Real Time Data Integration

Real-Time Data Integration

What is data integration?
The majority of industry media currency studies use the best and most appropriate research methodologies to accurately measure media audiences. As such, independent surveys result in independent or siloed analysis.  Data integration techniques enable the statistically joining of such surveys for the purpose of cross-media analysis as well as the analysis across media studies and consumer research studies.

Depending on which studies you want to work with, Data Integrations may be obtained from data suppliers themselves as Study Fusions, or from Telmar as a Custom Data Integration using Telmar’s MultiBasing technique.

MultiBasing is a modeling based technique that performs on-the-fly integration of the results from multiple surveys to perform cross-media analysis using any brand or product target against any or all of the media surveys which are integrated (cinema, magazines, newspapers, outdoor, radio, television, and web). Below-the-line media such as in-store advertising and direct mail can be included where suitable audience research is available.

Telmar invented their own data integration methodology in 2001 and it has allowed clients to:

  • Crosstab client proprietary surveys against industry media currency surveys
  • Load customer segmentation surveys for use with media currency surveys for Crosstab and planning
  • Create attitudinally based clusters on the industry product survey and analyse these against the TV audience currency survey for an added value TV buy
  • Media plan product or brand data using multiple media industry currency surveys for achieving a total communication plan

Integrating newly released audience data (i.e. updating) is straightforward – the main task is to code the ‘links’ into the dataset. There is no new ‘fused’ database to create, thus saving time and money.

What is Real-Time Integration?

Real-Time Integration is Telmar’s data integration method. It is similar to fusion in that it allows clients to cross-analyze surveys for the purpose of multimedia planning or multi survey analysis. However, it is clever in that it calculates (integrates) on the fly at run time using chaid analysis to choose the best linkage variables for the target audience(s).

When the surveys are combined, the data appears in CrossTab and MediaPlanner in the familiar codebook and input and output formats, so no software training is required.


Telmar does not create one dataset out of the various surveys, the software integrates the data and leaves the surveys intact. This means the integrity of each individual dataset is not compromised and there is no physical merging or fusing of the data.

First, looking at CrossTab in the example shown here, you will see that you can simply input a product or brand target as a column and then you can bring in your media from the appropriate industry currency surveys as rows.

The screen below shows the multi survey data in the MediaPlanner+ planning screen for combined reach and frequency.



Real-Time integration works in a similar way to fusion in that it uses common questions and demographics (linkage cells) to integrate surveys. It does this by using one survey (normally the brand survey) as the hub survey and the media currency surveys as donors. At the setup stage, all the appropriate common questions are used, but only the most relevant (identified by an on the fly chaid analysis) are used for linking at run time

An example: one survey – the ‘hub’ study – collects data on product usage, attitudinal and lifestyle characteristics, and other interests and behaviors including basic media consumption information, e.g.: “Did you listen to commercial radio in the last 7 days?”.

A second survey – the ‘donor’ survey – might be a media industry standard such as RAJAR, BARB or NRS. This media survey has wider and more complete (and, of course, industry currency) media consumption data in addition to demographic variables, and usually some lifestyle and psychographic questions.

Utilizing a combination of demographic, brand, and attitude information (questions on both surveys), Realtime Integration matches groups of respondents using all these variables and creates sensible and realistic links between respondents on the different surveys. Duplication between individual media vehicles (TV programmes and magazines for example) is accounted for, ensuring accurate reach and frequency results for integrated media plans.

*Real Time Integration and Multibasing are Trademarks of Telmar Group Inc. Copyright Telmar Group Inc., 2008

Excerpts from Erwin Ephron’s Article, The CPM Below- 2005

User Targeting Can Discover TV Program Value

Reach, Frequency, CPM are the common measures of TV planning.
Of the three, CPM is the most widely watched because it has the short-hand advantage of bundling cost and value into a single number.

But our focus on CPM has a price. Few planners believe the lowest CPM buy is best even when there is no sacrifice in Reach or Frequency. And yet few planners are able to fully justify paying more money for the same number of exposures. Think how much more rational TV buying would seem if there was tangible proof of the unique value of programs hidden right in the data we use?

There may be. It is an offshoot of “demo-profiling” with Telmar’s MultiBasing model and that is where we begin.


Demo-profiling is the cool way to target television. Since Nielsen collects very little user data, the practice is to take the demographic user profile of a brand from MRI (or the client’s own research) and assign those values to the demographic profile of a TV program.

The resulting weighted user-index of the demos becomes the program’s user index: A single number showing how well the program’s audience exploits the demographic profile of the product’s users (Table A). We call this “Demo-profiling.”

It’s much like seeing how the suit fits before you buy it.

Here program A indexes at 112 against brand A users (who tend to be younger) based on its demographic composition. If the Nielsen 18+ rating is 2.0, the profiled user rating would be 2.2 (2.0 x 1.12 = 2.2).

Brand A, Program A
Demo profile User Index (%) X TV Comp = Wtd. Comp
18-34 140   40   56
35-54 100   30   30
55+    85   30   26
Total 100  100 112



If we also have TV data from the user survey, as we do with MRI, we can do something even better. We can compare the MRI Adult rating of the program, to its user demo-profiled rating, to its user rating taken directly from the MRI study.

In the case of Program A, the MRI adult rating is 3.0. The user demo-profiled MRI rating (which is as we would do it for Nielsen) is 3.3. The MRI actual user rating is 5.0.

The question is “Why is there a large difference (3.3 vs. 5.0) between the user demo-profiled and the user tabbed ratings?”   When we think it though the answer is obvious. Demography is not usage. Although 18-to-34 year olds are more likely to use the brand, every 18-to-34 year-old does not have the same probability of using it. And different 18-to-34 year olds may be attracted by different TV programs.

Profiling Vs Tabbing, Program A

Rating Index
All 18+  3.0%  100
User Demo-Profiled  3.3% 110
User Tabbed 5.0%  167

A simple example. Upper income men 55+ index high as golfers.

However upper income men 55+, who watch PGA Golf on TV, should index much higher than the demo.

This difference between the program’s demo index and the program’s user index can be thought of as that program’s added value – its contribution in attracting users.

The next step is to apply this “program added value” to the Nielsen profiled rating. We recall that Program A’s Nielsen Adult rating of a 2.0 is increased to a user rating of a 2.2 because demo profiling shows a user index of 112.

We also recall that program A’s MRI profiled user rating is a 3.3 but its MRI actual user rating is a 5.0 (Table B). The difference is 51% (5.0/3.3 = 1.51).

This bonus of 51% is the greater user selectivity of Program A, which is used to adjust the Nielsen profiled rating from a 2.2 to a 3.3 (Table C).


Adjusting Nielsen

Rating Index
All 18+ 2.0% 100
User Demo Profiled 2.2% 112
Profiled + Program Value 3.3% 166
Program Added Value 1.1%   51

The 1.1 rating-point difference (+51%) is the program’s “AddedValue.” Its unique contribution to attracting users of the product that is not reflected in its simple demo profile (or in its demo CPM).1

Now on to the payoff question. Is this true? Do higher CPM programs attract more product users than their Nielsen demographics indicate? The analysis is still in progress, but the easy answer is “sometimes.”

MRI Index Nielsen Index
All 18+ 5.0% 100  3.0% 100
User Demo Profiled 5.7% 114 3.0% 99
User tabbed  7.7% 154   4.4% 146
Program added value 2.0% 40 1.4% 47

A real example of program added value uses “Attended 2+ movies in the past month” as the target and “Friends” as the program. It’s apt because Movie companies pay a premium for programs like Friends.

1 These extra users significantly lower the user CPM for the program. For example, if the program attracts 800,000 demo-profiled users at a cost of $100,000, the CPM is $125. Adjusting for tabulated users increases the user audience to 1,200,000, lowering the CPM to $83.

Get a Demo

Learn more about Real Time Data Integration

Telmar OOH (Out of Home)

Announcing the release of our new Out Of Home for the US market using the latest Geopath data.

Get to know our new product