“Real Time” in Display Advertising Doesn’t Really Mean Real Time
A lot of nonsense is spoken today regarding “real-time marketing” and “real-time bidding,” yet most of what is actually being bought isn’t really “real time” at all, and marketers need to know why that’s a problem. With the rise of the media exchanges came the separation of the media (the “where” to target) from the […]
A lot of nonsense is spoken today regarding “real-time marketing” and “real-time bidding,” yet most of what is actually being bought isn’t really “real time” at all, and marketers need to know why that’s a problem.
With the rise of the media exchanges came the separation of the media (the “where” to target) from the data (the “who” to target). The marketer was able to stop simply shouting their message at crowds, and instead became able to pick and choose the individuals they wanted to talk to. The promise was efficiencies unseen in marketing before.
There is however a problem. Whilst the media became real-time, the data did not, and the promise was broken.
The term real-time bidding, or RTB, is commonplace industry language. The term specifically refers to the media that is being purchased; it is the ad impressions themselves that are real-time. The data being used to determine which impressions to commit to, and at what price to pay, is mostly slow-time at best, and even worse, is often generically grouped based on broad segments that offer little value.
More than 50% of ad dollars spent today in RTB are for site retargeting. A visitor hits a page or site, leaves and is then retargeted. Each time that individual is seen again, a real-time auction can take place, and the impression is acquired.
Now think about a time when you have been on a brand’s website, and having looked briefly at an item, you have been followed for days, weeks or even months about that same product. For a moment in time you had an intent related to that product, but, all too often, your intent to buy wore out long before the ad budget did.
When you visited the site, a cookie was dropped on your device and you were added to a segment created by the marketer. Let’s say it classed you as a “browser.” You will remain in that segment for targeting until you complete a different action, and until that happens, the stalking will continue.
Every minute that passes since your visit, the data you left behind degrades, and with it, the marketer’s knowledge of your intent.
The wastage becomes further amplified with a prospecting, or customer acquisition campaign. Many of you likely work as marketers in industries that are very competitive, and are always asked to generate more new customers. There are data signals that can be used to suit most industries, but some are more real time than others.
Imagine I am looking to book a flight between San Francisco and Denver; my booking window is probably less than an hour for a business trip, perhaps a day or so for a vacation. During that time I will run some searches, visit some sites and perhaps look at some reviews. Within a short period of time, I will have worked out the basics of my schedule and be price shopping.
If a marketer is relying on data that is not really real time, that marketer will have missed my dollars before they knew I was in market, then, when they do start targeting, I have a negative experience with that brand because I no longer need them. One of their competitors’ benefitted instead.
3rd Party Data
A common proliferator of slow-time data is the group of companies selling cookie pools or segments of users based on broad interests. These can have a place if used broadly enough (and if the cost outweighs the data investment), but for a direct response campaign with any sort of time requirement, they are usually late to the game.
Such segments are usually built from a combination of data collected from multiple sources, analyzed in batches (often once a day) and then passed to a tool for RTB, which often takes another day before the cookies are synchronized. Unfortunately, those intender segments you have been buying are far from real time, only the impressions you bought against them were.
DSPs And DMPs
Drawn by the promise of real time, brands and agencies have brought in DSP (Demand Side Platforms) and DMPs (Data Management Platforms) to execute their spend. Typically these are disparate technologies that need to be synchronized together, and, due to their limitations, will absorb data and spit out segments from the DMP back to the DSP. This takes time, and we often see brands buying against data that runs in a midnight process for targeting the next day.
The result is that all the targeting data is slow-time, even their own on-site 1st party data, which has now being washed through several platforms and slowed down.
There are some easy solutions, and some more difficult.
Firstly, marketers must understand what they have and what they don’t have. They need to look at the ways they are buying, and follow the flow of the data themselves to see the time delays involved. At least with this knowledge, more intelligent decisions can be made around investments.
The next step is to consider the technologies and partners being used. Most of the many big brand off-the-shelf retargeting companies do not operate in a real-time data environment, and are creating the negative experiences talked about above. I’m a big believer in the concept of Programmatic Site Retargeting, which looks at individuals during the actual bid request when each impression is bought, and calculates their true value and their live intent.
The same goes for Programmatic Prospecting. Understand the buying cycle of your products and services, and then match it against the data you are processing. If the two don’t align, it’s time to make a change.
And finally, separate DSPs and DMPs can have their benefits, but also inherently break the promise of real-time marketing. Consider using the same platform for both functions and eliminate the time lag where possible.
Opinions expressed in this article are those of the guest author and not necessarily Marketing Land. Staff authors are listed here.