‘Real Time’: A Misnomer In Display Advertising

timmayer“Data-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.

Today’s column is written by Tim Mayer, chief marketing officer at Trueffect.

“Real time” – in terms of data or targeting – is an overused phrase in display advertising.

“The rise of programmatic display lulls us into believing that we are advertising in ‘real time,’” Dax Hamman recently asserted. In reality, it is often only the bidding process that is truly “real-time” in nature. Conversely, much of the data being used to determine bids and make decisions on which creative to serve to the user is based on “non-real-time” data.

There are two primary areas where data gets stale. The first takes place in internal systems that aggregate first-party data, including customer registration and CRM data, with marketing data about what a user has been doing on the website. The second instance occurs whenCRM and website data are not being mobilized to the edge (into the user’s browser), where it can be leveraged for creative decisioning in real time.

In the first scenario, many companies separate CRM and website behavioral data into silos, where they cannot synchronize with each other to get an actionable, 360-degree profile of a user. If this information is synced, it is commonly done daily, creating a lag between the data that is used for targeting and the data in the advertising systems’ databases.

To improve marketing performance and the true “real-time” quality of the data, a company must not only create a data warehouse where it can aggregate CRM and first-party behavioral data, but it also needs to ensure the data flows from its CRM database and website in real time. This requires continuous improvement of the IT process and can sometimes take companies a year or more to evolve to this desired state.

In the second scenario, many advertisers are either not targeting with first-party data (CRM and website behavioral data), or they are targeting with third-party behavioral data. In either case, the data used for targeting is not real-time in nature. When users target with third-party data, it can become especially stale. This third-party behavioral data is collected on a third-party site, which is tracked, and a user is placed into a specific segment based on his or her visit to a specific section of a site.

A specific advertiser may purchase a segment, for example, of all users who visited the SUV section of Edmunds.com. One of the users in the purchased segment may come to an auto brand’s site and begin looking at the station wagon section after only an initial glance at SUVs. In the first-party data-technology environment, if that behavioral site data could flow through seamlessly, the user would now be targeted with station wagon ads for the specific brand. On the other hand, if only stale third-party segment data is used, the consumer will still be shown ads for SUVs they are no longer considering, causing these ads to underperform.

In the ideal state, the ad server needs to take relevant first-party data and push it into the user’s browser where it can be leveraged by the advertiser to better target the user in real time, with creative based on the segment in which the user has been placed.

In the retail industry, for instance, users could be separated into four segments:

1. Net new/unknown: The advertiser has not seen this prospect before and a CTR-based offer should be served to try to get the user to visit the site.

2. Visitors (recent or lapsed): A value-based offer, such as a discount or coupon, should be used to lure the prospect to make their first purchase, establish the relationship and create some first-party data in the process.

3. Visitors (with known product interest): Offer-based messaging, based on knowledge of location or product, should be made.

4. Existing customer (with known product interest): An upsell, cross-sell or retention message should be used.

There are visibly high benefits to ad targeting by moving in-house first-party data closer to true “real time,” and pushing that real-time data out to the edge (within a user’s browser) so the ad creative can be decisioned in real time at the time of serving. This can lead to huge campaign performance improvements, especially in the purchase funnel and high-consideration purchase industries.

The constant chatter about “real-time” bidding with the rapid rise of programmatic bidding should not cloud our understanding that the critical data for superior ad performance is still not being utilized in real time for segmenting users or creative ad decisioning.

Follow Tim Mayer (@timmayer), Trueffect (@Trueffect_Tweet) and AdExchanger (@adexchanger) on Twitter.

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!


  1. Tim,
    Nicely written article. Dax’s comment is an important reminder of the slings and arrows of outrageous terminology: people think they’re doing real time decisioning based on real time bidding. When it comes to data, I think there are some additional things to consider:

    1) Using stale data takes a toll. It’s expensive to serve ads to users already converted (a waste of ad dollars and a hit on performance goals), and it hurts the brand because it makes the advertiser look like they don’t know the users (I just bought this! Why are you trying to sell it to me again?).

    2) To me, data warehouses aren’t as agile as a DMP. I think of a data warehouse like visiting a brick and mortar video rental store – it takes time to access what you need. A DMP needs to be efficient in moving data, which makes it more like streaming a movie online.

    3) Do CRMs and DMPs need to be integrated? Absolutely. But does it require Herculean IT resources? No. If an end user can sign into their account online, a DMP can bridge offline data with online interactions pretty easily using the first party key.

    4) The real problem with stale data comes from these Rube Goldberg-esque stacks we’re creating to carry out a marketing plan. The use of multiple vendors adds unnecessary complexity, which then leads to cookie matching loss, lack of real time engagement, and technology problems as systems don’t talk to each other. Syndicating segments is nice, but it doesn’t solve for real time. And don’t be fooled – when you see “marketing stacks,” it doesn’t necessarily mean the systems talk together in an easy fashion.

    What you are probably really looking for is an integrated DMP + DSP type deal where the two are fully integrated. When this is the case brands can go beyond pushing the real-time data out to the edge (within a user’s browser), as you describe. They are actually able to support real time decisioning in other channels as well, such as email, call center and even updating the CRM to fully leverage the offline data. One important thing to note: all DMPs are not created equal. When you’re looking at DMPs it’s important to ask the question: What is your real time capability across all channels?

    And finally, there’s using third party data for behavioral targeting. Having that update in real time in a cost efficient manner is a nut the industry has yet to crack, much like syndication in “real time.” But given the speed of change, it likely won’t be long.

  2. Ian
    Thanks for the kind comments and the comprehensive response.

    We see #1 a lot and in fact we see advertisers retargeting users with a better offer than they converted with or sometimes an offer they are not qualified for as an existing customer. People often go back and return or unbook and then rebuy at the lower rate which costs even more than just the wasted impressions

    On #2 and #3: there are a lot of ways to skin the cat but in the end it is getting to real time that matters. Some of this depends on the client needs and existing architecture.

    On #4: agreed that people are looking for more simplicity and less complexity. This complexity is leading to data lag but also to a massive amount of non-working media cost in our ecosystem.