TV-Driven Branded Search

Customer AttractionJoo, M.; Wilbur, K. C.; Cowgill, B.; Zhu, Y. · 2014Management Science

Topicstelevision advertising·online search·cross-media effects·search engine marketing·advertising elasticity·financial services

Your TV and search budgets sit in separate rooms, planned by separate teams. After a TV flight, more people may type your brand's name instead of a generic search term. Do you prepare your search coverage for them, or let the opportunity pass?

More TV spending is associated with more branded searches over the following four days.

Data chart

Branded-search response to TV spending is highest in the early morning

Percent increase in a brand's searches over the next 96 hours for each 1% increase in its TV spending, by hour of day.

Overall, 10% more TV spending goes with about 1.7% more branded searches.
Weekend hours run slightly lower than weekday hours.

The response of branded searches to TV spending peaks in the weekday early morning, is lowest in the late afternoon and rises again by midnight.

Model estimates by hour of day, read from paper Figure 7, which gives no table; almost all of each value is searches shifted from rivals rather than added to the category. US financial-services brands, 58,226 TV ads linked to hourly Google searches, October to December 2011; exact values .217, .154, .197 and .140, against .17 overall. This measures searches, not sales, clicks or return.

Data chart

The switch to branded keywords comes within hours; category searches stay higher for days

Hours or days after a TV ad airs until each search response peaks or fades.

The switch to branded keywords peaks about 6 to 11 hours after an ad airs.

The switch to branded keywords peaks within hours of an ad and is not statistically distinguishable from zero after hour 17, so branded-search coverage has to be ready soon after a flight. Category searches stay higher for about three days.

Hour-by-hour model estimates (paper Table A2 and Figures 5 and 6). The largest hourly estimate for branded-keyword choice is at 8 hours, inside a peak from about 6 to 11 hours; from hour 18 on, the estimates are not reliably different from zero. US financial-services brands, hourly data, October to December 2011.

Action guide

Approve one plan for TV and branded search.Require your TV and search teams and their agencies to deliver a single plan covering budget, flighting, creative and measurement, because TV spending and branded searches move together.
Give TV credit for the branded searches that follow it.When searchers switch to your brand's name, more paid clicks should come from branded keywords, which usually cost less than generic ones (the study did not measure clicks). Judging search on its own risks spending too little on TV and too much on search.
Protect branded-search coverage after each TV flight.Approve the search budget and coverage for the period that follows, and have your search team show how it will meet any rise in branded searches.
Judge each TV flight on both search responses, over different windows.Have your analytics team report category searches and the switch to branded keywords beside TV spending. Give category searches the longer window, because they respond more slowly.
Approve branded-search budgets based on your own TV plan.In a mature category, do not assume competitors' TV spending goes with more searches for your brand.

Evidence

A 10% rise in a brand's TV spending is associated with about 1.7% more branded searches over the following 96 hours.
On average, taking search share from rivals accounts for roughly 100 times as much of the increase in branded searches as growth in category search does.
One brand's TV spending has little positive association with searches for its competitors in this category.
The switch from generic to branded keywords peaks about 6 to 11 hours after an ad airs.
Higher category TV spending is associated with more category searches, a rise that builds slowly and lasts roughly three days.
The response of branded searches to TV spending peaks early on weekday mornings, at about 1.5 times its low on weekend late afternoons.

Key takeaway

More TV spending is associated with more branded searches, so plan TV and branded search together.

Limitations

The study covered TV ads and Google searches for established U.S. financial-services brands, October to December 2011, and counted searches, not clicks, conversions, sales or profit. TV for new brands or changing categories would likely add more category searches, and take less share from rivals, than it did here.

Source

Joo, M., Wilbur, K. C., Cowgill, B., & Zhu, Y. (2014). Television advertising and online search. Management Science, 60(1), 56–73. https://doi.org/10.1287/mnsc.2013.1741

https://doi.org/10.1287/mnsc.2013.1741

Exact values behind the rounded figures: branded-search responses of .217 at weekday 6 AM, .154 at weekday 4 PM, .197 at weekday midnight and .140 at weekend 5 PM for each 1% more TV spending, against .17 overall.

Read the paper ↗