PETER MAHONEY: There’s a huge challenge in mobile right now around monetization—it doesn’t monetize nearly as well as the other media platforms. And you also have the rise of the voice recognition systems like Apple's Siri, Samsung’s S Voice and Google Now, which popularize voice inclusion. We think that giving the ad creation tools to agencies in voice recognition dialogue responses we can increase engagement, increase conversational brand awareness, and that will lead to a higher monetization.
PM: We’ve leveraged a lot of the same technology backbone that we have around the voice recognition, natural language, dialogue management that you see in [Nuance’s voice-activated virtual assistant] Nina. What we’ve done is taken some of the same infrastructure and created the ability for people to build an interactive ad experience into a rich media ad unit.
So what led Nuance to get into the advertising industry?
PM: [Clients] were asking us to help with the creative copywriting and we said we’re a technology company, we don't really do copy writing. We can give them the tools to create the content and then we decided, why not build a whole voice recognition suite and write the responses to the spoken query for advertisers and let them embed the experience?
The agencies and brands are really excited about this because the whole concept of the creative that we’ve seen in mobile advertising is pretty lame right now. They’ve taken desktop ads and tried to squeeze it down into basically little banners which doesn’t really translate for people.
Can you show me an example of a voice-activated ad?
MIKE MCSHERRY: We initiated a banner ad with a fictional brand, Alpha Deodorant, but it could be a full-blown rich media ad. The campaign that we created was around don’t sweat the petty details, ask the Magic 8 ball.
[Using an iPhone, McSherry taps on an ad with a Magic 8 ball image.] The ad asks, "What do you want to know?"
MM: Should I call her?
Magic 8 ball ad: Umm, no. But on the off chance that you run into each other in person, use Alpha. Does anyone else smell a sure thing?
MM: We’ve scripted in 50 different responses to various questions. We’re picking off what the user says and providing a scripted response back. Or it could even be an Amazon ad for Father’s Day. [The ad could have some text like] running late, forgot what to buy? Click here and then say, “Black and Decker tools on Amazon.” And that would take you straight into that section on the Amazon site. Ultimately it’s up to the creativity of the ad producer what sort of experience they want to create with this engagement and drive the engagement level on that front.
How would a consumer know to ask the ad questions?
MM: Because the ad is asking you to say something, there are two microphone icons. Even though more education is needed, people are getting more and more familiar with the use of microphones and voice dialogues.
What’s the estimated market for voice-enabled ads?
PM: The estimated market at this point lies in the demand and the interest of the partners that we’ve lined up. The agencies are getting the brand commitments on that level. Where we have seen the most interest actually has been in the ad agency side where it provides a new creative dimension to building an ad. It takes it to a new level of experimentation whereas I think voice recognition to date has mostly been limited to large OEM players. I think this almost democratizes access to using the voice tools at a level where any ad can be created around it.
How many languages does it include?
MM: It covers 40 languages [such as] English, Russian, Hebrew, Turkish, Japanese and Mandarin, a lot. And we’ll be adding more languages.
Are there any plans to combine this with geo-location data?
MM: Yes, and that’s where the ad networks come in. They have the ad targeting data. We’ll let the ad networks do the targeting element but we’ve got the interface element to this. Voice recognition is 95% gender accurate, so on that front we can contribute that into the mix of data sets against the user and when the ad creators are building the ad they can also assign keywords associated against that. So say a Harry Potter ad has the words Hogwarts, Dumbledore and Slytherin. These words are very specific to the Harry Potter experience but generally speaking they don’t get much use outside of that. But you can assign those keywords against that user for use in other voice application environments from the desktop to the mobile, car, et cetera for certain language models up in the cloud.
Would your platform allow advertisers to store the responses to the ads?
MM: Yes. We capture everything and send the responses back to them. Initially it’s probably going to be in an Excel file, here’s the campaign response but that’s another area that we’ll be developing. We’ll provide access to things like polling results and other metrics. Many of these partners will have that as well. They’ll have their own systems. We’ll continue to work with the ad networks on how we can better integrate their systems to share more and provide visibility into the campaigns and response rates, et cetera.
PM: One of the things we’ve been saying is instead of cost per click you get to a metric of cost per conversation. For example, we had people go five levels deep in that ad, or we had a 40% response rate, or we gained this much data inside of what our customers think about XYZ.
What about sentiment analysis?
MM: I don’t know if we’d provide sentiment analysis but if you look at the Magic 8 Ball example, you could find out if more people saying, “Does she love me or does she hate me?” And I guess you could infer things from that. All the dialogue responses are written on the server side so you can change campaigns on the fly. And again to the gender accuracy, I know I’m talking to a female or male when they return this result to this question or this page versus that, as well.
Will these voice-activated ads also work on a desktop or a laptop with a microphone?
MM: We’re launching on mobile devices for iOS and Android users. With mobile devices, the microphones are optimized for voice and noise cancellation capabilities. When you extend into the TV domain there frankly aren’t that many voice recognition TVs out there and you certainly don’t reach a billion users, so you can do some trials on it but no advertiser wants to essentially reach 2% of households. PCs are interesting and you could embed a web banner but it won’t read the microphone capabilities on the device. Flash will allow you to do that but then it pops up with a warning, hey, this application’s trying to access your microphone, which disrupts the ad experience.
What’s your business model for Voice Ads?
PM: For this it will be a CPM-based media, so there will be a charge, like a rich media unit charge on top of the ad impression that we’ll be selling. We’re signing deals with ad networks. This will be in a creative CPM cost on top of the standard rich media ad cost. We’ll see what the market will bear. It’s certainly not an exorbitant charge, it’s a small percentage since we want to encourage the use of voice in multitudes of experiences.
So how soon can we expect to start seeing voice-activated ads?
PM: We expect in the next couple of months to be able to see campaigns out in the market. There are a bunch of agencies talking to their brand customers already about the concept and they’re really excited about it.