« Blog Home

Four types of signals that can personalize your dynamic ads (beyond dynamic retargeting)

August 18, 2016

From contextual information to first-party data, there are lots of ways dynamic ads can be personalized and made more relevant and effective.

What's the best way to personalize a dynamic ad? By a person's gender? By their local weather? By how many children they have? By the websites they read?

Overwhelmingly, for the last few years, the answer has been "by the product(s) a customer looked at on my website," otherwise known as dynamic remarketing (AKA product retargeting, personalized retargeting, etc). But despite the huge growth and popularity of dynamic product retargeting, dynamic ads can be personalized based on a wide array of signals about the customer.

What is a "signal?"

Out of all the possible messages an advertiser could deliver to a customer, signals are bits of information about a customer (or group of customers) that help an advertiser (or a dynamic ad unit) select which message that will have the greatest possible relevance and/or appeal.

A signal could be something going on around a customer (e.g., weather, a concert), something a customer did (e.g., visited a store, subscribed to a newsletter), or a group a customer belongs to (e.g., retirees, parents), just to name a few examples.

To better understand the pros and cons of various types of signals, and when they might be employed most effectively, we'll group signals into four categories:

  1. Contextual — something going on around the customer, from which we can make inferences
  2. Demographic — some basic attribute that describes the customer, such as age or gender
  3. Behavioral — something the customer did
  4. First-party — information the advertiser keeps about the customer


Contextual signals

Contextual signals relate to the world around the customer and what customers are currently experiencing. Using real-time contextual cues, we can make inferences about a customer.

A customer is… in a city with hot weather in the forecast therefore we might infer that he or she… will be more likely to purchase warm-weather clothing.
reading an article at may be interested in heartburn medications.
in a ZIP code with a nearby retail storefront might find it useful to see the contact info for that storefront.


  • Wide audience reach and available impression volume
  • Contextual data can often be accessed at no added cost
  • Can be used in both prospecting and retargeting campaigns
  • Limited risk of violating a customer's privacy


  • Inferring contextual relevance is imprecise. Advertisers must show ads to large swaths of uninterested customers in order to reach the customers who are interested.


Demographic signals

Demographic signals describe a customer's basic traits, such as gender, age, and marital status.

One could argue that demographic signals don't belong in their own category, because they're almost always aggregated from the other three categories (contextual, behavioral, first-party), but for the purposes of this post, it's helpful to consider demographic signals separately.

A customer is… female, 15 therefore we might infer that they… will be more likely to purchase ladies tops.
male, 40, with an income over $75,000 a year will be more likely to purchase men's suits.
a parent of two children might be interested in buying a bunkbed.


  • Demographic signals are widely available from multiple sources
  • Wide audience reach and large available impression volume


  • An imprecise way to personalize ads. Not all 15 year old females like the same things.
  • While useful for advertisers with broad appeal (retail, automotive, etc), not always useful for advertiser with niche products that don't map easily to demographic buckets
  • The source and accuracy of demographic data can be opaque


Behavioral signals

Behavioral signals describe something a customer did. Fewer inferences need to be made, so advertisers can draw far more precise conclusions about which products, messages, and offers ought to appeal to which customers.

A customer… looked at athletic shoes on a retail website therefore we might infer that he or she… is shopping for athletic shoes.
bought athletic shoes on a retail website might be interested in buying gym socks.
performed a web search for 'cabo san lucas flights' might be interested in Cabo San Lucas travel offers.


  • If captured correctly, behavioral signals are very precise
  • Savvy advertisers can maintain very tight control over what events are tracked and how that information flows through an ad tech stack


  • Often only available in remarketing campaigns
  • Reach is capped to customers that actually take the actions being tracked (e.g., product pageviews)
  • Can be resource-intensive to set up, requiring investments in data feeds, tag containers, event trackers, DMPs, and so forth


First-party signals

First-party signals are based on information that a business has stored (perhaps in a CRM system or DMP) about a customer. This might include purchase history, opt-in preferences, and customer service interactions.

Businesses have been using first-party signals (AKA "first party data") to target and personalize marketing for a long time. When a local merchant makes sure to stock your favorite brand of bottled water, that's an example of actionable first party data. What's relatively new and still in its infancy is to use an advertiser's trove of first party data to personalize and fine-tune digital display ad creative.

A customer… purchases Nike shoes about once a year, and last made a purchase 11 months ago therefore we might infer that he or she… may be interested in current Nike inventory.
opted in to receive a retailer's "What's New" newsletter might be interested in new products and offers.
has a 6 year old daughter. might be interested in girls' clothes and children's toys.


  • Hands-down, the most precise, reliable type of signal
  • First-party knowledge of a customer builds over time, whereas many other signals are "point in time" data that are unable to shed light on or draw from past events or circumstances


  • First-party signals are the resource-intensive to gather, organize, store, and make actionable
  • Reach is capped to identifiable customers and prospects already captured in an advertiser’s CRM system or other databases
  • First-party signals have the highest potential to be misused in ways that violate customers’ privacy, therefore must be used with strict oversight


Broaden your horizons

Hopefully you've been convinced that dynamic ads can mean much more than just retargeted product listings.

The next step is to determine which signals are truly relevant for your business and can be turned into messaging that will make your dynamic ad creative more relevant and compelling to your customer base. Setting up more bespoke forms of ad personalization will likely require some time and investment, but it will pay off in the end when your competitors haven't progressed beyond basic "look at Product X, retarget Product X" dynamic ads.

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