Early this year, you likely saw the slew of cross-media optimization studies analyzing the implications of increasing and decreasing the online share within an overall campaign spend. Without a lot of effort, you quickly grasp the motivation: These research efforts are aimed squarely at the heads of traditional media planners. One can only ask if these online/offline reach/frequency studies are just another attempt to validate the medium, or have we finally struck a nerve with marketing decision makers?
We’ve crossed a chasm. As a participant in the development of one of these studies, I’ll give you the inside scoop. Here’s a look at what we set out to do and where we ended up.
Back in August 2002, we started planning a case study approach. We used industry-accepted media planning tools to assess how online media inclusion positively contributes to achieving advertisers’ stated objectives.
Charlie Buckwalter of Nielsen//NetRatings and its sister company, IMS, provided the data crunching and analysis. Kathryn Koegel of DoubleClick found the bucks to fund this little endeavor. (While naming names, I have to shamelessly plug Charlie, Kathryn, and their teams. Working with these folks was been one of the best collaborations I’ve had in the online industry. They worked their tails off. Mucho thanks.)
In return for our participation, we obtained client approval to provide advertising schedules for two brands for Q2 2002. We ponied up background and post-buy data, including:
- Detailed definition of the campaigns’ target audience
- Overall spend in print, network TV, cable, and Internet
- For network TV/cable, a list of networks with gross rating points (GRPs) per daypart and cost per point for each daypart
- For print, a list of publications with number of insertions and pages per publication
- For online, the number of impressions per site and total cost per site
The Rocket Science
So with that data, we set off to create various media mix simulations:
- Reallocate dollars from offline to online media.
- Retabulate target reach/frequency at low, medium, and high Internet spend (e.g., 2, 7, and 15 percent). By the way, these are not recommended target spending levels — just broad enough breaks in the data to determine impact.
- Examine the impact of the reallocation on the target audience delivery: offline, online, and total media.
- Reach our goal of showing the implications for future media strategy.
For the Media Rocket Scientist-Types
If you’re really into analytics, you’re probably wondering how we approached multi-media reach estimation. A few realities and parameters we worked with:
- No single-source research exists to adequately estimate multi-media schedule campaigns.
- A random duplication technique, although inadequate, was used to piece together audience reach across media channels. Though data integration techniques such as fusion have emerged to provide more precise estimates than random duplication, but they’re relatively new in the U.S. and still being tested.
- To get around this, we developed proprietary algorithms to estimate cross-media reach based on media exposure patterns found in MRI data.
The nitty-gritty algorithm details:
- Cross-media duplication was estimated through MRI media usage quintiles, using probability technique.
- Each target was evaluated independently. Media consumption relationships were applied to individual advertising schedules, critical for estimating multi-media quintiles.
(Whew. Time to pop an Advil. If you want more details, let me know. Now, on to the fun part.)
What We Found
Using data from an American Airlines campaign, we looked at the target audience composition for the specific campaign delivery and saw:
1)The online portion of the target is younger, male, and more upscale.
|Airline Target Profile|
|Total adults = 100 Index * Used Web in Past 30 Days|
|Source: MRI Spring 2002|
2. The online target watches less TV and reads more magazines.
3. When we performed a quintile analysis for GRPs and frequency, we saw a nice shift from high frequency cells to the lightly exposed.
4. Without giving away too many details, we were able to determine we could reach, with the same dollars, an incremental 3 million customers who would otherwise not have been affected by the schedule.
Where we net out is adding online to the mix results in reaching more of our target who are lightly exposed to our campaign. Sounds like a good result — more for your money, right?
I challenge the notion that just by adding online, incremental lift in GRPs and effective reach/frequency is the end goal. We’re just now beginning to understand a growing segment of media users who are what I like to call “info on-demanders.” These consumers prefer to access media that provides them with information they need when they want it. This is evidenced by the growth in online media usage, targeted print and select cable.
A quick personal example: I can no longer stand to sit through the six-minute weather segment on the 10 pm news and watch a talking head chatter about this, that, and the other. I can cruise through my own, personal forecast in 30 seconds on weather.com.
The importance of understanding this media behavior is clearer now that we can validate this segment as more affluent, better educated, and not watching a lot of TV.
At the end of the day, I have to ask, “Mr. Marketer, if you don’t include online in your media mix, are you willing to let the competition have 100 percent share of voice against some of your best customers?”
Next time you debate cross media effectiveness, don’t be too quick to cross out online. You might just cross off some of your best customers.
Video consumption keeps increasing and Facebook is serious about a video-first world, encouraging us all to explore its full potential. Ian Crocombe, ... read more
Mike Andrews Ph.D is Chief Scientist (Forensiq) at Impact Radius, and is carrying out some fascinating work around digital marketing and ad ... read more
A new organization, The Coalition for Better Ads, has been launched to “leverage consumer insights and cross-industry expertise to develop and implement ... read more