Why such a fuss over a report that's 18 months old - ancient by digital signage standards? Two reasons. First, the results that they claim are excellent, and if they continue to hold over time, demonstrate the true potential of digital in-store advertising. Second, the experiment features all of the controls and constraints that allow meaningful analysis of the network's performance, and can serve as a good model for others who plan to conduct the same sort of tests. For example, let's look at the test setup that CompUSA used:
Background: 182 CompUSA stores participated. Stores were divided equally into a group of 91 Test Stores that played participating sponsor advertising within the CompUSA TV program (7 advertisers participated) and 91 Control Stores that did not include the 7 advertisers' commercials as part of their monthly program. Comparable store size, market and sales were used to provide accuracy of the test.The first thing that you should notice is how they attempted to isolate the "placebo" effect of simply having the screens in place and showing generic content, versus having the screens running with relevant ad content. Unlike many other reports that I've read, where stores with screens running ads were compared to stores that didn't have any screens, every store in this study was equipped with the screens. The only difference was the content displayed on them. Given the large number of plasma screens, LCDs and computer monitors that the typical CompUSA has on display, it could be that a few extra digital signs woudln't have any significant impact on consumer behavior. However, without taking this extra step, it would be impossible to know for sure. Additionally, they set up an extra 43 stores to display the same content (including advertisements) on the wall of computer monitors typically found at the back of the stores, but without any additional digital signage screens. This further isolates the effect (if any) of having the additional dedicated screens in the stores.
The results of the 90-day trial can be summarized as follows:
- There was an aggregate sales lift of 29%
- All sponsors for the test saw increases during the first 30-day period that the ads were run
- Every month, the test stores (those showing the ads on digital signs) outsold the control stores (those with generic content on their digital signs, or those without the digital signs at all)
- Advertisers who purchased the most time on the network saw the greatest sales lift
- Employee brand awareness increased 23%
One last thing that I wanted to work through was the rate card that they provide. CompUSA claims that advertisers saw increases for all ad formats, from quick 15-second spots to longer 3 minute infomercial-style productions. Their typical loop lasts 30 minutes, which coincides nicely with the stores' claimed average visit length of 35 minutes. And according to In-Store Media Networks (the company that runs CompUSA TV), the 245 CompUSA stores host about 8.8 million shoppers per month (which works out to about 36,000 shoppers per store per month, if you're interested). Excluding the holiday period, a 30-second advertisement on the network costs about $10,000. Thus, the quick-and-dirty CPM for a spot running once every 30 minutes (and thus optimally reaching every visitor at least once during their visit) is about $1.13. During a holiday month, the rate doubles, and a 30-second spot costs about $20,000. While it would appear that CPM would also double to roughly $2.26, the holiday months (according to the rate card) last for 5 weeks. And as anybody who has been to a CompUSA between Thanksgiving and Christmas knows, store traffic increases dramatically as well, so it's likely that the CPM is only incrementally higher than during the rest of the year.
Now, let's go back to that testimonial from Symantec above. They claim that they saw a 107% uplift during January of 2005, which according to the rate card is a holiday month. Considering a generous capital outlay for content production of $100,000, and another $20,000 to purchase the airtime on CompUSA TV, the total cost of marketing through the medium would have been $120,000. Guessing that Symantec's average product costs about $50, and also guessing that they make about 20% gross margins on boxed software sales, that would mean that they would have to sell an additional 12,000 licenses during that month to break even (that's only about an extra 49 licenses per store). If we take the cost of content production out of the equation, they would only have to sell an additional 2,000 copies (or about 8 per store) each month to break even.
Since we don't know the true cost of production, actual gross margins, or how much software Symantec sells in the typical CompUSA store, it's hard to calculate ROI. But given the quick figuring above, it seems evident that as long as they are able to keep content production costs under control, they will see very healthy growth in their bottom line just by advertising on the chain's digital signage network.
How does this information help others who have deployed (or are deploying) digital signage and narrowcast networks? First, CompUSA's testing methodology makes a strong case for doing a proper control experiment to see how digital signage displays contribute to sales lift. Second, we can see what a national brand's digital signage rate card looks like, and get an idea of what some companies are willing to pay for in-store TV on a CPM basis. Finally, thanks to the testimonials, we have additional proof that big companies support the medium as a viable form of in-store advertising.