Cy Khormaee is VP of Product at Attentive, an SMS-first software platform and leader in the conversational commerce space. Cy’s role helps create “magical” mobile conversations between brands and their customers through personalized, customer-first text messaging.
In this interview, Cy shared insights into how AI is transforming and personalizing SMS messaging for marketers, how Attentive uses edit distance as a metric to enable the product’s continual improvement, and how segmentation can uncover the “why” behind trends in data.
At a high level, Attentive enables brands to engage with their customers on mobile. While the core of our business is SMS, we’ve recently been expanding to email as well. My role is to help our customers create great experiences for their audiences across both of these channels.
I think about how we can accomplish this in a couple of ways. First, we are leveraging AI to personalize the experience. The opportunity to evolve from sending the same message to everyone to creating bespoke messages tailored to individuals represents one of the largest opportunities in marketing today. And, it’s a big reason I came to Attentive.
Second, we are expanding those experiences from just SMS on mobile to email. We want to enable our customers to have those conversations in a coordinated way across all the channels their audiences are engaged with.
We operate as a B2B company and create valuable partnerships with a wide array of B2C (and some B2B) brands in the digital retail sector. Our customers represent a broad spectrum of industries, ranging from fashion and home decor to restaurants and electronics. While our primary clients are brands, the ultimate beneficiaries of our partnerships are the subscribers, since end-consumers have chosen to opt-in and receive messages from our customers. It’s important for us to create an amazing experience for both brands and consumers to ensure the conversation can continue.
A great aspect of personalization we’ve developed is something we call Attentive Concierge, a conversational AI that can have a dialog with you about a product or a brand. Let’s say I’m interested in buying a new wetsuit, but am curious if it’s good for 50-degree water. As a consumer, I can ask the AI over text and quickly chat about any other questions I have before making a purchase.
A big problem this solves is information retrieval. Lots of these details can be difficult to look up on a website by reading the description. This may prevent many consumers from taking the next step. But now, we have a text message channel where, among the marketing messages, a consumer can ask for any other information they need before completing a purchase.
That’s something I’m particularly excited by. I loved this experience myself, personally. I like technical specs on things, so not having to hunt those down and being able to ask an AI questions over text is a pretty magical experience.
I think there’s maybe two stages here. The first is personalization — creating messages tailored to the individual. Second, I believe conversational commerce will take off. It’s become de facto in Asia, fairly common in the EU, but is still fairly uncommon in the US. Humans are quite similar in nature, and if some humans have adopted one thing pervasively in a place, it’s likely a matter of time for others to adopt it.
When people are playing with ChatGPT today, they’re starting to understand the value of conversational AI. This should help shift their expectations of this experience to other areas. My hope is that in five years, it’ll become a standard mode of commerce where it would feel strange to not have the option of this experience while shopping online.
Then — and this will take longer — I predict this will be integrated across all the channels. With your consent, the information will be more and more integrated, and your needs will be anticipated. The recommendations will get even better, and you’ll be recommended exactly what you want, or things that will delight you that you didn’t even know you wanted.
One of the best examples of innovation from our AI team is Copy Assistant. One of the biggest problems marketers generally face is the blank page problem. Every day, every week, a marketer needs to sit down and figure out what to send to their user base to get them engaged. It gets really challenging to send new, fresh, innovative content.
One of our first forays into artificial intelligence was to create a tool that looks at all the data we have about a brand. We look at what their products are, their content, their past messages, and we recommend new messages based on all of that.
In just a few months, Copy Assistant was adopted by the majority of our customer base. I think that really is a testament to two things: that it solves a real, key problem of writer’s block and that it’s easy to validate.
I like to think about metrics in a hierarchy where you’ve got a top hero metric that can be broken down into different component metrics to measure more granularly and make sure you’re continually doing the right thing. Typically, as you go lower in the stack, they’re easier and more frequent to measure.
You’d like to have all of your product goals perfectly aligned with your revenue metric. That’s typically not possible, and it has a slower feedback cycle (slowing the pace of iteration). So, you break down those metrics into things that enable faster iteration that is focused on a specific area of the business.
Our hero metric is revenue generated for our customers, and underneath that, we think about what things are stopping this process. Usually, it’s not sending enough messages, or the messages aren’t high quality enough, which are both functions of marketers’ time. Does the marketer have enough time to send a message? Does the marketer have enough time to send a high-quality one? In Copy Assistant, those two things break down to edit distance, which actually fixes both — you’re creating a message that the marketer likes and thinks is high enough quality, in zero time.
It’s very hard to generate big movements in revenue per subscriber or per message, but I can measure edit distance on a daily or per-message basis and make rapid changes to it.
It’s interesting to think about, and as we evolve, we’re doing more and more work — for example, in the EU — around how this translates to other languages like French or German. I would generally expect a measure we’ll use is called Levenshtein distance — how many operations between moving or changing characters it takes to get from one word to two words. So, from “dog” to “dodge,” what are the characters that have to change? For example, in Romance languages, it’s likely to be very similar, whether it’s English, German, or French.
The cool thing about AI is the difference in making that transition. For a typical product, there’d be a huge internationalization of localization efforts. For AI, it may only matter a very small amount or not at all, because we’re purely looking at things like pattern matching and the terms that really matter.
It’s an exciting time to be applying AI to this space. This revolution really kicked off six or seven months ago, and I think we’re maybe only 1 or 2 percent of the way through the maturity curve we’ll see.
Yeah, I mean, there’s a bunch of different ways we can segment. The first thing that matters is really, what am I testing for? For a global change across our platform or for something that’s customer- or channel-specific? Those all inherently create their own segmentations.
Segmentation is really how to get to that “why.” For example, we can split this up and look at traffic from France versus Germany versus the U.S. If we see one country’s traffic particularly high, we can assume that’s probably the leader and look at that segment’s performance and how it influences the aggregate.
To start, we look at overall customer revenue generation, our hero metric. Then we’ll look at key anti-metrics like opt-out rates. It’s important to keep both in mind and balance when observing customer behavior.
For example, if a customer were to send more messages and, potentially, less relevant messages, it may increase revenue but might also increase the opt-out rate. We want to balance those two and, obviously, decrease the opt-out rate with as much relevancy as possible and increase the conversion rate with as much targeting as possible.
Some of the privacy regulations around telecommunications spam and scamming are the foundation of why this company exists. All of the subscribers that our customers message have deliberately opted in and asked to receive messages. That positions Attentive in a very unique place from a privacy standpoint by enabling our customers to obtain direct consent from users to message them.
Everything Attentive does is for users who have deliberately opted in. We use extensive audit logging and security controls along the way to ensure this is consistent. I foresee the efficiency and importance of platforms in the space growing as the importance of privacy becomes more recognized and valued. My belief is businesses based on delivering messaging without consent are going to be continually hampered by these regulations.
It’s almost like a whole new channel and relationship between consumers and marketers that will be enabled by AI. There’s one marketer for millions of consumers — there’s no way to send hyper-relevant messages. You can’t please everyone with exactly the same message, we know that’s almost universally true. With LLMs, we can work in partnership with marketers and use everything that brand knows about a user to send them the most relevant message possible. That’s a really exciting future.
LogRocket identifies friction points in the user experience so you can make informed decisions about product and design changes that must happen to hit your goals.
With LogRocket, you can understand the scope of the issues affecting your product and prioritize the changes that need to be made. LogRocket simplifies workflows by allowing Engineering, Product, UX, and Design teams to work from the same data as you, eliminating any confusion about what needs to be done.
Get your teams on the same page — try LogRocket today.
Authentic leadership occurs when you’re genuine and act consistently with your principles and values, as well as with honesty and integrity.
Ajoy Krishnamoorthy, Chief Product Office at Cin7, discusses data-driven decision-making and his experience launching v1 products.
In our pursuit of innovation, we must be aware of the crucial aspect of letting go — the decision to shut down a product feature.
Justin Kim, VP, Product at Vimeo, discusses the importance of applying ruthless prioritization across the board and curating a clear vision.