« Blog Home

It's time to get more creative with your product feed

August 23, 2016

Want your dynamic ads to do a better job of matching the right products to the right viewers? Start including more than just basic data in your product feed.

NOTE: This blog post is a helpful primer on the types of signals and cues that can help dynamic ads make intelligent decisions about which messages to show to which viewers.

Dynamic ads have the potential to be a lot smarter than what you see in typical dynamic remarketing ads where the customer is shown the same products they just looked at.

But if SKU-level data is all that's in your product feed, then your dynamic ads will never be able to take a more holistic approach to connecting the right products with the right customers.

In this post, we'll look at a hypothetical sports retailer's product feed and dynamic ad creative, and how they can work together to power smarter ad creative.

Dynamic ads need run-time data

Let's imagine a dynamic sporting goods ad that has the following viewer data available for every ad impression (i.e., "at run time"):

  • SKUs viewed (but not purchased)
  • SKUs purchased
  • Gender
  • Approximate age
  • Approximate income
  • Publisher page URL (the URL of the web page where the ad is being seen)

This kind of data is available using various ad tech stacks that may combine DSPs, adservers, and DMPs, but the purpose of this post isn't to be a technical how-to, so we'll just assume all that data is available in real time every time an ad impression loads.

Dynamic ads need rich product data

Let's also imagine a sporting goods product feed that includes much richer data than basic SKU-level product details, revealing the connections between various products and between products and viewers.

  <title>Red running shoes</title>
  <complementaryCategories>Athletic Socks,Running Shorts</complementaryCategories>
  <category>Running Shoes</category>

Connecting products to customers

Let's now run through a few hypothetical ad impressions and see how richer data might help our dynamic ad choose the most compelling products.

Example 1 – Retargeting a customer post-conversion

In this example we look at a customer who has purchased an item. There may an opportunity to bring that customer back to make a follow-up purchase. Product feed data can help the dynamic retargeting creative select a complementary product that's well-suited to the customer.

Viewer attributes::

  • SKUs viewed: none
  • SKUs purchased: 77403270
  • Gender: male
  • Approximate age: 40
  • Approximate income: $125,000
  • Publisher page URL: N/A

Now let's look at the product attributes of what this customer purchased:

  <title>Wireless earbud headphones</title>
  <complementaryCategories>Smartphone Cases,Arm Bands</complementaryCategories>


When an ad impression runs, the dynamic ad can use a few simple programmatic rules to find a product that fits what is known about the customer and is likely to be compelling:

  • <category> NOT EQUAL TO 'Headphones' (the customer just bought headphones)
  • <incomeMin> LESS THAN 125000
  • <complementaryCategory> EQUAL TO 'Smartphone Cases' OR 'Arm Bands'
  • <gender> EQUAL TO 'male' OR 'all'
  • <age> range INCLUDES age 40
  • SORT BY <popularityRank>

With all of these attributes and rules in hand, the ad can perform a simple lookup to find, say, a particular Smartphone case that fits all the criteria for the customer:

  <title>Black smartphone case</title>
  <complementaryCategories>Arm Bands,Headphones</complementaryCategories>
  <category>Smartphone Cases</category>

Example 2 – Prospecting on fitness sites

In this example, our hypothetical sporting goods advertiser is running dynamic prospecting creative across a number of fitness sites.

We can assume that many of the fitness site visitors are in-market for sporting goods, but there is still the question of how the dynamic ads will decide what products to show.

Luckily, our ad has access to quite a bit of data related to each ad impression and ad viewer.

Viewer attributes:

  • SKUs viewed: none
  • SKUs purchased: none
  • Gender: female
  • Approximate age: 50
  • Approximate income: $85,000
  • Publisher page URL:

The ad would first use the publisher page URL to perform a keyword search against the product feed and perhaps find matches on products associated with the keyword 'swimming'. This might include bathing suits, goggles, and swim caps.

The ad then feeds this information through its decisioning rules:

  • <keyword> FOUND IN 'swimming-is-the-best-form-of-exercise'
  • <incomeMin> LESS THAN 85000
  • <gender> EQUAL TO 'female' OR 'all'
  • <age> range INCLUDES 50
  • SORT BY <popularityRank>

After the ad performs a lookup according to the criteria and rules, it might find a swim cap that is a good match for the viewer:

  <title>Yellow swim cap</title>
  <complementaryCategories>Swimsuits,Swim Goggles</complementaryCategories>
  <category>Swim Caps</category>

Use your imagination

There is no one set of product attributes that all product feeds should contain. Find out what run-time data is available to your dynamic ads (see this post for some ideas) and then think of what kinds of product attributes you would need to actually make use of that run-time data.

For example, if your dynamic ads have insight into a viewer’s ZIP code, then maybe you should include in-store product availability as a product feed field, organized by ZIP code.

And note that these concepts do not just apply to retailers. They can be applied to any industry vertical where dynamic ad creative is being used.

Contact us to learn more about the Canned Banners Dynamic Ads platform.