Amazon VCPM Sponsored Display Ads - Weird Reporting? Here's What's Happening...
Updated: 2 days ago
VCPM advertisements (offered by Sponsored Display Ads - aka SDAs) are some of my favourites. You pay per thousand impressions rather than clicks. That's not why I like them. I like them because if you get them right, they deliver impressive results, and as they often appear off-Amazon, they don't compete with current ads.
But there's a problem...
This became clear when I started seeing some strange reporting data for a client. Here's an example...
If I clicked on the TARGETING tab in Campaign Manager, which I do whenever I want to see an overview of where the ad spend is going, I saw that one VCPM campaign had 23 orders with 502% ACoS for a 30-day period!
You don't have to be a PPC specialist to know that this isn't good.
And yet, I was managing these campaigns daily. I would have noticed something like this. So, I did a deeper dive by looking at the campaign itself. The campaign metrics were 318 orders at 36.43% ACoS. Pretty good for a new campaign. But which is right 500% ACoS, or 36, 23 orders, or over 300?
How Amazon is Delivering VCPM Data
It is possible that Amazon is trying to present the data at campaign level in a better way as it wants to push this type of off-Amazon selling.
Okay, so what really is off-Amazon selling?
Well, first of all, if you go to a review site, a website, blog, or landing page where there is space for product advertisements, you will at some point see an ad for a product that you have previously viewed on Amazon, or one that Amazon thinks you will be interested in.
So, I was looking for a kayak recently, and was surprised to see one of my client's ads showing there as a "hey, you looked at this recently, don't forget to buy it."
These ads are different than regular sponsored product ads (SPAs) because SPAs appear during a search when the shopper is actively looking on Amazon for that type of product. Off-Amazon is an interruptive ad, where it tries to grab your attention and remind you to buy something at a time when you aren't shopping.
And there is an important difference here in how this data is captured by Amazon.
Viewed Conversions vs Clicked Conversions, Know the Difference?
Through cookie tracing and IP address tracking Amazon ads are being monitored off-Amazon. Also, I remember seeing a banner notice recently while working in Campaign Manager, that Amazon may decide to show some of your ads off-Amazon whether or not you have actively chosen to do so. And this is a sign to an important message - this is where Amazon sees most new growth.
So, I am looking at a kayak blog and there is an Amazon ad. How does Amazon know that I have seen the ad and not simply scrolled passed it? How does Amazon know that I have actually "viewed" the ad and it can therefore measure it as an active view? Heatmap tracing, mouse movements? Maybe. Perhaps websites need to agree to specific types of cookies to deliver that information. Here is how Amazon determines a "view"...
With SDA campaigns targeting VCPM advertising, you will likely notice that there are far more "impressions" than there are "views", so this means that most "impressions" aren't meeting the above criteria of being in view for more than a second.
What is important is that we understand the data that it is given to us so that we can make meaningful decisions, and some of how this data is delivered is shrouded in mystery because Amazon isn't making it clear as to how it is differentiating things when, for example, it combines "viewable impressions and clicks" as a percentage together. See below:
Important - If an ad is shown off-Amazon and the same person returns to Amazon to buy your product (but hasn't clicked the ad on the blog post, for example), then Amazon measures this as a "Viewed Conversion" and not a "Clicked Conversion". So, a "Viewed Conversion" is a sale that is later attributed to the VCPM ad but where the ad itself wasn't clicked. Therefore, Amazon is claiming a sale against this ad based on unknown criteria, and possibly when a shopper hasn't even noticed the ad.
Mmmm. Interesting. Whether I am at ease with this really depends on how the shopper returned to buy a product. If they had saved it in their favourites, or went through their history to find it, and happened to be reminded by the ad then okay. But what if they had originally clicked a regular SPA ad previously and didn't acknowledge seeing the ad? Well, Amazon is making some informed decision and then if certain criteria are met, the 7-day, and 14-day sales attribution windows are used to align the sale to an ad.
For SPAs, we have a 7-day window, and for the rest we have 14-days. An attribution window is not a delay in delivering data, it is Amazon including purchases within the relevant timeframe and attributing that deferred sale to your original ad click. And, I am noticing longer attribution windows, sometimes up to 21-days. So, which ad gets the click [cost], and which gets the sale? Are we potentially reducing bids on SPAs because of low CVR when people are returning outside the 7-day window after seeing an SDA you also paid for?
Perhaps we are paying twice for an ad, but perhaps we sometimes need to. And perhaps this has always happened. I understand that's the nature of advertising. As long as we get the sale some way, and as long as we make a profit, then it all works. But as PPC managers, or people using PPC management software, we need to factor in these attribution windows before making bid changes.
Returning to the above numbers - 23 orders at 500% ACoS - I see this data as meaning that 23 orders came from "Clicked Conversions", i.e., they clicked the ad that they saw off-Amazon and then purchased your item. Whereas the 300+ sales at 36% ACoS were later attributed to the VCPM campaign as "Viewed Conversions". In this example, around 300 sales were attributed to VCPM ads that were not clicked. That's quite a difference. And it begs the question, how do we assess the success of a VCPM ad? And are we making accurate bid adjustments to other campaigns when a potential 21-day window can attribute a sale against a VCPM that may not have been actively noticed?
And of course, if you are targeting competitor ASINs with VCPM ads and a person having shopped around suddenly sees a reminder to buy your product while they are reading a product review website, well, you are in the hot seat. And it's super-hot because you are advertising in a space where your competitors are not. Most unlike regular SPA and SBA ads.
And that's why I like these ads. They usually convert well and have a low ACoS - think 10-20%. They do need a lot of attention at first because getting the bid right is quite an art!
The Future for VCPM Advertising
This feels like the dawn of something quite significant. With some of Amazon's Demand Side Platform Ads (DSP) ad types cascading to regular brand sellers, and the fact that a blog post or similar doesn't want to be crowded out by a whole host of competitor ads for the same type of product, I see winning ownership of this ad space as being vital to scaling your business.
If you want a tip right now, here's what I recommend... although these ads aren't always off-Amazon they are worth doing. Get your best keywords and type each one of them into Amazon. Target P1 to P3 ASINs that appear in the results via SDA with VCPM targeting. Do a separate campaign for each page. To target VCPM you need to select "Optimize for Reach". Remove "Similar to advertised products" which is chosen automatically. I would suggest a bid of $3 (very low bid) to $6 (reasonable bid) per thousand impressions. Then monitor this twice a week to ensure that nothing is running away with your money, and bid manage into decent results.