What Display Advertising Can Learn From Politicians
In politics, the act of drawing, reviewing and re-structuring district lines for voting purposes to ensure that like demographics are aligned is a common practice. These lines appear bizarre when viewed on a map due to the non-uniform nature of redistricting. Why don’t politicians just use ZIP codes? Politicians learned many years ago that zip […]
In politics, the act of drawing, reviewing and re-structuring district lines for voting purposes to ensure that like demographics are aligned is a common practice.
These lines appear bizarre when viewed on a map due to the non-uniform nature of redistricting. Why don’t politicians just use ZIP codes? Politicians learned many years ago that zip codes are not demographically sensitive.
While it is fair to assume that there are some similarities with you and your immediate neighbors, within a few short blocks there are often stark contradictions across virtually every demographic data point.
ZIP codes are not even real areas. They are, in fact, a series of delivery points. Even when treated as areas, they don’t have the density to relate to urban communities.
Somewhere along the evolution of display advertising the “ZIP code” in its assorted repackaging forms became the standard for location and demographic based targeting.
Geo targeting is one of the most common campaign targeting methods in running a display ad campaign. It also happens to be one of the most technically misunderstood features in our industry.
In fact, I recently surveyed more than a dozen client services and media-buying professionals on this topic. Virtually everyone who buys display ads routinely include one or more specific geographies of varied sizes for varied reasons in their ad buys. It is also true that very few of these media buyers when asked directly could tell me how their vendors pull off this technical magic or confirm with any confidence that the actual delivery occurs with any known accuracy.
In The Beginning…
The technology behind most geo-targeting deployed by ad networks, display ad servers and demand side platforms is a first generation technology based solely on a consumer’s IP address (geographic IP targeting). This is almost always deployed down to the ZIP code or town level.
With more than 4 billion IP addresses in existence, accuracy is a luxury. This is especially true in light of the fact that IP addresses are constantly being assigned, allocated, reallocated, moved and changed due to routers being moved, enterprises being assigned IP addresses, enterprises moving and networks being built or changed.
Getting anywhere near accuracy requires complex algorithms, bandwidth measurement and mapping technology. Even then, a highly-tuned delivery mechanism is necessary. Finally, each address must be periodically updated to reflect changes in the IP address information, without invading a user’s privacy.
The truth is that for most companies, the accuracy of the data used behind the scenes for geo targeting and the closely associated demographic targeting is relatively poor. First-generation geographic IP targeting is the product of a location applied to IPs.
While IP-based geo-targeting is accurate in pinpointing the geography of an IP, the limits of the data’s relation to advertiser intent are significant. The precision of IP targeting is further diminished by the discrepancy between the user’s actual location and the location of the IP.
When IP targeting was developed, there was no real consideration made for use cases. When the ad industry creates use cases, they are coerced into a one-to-one relationship with the IPs. For example each IP has one city, one ZIP code, one demographic, and one confidence level.
If you decide to dig into this with your vendors, you will often find that the underlying data is often considered proprietary, or involves many heuristics that flatten semantics, causing information loss between the data source and the use case. A common example of this is simply assigning the location to a particular IP address based on a point-in-time inspection or based on where it appears most often when pinged. This is hardly a reliable tactic.
The Evolution Of Second Generation Geographic IP Targeting
Luckily, when I dove into this subject more deeply, I learned that some of the best and brightest in our industry have been following second generation IP targeting that many describe as “Beyond Location”.
Unlike its predecessor, it is designed to satisfy real needs of brand and direct response marketers from the start.
The reality is that IPs and locations are not a fixed one-to-one relationship in the real world. Locations and IPs are in a dynamic relationship that may be deployed and redeployed arbitrarily.
The key to second-generation geographic IP targeting is technology logic that follows IP movement and deployment models over time. When followed closely predictable patterns emerge.
These patterns, while hidden, are very real, and may be profiled. The history of IP deployment is then used to predict the likelihood, or bias, that an IP qualifies some condition or property denoted as the desired target.
Companies accessing this level of detail can then place a value on the potential consumer for the specific marketing case needed to maximize engagement.
Furthermore, the evaluation can go beyond whether to engage in a particular way. This is about newer, more specific and more accurate methods for matching the message to the audience to achieve higher conversion rates.
Marrying Better Element-Level Demographic Data
With this newest evolution in geographic IP targeting, online marketers can obtain whole new levels of accuracy in element-level demographic targeting.
Finally free from the boundaries of ZIP codes; savvy marketers are free to adopt more effective geographic models liked those used by the survey industry. In these models, neighborhood differences become exposed, yielding more significant target margins.
So with the proper expression of geography available with second generation geographic IP targeting, anything related to geography may be used such as any third party or proprietary data set that has any kind of location information may be used by way of IPs.
Second Generation Geographic IP Targeting:
- Respects privacy — location is mapped to centers of streets much like Google Maps.
- Accounts for exchanges removing the last octet of IP addresses — allowing continued targeting without compromising privacy.
- Offers insights into the probability of accuracy — allowing reach to be dialed up or down in reference to accuracy. Something not available in 1st generation systems.
Specifically, a survey, like the US Census Bureau’s American Community Survey, may be related to IPs, so hundreds of demographic qualities may be targeted, allowing demographics missing from other sources to be targeted.
Marrying highly detailed, current and accurate survey data with the newest second generation geographic IP technology results in increasingly precise and accurate delivery within display campaigns.
Marketers are able to target very specific demographics across reliable geographies infinitely smaller than a ZIP code such as neighborhood blocks.
For example, as you can see in the accompanying infographic, there are 176 traditional ZIP codes in the greater New York City DMA.
Now a marketer can view the same geography as 6,494 Geo blocks or neighborhoods.
Combine this greater location accuracy with demographic data such as that obtained through the Census Bureau and you will find amazing new combinations of element-level data to reach your ideal customer.
A car manufacturer such as Ford can target economy cars to geographies with longer commute times and leases to geographies with shorter commute times. Ford may also feed in data on where Mustangs are purchased and then target those geographies.
A restaurant chain can evenly distribute display ads by IP proximity to the franchise locations and by franchise quota. This prevents them from having to define inner metro boundaries that may be gerrymandered or unfair.
High-end retail stores or Financial Services firms can target interest and divided income to locate large investment portfolios.
In politics at all levels, candidates can feed in donation addresses and target donor areas that are difficult to reach on foot. Extending reach to congressional districts, providing the ability to target businesses or people — thereby minimizing the wasted impressions blanket ZIP code/DMA targeting can cause.
Another application involves the home equity metric, giving marketers the ability to target to individuals with:
1. Good home equity — Percentage of households that have homes whose values back the median mortgage payment.
2. Good cash flow — Percentage of households that have income that can cover expenses at twice the median mortgage payment.
Any “equality” like the one mentioned above can be measured. For example, sex equality such as identifying areas where women have higher equality.
Data regarding how old the buildings are is available as well, allowing a replacement water heater company to market to areas that have a high percentage of buildings that are reaching the age in which heaters should be replaced.
Advertisers should follow the lead of politicians and take advantage of the next generation demographic targeting capabilities second generation geographic IP targeting presents. Marketers now have the freedom to target beyond previous metrics, the ability to go “more local than local” and to target rich demographic data with higher degree of accuracy, at levels that make ZIP code targeting seem like throwing darts.
Opinions expressed in this article are those of the guest author and not necessarily MarTech. Staff authors are listed here.