AI Adbot Launch and Growth Strategy
Now that all the copywriters that were psyched over my last post are diligently preparing for the onslaught of AI AdBot briefings lets now focus on measurement and analysis.
I can wax lyrical about the raw potential AI AdBots bring our illustrious industry.
The real time 1:1 engagement that a conversational medium affords us is absolutely mind blowing.
Who would have thought that social messaging app platforms would have ultimately become the fertile soil from which eCRM would rise like the phoenix from the ashes of direct mail, email campaigns and targeted banners?
Having said that, let’s examine one of the more exciting opportunities AI AdBots bring which is an opportunity for deeper and more complex ways to measure stuff.
And we all know how much us ad folk love to measure stuff.
How can we measure performance and predict growth for AI Adbots?
At first I got that “oh no its that crazy creative technology guy again and he is probably chasing down another one of his hair brained ideas and is going to pull me in” look, so I put down one of my drinks and asked more succinct questions.
What is the best way to measure AI AdBot activation and retention?
What AI AdBot metrics are most important to track?
How do we attribute new AdBot installs?
Kevin then started to take me a little more seriously.
At first I thought we could base measurement loosely on traditional web and mobile analytics and user behavior.
But then I put down my other drink, and postured that since this is such new territory and there are really no rules yet, why don’t we make up some mind blowing metrics that will have the industry salivating?
So with drink in hand, Kevin and I walked over to the back couches overlooking Columbus Circle and started to chat.
We decided that we should create a set of AI AdBot events and get them into (I can’t believe I am saying this) a funnel so we can better understand how people find, use, and interact with AI AdBots.
We would then start to plot out what those customer expectations and experiences would be from an outcomes perspective.
Wait… no, thats not right.
Can we create a coherent AI AdBot journey? Maybe.
Could we accurately identify (within a conversation with a robot) all of the moments of truth where a customer derives value? Those key events, interactions, engagement signals, etc.
In order to do that we needed to be able to name all of those interactions.
Then I asked young Kevin, “can we create our own naming conventions?”
For example can we make up bot events like “API ACTION TAKEN” or “MULTIPLE VERBS USED” or “INTRODUCTIONS MADE” could we measure sentiment like “SARCASM FREQUENCY”?
Once I managed to convince young Kevin that I wasn't drunk or crazy and after I hid his iPhone so that he couldn’t pretend that he had something more important to do, he shot me a question.
What is this AdBot’s primary objective?
Without thinking I simply said “duh! to drive sales” — I think Kevin might have actually been impressed with my quick retort.
Kevin then said, ok cool, lets think about how we can create a data-driven launch strategy.
He completely lost me.
But I was listening…
Kevin suggested we create a use case. Maybe loosely based on one of our clients.
So let’s say that we have somehow determined that smart phone customers really desire the ability to personalize the software and customize the hardware for the mobile devices they want to buy.
I know I would.
If that is the case then the primary objective of the AI AdBot would be to first engage people (lets say on Facebook) who have shown some interest in getting a new personal device.
We could then use customer FB profiles to teach our AdBot about what some of those product personalizations and customizations would be based on profile data.
Then we could easily target those customers to engage with our AdBot on FB Messenger.
Some paid media would get the message out that our chatty AdBot could help make mobile device customization and personalization easy and seamless.
This would help us learn if our AdBot was more effective in delivering value versus a web-based or mobile app process which requires a lot of self reporting and we all know that sucks.
Then young Kevin sobered me up with his next suggestion.
Brand X’s AI AdBot must be equipped to manage customer service as a CRM tool because it will be disastrous if this AI AdBot gains a user’s trust, user buys said product or is very close to buying it and then submits a customer service inquiry and the AI AdBot doesn't have the intelligence to support those kinds of requests.
All of a sudden our conversational cyborg is now exposed to be one dimensional.
Kevin convinced me that customer service has to be foundational and integrated throughout.
According to NewVoiceMedia, poor customer service in the US costs companies $41 billion each year. Also according to Zendesk, 40% of customers switch loyalty because a competitor offers better customer service.
Kevin’s next statement started off with two words that I personally hate, “Potential Barrier.”
Kevin proceeded to question me on how we were going to drive users to engage with this AI AdBot.
Were we simply going to assume that we would get a higher conversion rate due to its newness and chatty personalization methods vs traditional methods and channels?
Kevin proceeded to answer with another good question.
What is the incentive/convenience factor here for brands that we need to highlight?
I immediately answered “how about optimization & growth?” — you could answer almost any question in advertising with that response so I just instinctively used it.
Turned out I was spot on…
Kevin went on to say, that to effectively understand whether the Adbot is successful in achieving its goals, a new kind of measurement framework based on traditional web, mobile, and social analytics needs to be applied.
This would help to identify key triggers during the conversations these AdBots are engaging in and then measuring the kinds of information they are learning from in all of these unique simultaneous conversations.
Kevin went on to say that as consumer data flows in, the measurement framework will be continually optimized to specialize in analyzing all of the AI AdBot activity on both the quant and qual level.
This could include such things such as sentiment, purchase intent, secondary requests (like ask UBER to deliver me the device now), content shared, conversational subject matter, frequency of engagement, etc.
Kevin then suggested that we continue to test and evaluate our AI AdBot with new tactics like branded content and supplement that with competitive/industry trends.
Kevin, young Kevin, so bright and un-jaded.
How does all of this stuff you are saying help brands?
Young Kevin is pretty bright…
The Adbot has potential to become core to the business, tying in content, commerce, products comparisons, utilities and services.
It’s a brand new channel with the same business objective of driving sales but from the consumer perspective, it’s one place to receive service, offers, entertainment, and even out-of-category utility value.
Kevin didn't just stop there, he went on to say:
It’s a completely new approach that requires corporate reorganization, adoption of new technologies, and buy-in from all key stakeholders in order to ensure success.
Ok young Kevin, so bright and eager, but where do you start?
Kevin suggested we start with a proof of concept by demonstrating the user experience on the most basic/simplest level — a conversational mobile device personalization tool with some basic customer service integration.
Kevin then suggested we develop an AI AdBot vision and product roadmap.
Then in true millennial fashion he concluded by saying:
Operate like a start-up!
I stood up, clapped slowly and then gave Kevin a really awkward hug.
Kevin and I are now really good friends. I introduced Kevin to one of my favorite copywriters who is now interested in learning more about how she can write scripts for these new AdBots.
Now the three of us (Technologist, Analyst and Copywriter) are going to ride off into the sunset and build an award winning AI AdBot that will charm the pants off of social network users everywhere.
In all seriousness, this is a very exciting time. There are no rules yet, no right or wrong in how to effectively measure a 1:1 conversation being had by a robot and a human.
However it will be exciting to see how deeply we can measure these interactions and identify and even create new metrics that will help these AI AdBots become extremely effective over time.
I will leave you with this amazing quote that I feel we should all keep in mind when developing AI AdBots.
“Here lies the challenge in finding good salespeople. You need excellent empathizers who aren’t so empathetic they can’t close a sale. And you need people with strong ego needs who can still take a moment to figure out what another person wants. They must be aggressive enough to close, but not so aggressive they put people off. Too much empathy and you’ll be a nice guy finishing last. Too much ego drive and you’ll be scorching earth everywhere you go. Not enough of either and you shouldn’t be in sales at all. It’s a miracle anyone can do this job.”
― Philip Delves Broughton, The Art of the Sale