second-price auction to allocate winning CPC bids, but the reality is slightly more complex. The CPC bid is not the only factor Amazon uses to administer which product to display in the sponsored position. Another factor is predicted conversion rate. The expected conversion rate is an indicator of relevancy and is also used for organic ranking. We also believe that there are other undetermined factors that Amazon is using. Combining those factors will help Amazon predict the revenue and the value of displaying a product on a given search term.

Moreover, Advertising will provide Amazon with two revenue streams: sponsored clicks and sales commissions. Each time a shopper clicks on a sponsored product, Amazon charges a CPC and every time a sale takes place, Amazon receives a commission. Therefore, Amazon is looking for the most qualitative ads/products in order to maximise revenue streams from sponsored clicks and conversions. The Amazon algorithm will enter the various eligible accounts and products in a profit-maximising competition and display the winners for a fixed search term. The competition will arbitrate which account, product and campaign will be displayed.

Common practices to avoid

We now have all the necessary tools to explore six common practices sellers must avoid.

 1. Blacklisting bad keywords
The first is blocking relevant "bad" keywords. What defines a keyword as bad is subjective and might depend on various criteria such as an increase in advertising cost of sales (ACoS). Keep in mind that we are not addressing irrelevant keywords which Amazon will automatically block. Our recommendation is to avoid using blacklisting and rather considerably lower CPC bids for underperforming keywords. We invite you to conduct the following experiment to prove that blacklisting might not be the most profitable solution. Start by setting a threshold for "bad" keywords like "all keywords with an ACoS twice as expensive as our usual levels". Next, choose one or more past campaigns to investigate the wrong keywords. Finally, divide the data into two different periods—first, find keywords with ACoS above the threshold and second, compare performance metrics with and without those keywords.

We also experimented over two periods of five weeks (figure 3). On the plot, we flagged keywords in the first period and graphed advertising sales with and without them in the second period. The table computed three metrics: ads spend, ads sales, and ACoS. ACoS are almost identical, but the ads sales were slightly higher (8.5%) when keeping bad keywords. Therefore, for nearly the same ACoS, keeping bad keywords might increase sales. The key is to manage wrong keywords by reducing bids. Overall, all keywords have an optimal CPC at which they are profitable.

Figure 3: impact of blacklisting underperforming keywords

 2. Abusing phrase and broad match
The second common practice is overusing phrase and broad matches. If you use phrase or broad for every performing keyword, you are not narrowing down keywords to a single bid. Therefore, bidding against yourself on keywords matching the same user search query lets Amazon charge you with the highest bid.

Let's understand these overlapping matching types with an example. Imagine a shopper enters "Pink tall vase" as a search term, and you set "Vase" as broad, "Tall vase" as phrase and "Pink tall vase" as exact for the same ASIN. Consequently, Amazon will associate the impression and sponsor position to the highest bid between the keywords. In other words, it is a traffic of search terms addressed from different entry points; Amazon will select the most expensive entry leaving you with a higher ACoS for the same performance. So one solution is to negate the inappropriate entry points.

Moreover, to prevent this problem in the first place, we recommend avoiding using hundreds of keywords with different matching types. It is more profitable to have fewer well-configured keywords than thousands of not controlled keywords. All manually targeted keywords should be coupled with their negated version; check if an exact or phrase keyword in an ad group will be matched multiple times.

 3. Stopping costly campaigns
The third practice is stopping costly campaigns. The mistake is to focus on ACoS over one period and decide to stop expensive and underperforming campaigns. However, the issue stands with the campaign's configuration and not necessarily the campaign itself. The performance of a campaign may diminish because of variation in competition or several different factors. During that period, ACoS can reach incredibly high levels. Considering the campaign below (figure 4), the ACoS went from 30% to 160% in a week. Some sellers look at this particular week and stop the campaign. However, simply decreasing the bids is enough to return to profitability, and after the spike, the ACoS returns to normal levels. Therefore with the proper structure, any campaign can be profitable. In case of spikes in ACoS, we urge sellers to check the data's significance, find the underperforming search term and adjust the CPC.

Figure 4: evolution of ACoS and Bids for a given campaign

4. Combining bid adjustments
The fourth mistake we flagged is carelessly combining bid adjustments. In seller central, sellers can select various bidding strategies. One option is setting dynamic up and down, letting Amazon double or halve bids. Another option is to adjust bids by placements. On some positions, such as top of search or product detail pages, sellers allow Amazon to raise bids by a percentage to increase impression chances. Although top of search and dynamic biding perform well, we caution sellers to combine them as bids can be multiplied twentyfold (double bid & 900% top of search). A default bid of $2 can reach $40, and if a competitor bids incredibly high, the seller might pay slightly under $40 for a click (figure 5). All in all, while combining adjustments, make sure you are ready to pay the maximum bid.

Figure 5: bid increase when combining bid adjustments

 5. Trying to get the first position with insane bids.
The fifth mistake is bidding very high and flat to guarantee a sponsored position. A common belief is that sellers can bid high to secure a win as they will only pay the second-highest bid. This method can be hazardous and does not always pay off. We mentioned earlier that bid is not the only variable used to estimate the value of displaying a product in a sponsored position. In reality, Amazon will start by checking ads' relevancy and product ranking (i.e. the likelihood of conversion). Next, Amazon looks at predicted profitability, and the bids. Therefore, a high bid does not guarantee display if relevancy and estimated conversion are poor.

What if I bid one euro and I pay twenty cents a click. Therefore my competition is bidding very low; I can increase my bid so as to be safe and guarantee the sponsored position. However, competitors' will gradually increase their bids to catch some traffic. Therefore, as the bids surge, you will be paying higher and higher for the same traffic. Overall avoid overbidding because, in the long run, cost will increase with competition.

6. Sponsoring the same ASIN in multiple campaigns
The last mistake is sponsoring the same ASIN in multiple campaigns. The logic is similar to abusing phrase and broad matches (mistake 2). When targeting similar keywords from different campaigns, you are overbidding yourself. For example, when a $8 bid for campaign X and $6 for campaign Y occur for the same product, the $8 bids win and pay $6 (figure 6). Without campaign X, the $8 bid wins and pays $4 (i.e. $2 less), so the overlapping bids are harmful. Duplication errors occur when numerous campaigns are created; the structure becomes complicated, and sellers will lose a sense of what is in each campaign. The takeaway is to have only one campaign for a given product on a given keyword.

Figure 6: Sponsoring the same ASIN in multiple campaigns

To sum up, we presented six typical practices to avoid in Amazon PPC: blocking wrong keywords, abusing phrases and broad matches, stopping costly campaigns, combining bid adjustments, trying to get the first position with insane bids, sponsoring the same ASIN in multiple campaigns. Therefore, we urge sellers to take the necessary precaution against the six practices to optimize their PPC campaigns and save costs.

If you wish to discuss this topic with us, feel free to reach out here.

Share this post