Amazon PPC Best Practices: Summary of A Year Long Experiment

2019-01-16

In 2018, SellerMotor invested over $20M on Amazon ads and collected over 50M rows of performance data. In this article, we are going to share some major trends we found from this year-long experiment.

 

As an Amazon seller, you might wonder about the best practices on Sponsored Ads, including:

  • 1、How to determine budget allocation between Manual Campaign and Auto Campaign
  • 2、Which targeting method is more effective, Keyword Targeting or Product Attribute Targeting?
  • 3、Which set of keywords to focus on, broad or long-tailed ones?
  • 4、Broad Match, Phrase March or Exact Match?

We hope to answer these questions in this article.


1. Traffic volume: Black Friday the highest, Prime Day the second and Christmas is mixed

 

What is the most important thing about Advertising? Traffic volume.

We analyzed the monthly advertising impressions and clicks on Prime Day, Black Friday and Christmas in 2018. As we can see clearly from the chart above, the traffic reached its peak of 2018 during the period between Black Friday and Cyber Monday. The daily average traffic during this period is almost twice of that of Prime Day.

It is noteworthy that both impressions and clicks experienced a steep decline after Cyber Monday. Traffic volume bounced back from 33.4% to 84% towards the end of December. Therefore, when planning for 2019 PPC strategy, sellers should plan for at least 2-3X PPC budget and product inventory in advance for July, November and December.

2. Orders Peak on Tue, Weekends lower than Weekdays 

 





Besides impressions and clicks, we also analyzed ad attributed orders and total orders in 2018. We find out that the proportion of ad sales in total sales ranges from 10% to 85%. Certain product categories have relatively low organic traffic and therefore high dependence on ad attributed orders.

 

As for specific days of week, we can observe that weekend sales are in general lower than weekdays, while Tuesday orders are the highest on average. This could be counter-intuitive to some. While doing inter-week budget planning, you can make adjustments accordingly.

 

3. When done right, Manual Campaign is far more effective than Auto Campaign 






Compared to auto campaigns, manual campaign have on average much higher click-through-rate (CTR) and conversion rate(CVR). Although the Cost-per-Click (CPC) is also higher, manual campaign has a lower ACoS and is much more conversion effective.


This is because when equipped with thoroughly tested keywords (as we do with SellerMotors SmartPPC algorithm), manual campaigns have much greater focus and precision when it comes to targeting. We recommend that sellers use auto campaigns to draw sizable traffic and test for keywords at the beginning, but should gradually move effective keywords and allocate certain budget towards manual campaigns over time.



4. Product Attribute Targeting is Necessary


Amazon launched Product Attributes Targeting (PAT) in Beta in early 2018 and launched it to everyone in Q4 2018. Some sellers were skeptical about its efficiency when added on top of manual campaigns based on Keyword Targeting. We tested it on over 500 product listings across 8 major categories and here’s the result.


According to our research, PAT is necessary. From the charts below we can clearly see that compared to Keyword Targeting, PAT generally has higher CTR, lower conversion rate, much lower CPC and lower ACoS. When budgeted the right amount with thoroughly tested targets, PAT can increase overall ROI of a PPC campaign.

 




 

 As we know, Product Attribute Targeting Ads are shown below product details in a listing page. The majority of traffic for these ads comes from buyers of related products. Therefore when the right product target is picked, the traffic can be highly precise. It is highly possible for your potential buyers to click your ad when viewing a competitors' product, when your product is cheaper, has a better review, or has a more attractive image.


However, the cost-effective merit of the PAT needs to be revisited after a while. As we can see from the chart above, traffic cost(CPC) becomes higher over time, but the relationship between the cost of keyword targeting and PAT remains unclear. If the gap closes over time, we then need to re-evaluate the effectiveness and budget allocation best practice regarding PAT.



5. Long-tailed Keyword: Treasure or Trash?


Many articles would recommend Long-tailed Words for Amazon ads. Are Long-tailed Words really effective? How “long” makes a keyword most effective? To answer these questions, we divided over 1.5M keywords into 14 groups based on their length and analyze their performance from different aspects.


We immediately found that single-worded keywords are rare and ineffective due to the fact that they are too broad and for the most time way too expensive. Therefore, we exclude single-worded keywords from further discussion.

 

 



From the charts above we can see that traffic volume (impressions, clicks, cost, attributed sales, etc.) peaks around keyword length of 2-4 and drops exponentially when keyword length continues to increase. It is clear that most buyers use keywords with 2-4 words in length. Few sellers would bother to even target keywords exceeding 6 words.


However, CTR and CVR peaked at keyword length=10. It is clear that as the length of keywords increase, targeting would be exponentially more precise, and thus CTR and CVR will increase as ACoS decreases. Therefore, extra-long keywords have rare but solid use cases.

 

In conclusion, we should balance our goal of traffic size and conversion when optimizing our ads. Long-tailed word can generate high conversion rate but the traffic it brings is limited. Investing too much budget and time on Long-tailed keywords may not be worthwhile. It is critical for PPC practitioners to have a solid estimate of the traffic size a long tailed keyword so that she/he will know when to turn to broader keywords as a best practice.

 

6. Use Broad Match for Keyword Testing and Exact Match