Customer Success Secrets from Pinterest

Michael Krigsman: Wow, oh wow, do we hear a lot about customer experience these days. But, when we hear the buzzword "customer experience" and people talk about placing the customer in the center using the customer as a reference point, what does it mean in practice? How do you do it? When you move away from the buzzwords and you actually start executing real plans for real customers, things become much harder. Today, on Episode #290 of CxOTalk, we are speaking with somebody who knows. We're speaking with Dutta Satadip, who is the global head of customer operations at Pinterest and, previously, he was at Google.

I'm Michael Krigsman. I'm an industry analyst and the host of CxOTalk. And, before I forget, be sure to subscribe on YouTube. There's a tweet chat taking place right now using the hashtag #CxOTalk.

Dutta Satadip, how are you? Thanks so much for being here.

Dutta Satadip: Thank you, Michael, for having me.

Michael Krigsman: Dutta, tell us about your role at Pinterest.

Dutta Satadip: I joined Pinterest recently. Just to give a quick background, Pinterest is in the business of helping users discover what they would like to do and then do them in real life. And, I joined Pinterest from Google because I think there's a big opportunity combine customer experience, bringing the customers at the center of how we operate as a business.

Michael Krigsman: How do you define customer experience? Let me ask you that way.

Dutta Satadip: The customer experience is effectively the sum of all the interactions that any customer or user has with the business. And, the reason I say it's all the touchpoint is because it's just not when a customer calls for help, but it starts from the first time anybody engages with your business and the lifetime relationship they build with your business. It's not just transactional; it's a combination of transactions, summing all of them up, and the overall relationship that it builds together.

Michael Krigsman: It's interesting to talk about transactions because, when we talk about customer experience, people tend to talk not about transactions; they talk about empathy and feeling. And so, what do you mean by transactions? Maybe draw those distinctions for us.

Dutta Satadip: Absolutely. I think, traditionally, a lot of customer experience work has been around understanding customer satisfaction, understanding net promoter scores, and these are great, I would say, backward-looking indicators, scores, and measures of how customers engage with the business. What has also been part of this journey is, for the most part, these have been mostly centered around the customer service type of organizations, which is mostly: I ask a question and I get a response, and how well I do in that.

I think, when we talk about empathy, a lot of this is around, specifically, are we listening to customers well? Are we answering their questions well? Are we empathizing and putting ourselves in the shoes of the customer so that they can get a very good outcome for an individual transaction?

I think, as businesses have grown and evolved, most customers now have to focus on not just acquiring and answering questions, but a lot of it is around customer retention. And, thereby, now the shift has become, okay, what is a great customer experience? What is it that we are doing to create customer outcomes? When you combine experience and outcomes, that's kind of customer success and what makes it really customer operations. Those are the great external things for customers, but I think the big shift that we're all trying to work towards is, how do we, as businesses, understand these goals and put the customers in the middle of all of our business operations and how we execute internally?

Michael Krigsman: I think everybody talks about doing that. First off, then, why does everybody talk about that? Then, maybe we should discuss the challenges of actually doing it in practice.

Dutta Satadip: I think this is one of those things, right? It's like motherhood and apple pie. It sounds amazing. It sounds good. But, it's extremely hard to do.

Traditionally, the way most organizations have evolved/grown is, there are executives, there are functions, and each of the functions are incentivized to perform one or two things and do them really well. Those types of incentives typically drive some version of either siloed organizations or siloed incentives. That is why it is super hard to bring in the concept of customer in because, when we talk from the perspective of a customer, they honestly don't care whether marketing's goal is X and customer goal is Y. They want a seamless experience no matter who they touch, right from the start when somebody reached out to them to do something and ask them to invest in their company, in our case in the form of an advertising campaign, all the way down to when there is a billing issue, and somebody needs to resolve that.

I think it's that connecting of the dots across the different organization silos and the measurement and the metrics that go around that have to now evolve to being more cross-functional. That's the pivot that we are working towards.

Michael Krigsman: These are organizational issues then that interfere with having the right kind of customer experience.

Dutta Satadip: I think that is probably one of the biggest drivers and then the biggest challenges in making this vision real. It is a lot about organizations. It is definitely a lot about understanding the challenges in a data-driven way, in an objective way and, ultimately, it is also about the overall culture of the organization.

Michael Krigsman: When you talk about it in a data-driven way, that seems like a very key point, especially today.

Dutta Satadip: Absolutely. I think one of the things is, back at Google when I started there, I was there for seven years. When I started initially, Google was growing really well and, one of the things I observed was, we celebrated a lot. But, what we celebrated were stories: the time that we saved the customer, the time that we got the great deal. The challenge with stories is, we don't know whether it is a one-off great instance and an example or an actual pattern that we need to suss out, scale, and do more of.

Where data comes in is to be able to help us, to graduate us from these amazing stories to some sort of stats, right? This evolution from stories to stats is what data provides us. It helps us identify, is it truly a problem; is it a meaningful problem; and is it something that is going to be repeatable across wide ranges of customers, wide ranges of internal business processes, and then be able to solve that?

Michael Krigsman: That's really fascinating. Data helps marketers move from stories to stats. Then what? What's the next part of the chain?

Dutta Satadip: I think most of the data in this journey that we start with is often backward-looking, and that's how the journey is. It's nothing bad or good. We have data, we start to analyze the data, but it's typically backward looking.

The true opportunity is to take all of this data and convert it into a science. The amazing thing is there's a lot of technology available that wasn't available before to actually convert this data into forward-looking insights. On the high end of this, you could apply machine learning algorithms, et cetera. But, on the lower side of it, even getting to basic stats that look at making predictions, making recommendations, is a very, very good start.

The journey that I would summarize is, if you're at the story stage, how do we get to stats? And, if you are at stats, how do we convert the stats into a science?

Michael Krigsman: Okay. How do you convert the stats into a science? [Laughter]

Dutta Satadip: It sounds like a great story but, unfortunately, it involves a lot of nitty-gritty, crazy amount of detailed work. So, let me give an example of how we are going about it. Ultimately, I believe these journeys are successful when we understand what the purpose is we are solving and what is the problem we want to solve. Trying to boil the ocean usually never ends up in the right position. Trying to select a technology platform before understanding the business problem typically does not end up in the right direction.

One of the challenges that we are working off at Pinterest is, how do we ensure we're a growing business? We have over 200 million active users on Pinterest. We want to grow these numbers. We want to make sure our advertisers and our pinners find value in the system.

As a newer platform, we've been a company only since 2010. We're going and acquiring a lot of new customers. Now, as we all know, to acquire new customers, our sales teams should be spending more and more time with customers. When we look internally, we see a lot of time being spent on internal activities versus on external activities. So, our goal is to have the best customer experience. But, if our sales teams are not really talking to customers, understanding their requirements, we are probably missing the mark somewhere. That's the basic, core problem we are trying to solve here.

What does that mean? For us, it means truly unpacking what is driving inefficiency I our system? At this point in time, we're in the process of making sure that we are building that one view of the customer, all our customers' information, not only what they have spent with us and all the financial stuff, but also connecting that with all the issues, et cetera, that they have seen so far. How have we tried to work with them? How are our product adoption metrics looking? Give a 360-degree view of that customer in one place because, if we've tried to solve that problem, now we have the foundation to go back and understand very different types of questions, specifically questions like: What kinds of services; what kinds of help can we provide to customers to maximize their outcomes?

For us, outcomes is campaigns. Is it, we need to give them a better understanding of measurement? Is it, we need to give them a better understanding of their creatives, which is the image that shows up in the ad? Those are the types of conversations we want to build towards, but our first and the foundational step is having a view of the customer in one place.

Michael Krigsman: Is this the difference between customer experience and customer operations, or is this how customer operations support creating the right kind of customer experience?

Dutta Satadip: I think this is the beginning of the difference. This whole data-driven approach is, I think, one of the key hallmarks. You're absolutely right, it's being done in service of making sure that we are able to deliver not only a great customer experience but also have great customer outcomes.

To make it real, when we talk about experience, it's probably not the best experience if somebody walks into a meeting with an executive not knowing that there was a trouble running their campaigns and three of their campaigns had some sort of an issue right before the meeting. Chances are they executive who was going to have a conversation knows about that and is going to bring that up. But, that's sort of in service of experience. In terms of operations, if you have done something really well, having that snapshot tidbit right at the hands to make that conversation happen is in the service of customer outcomes.

Michael Krigsman: I want to remind everybody that we're speaking with Dutta Satadip, who is global head of customer operations at Pinterest. Right now, there is a tweet chat taking place using the hashtag #CxOTalk. Please, join in, and you can ask your questions.

Dutta, as you are relatively new to Pinterest, and as you think about the investment decisions, investment of time and money, that you need to make in terms of furthering customer operations and the support of customer experience, how do you build the framework for making those kinds of investment decisions?

Dutta Satadip: That's a great question. When I joined Pinterest, one of the things I knew was, even though I came from Google and I was I a similar business around online advertising, one of the things I did was actually to go and talk to everybody inside and outside the organization. I talked with about 40 to 50 sales team members to understand, what is it they did with customers; why were customers buying from us; what were the friction points?

What that allowed me to do is to get a very bottoms-up understanding of the challenges that are pertinent to Pinterest at this point in time for the business circumstances, conditions that we are in. Understanding that allowed me to actually stitch together what I talked about before a little bit, [which] is, what are the touchpoints that matter? Specifically, how are we working with our customers from start to finish?

The way we articulated it out is, we have effectively a few phases that we go through in our engagement interaction with our advertisers, number one. We are doing some version of planning with our customers, right? Our customers know about us. They have expressed some interest. We're raising awareness, but we're doing some planning. We want to understand what they want to do.

Then, we want to go and do some version of a pitch, right? We construct a pitch. We go to a sell, do one, and so forth. Our business is advertising. We get commitments on how much they would like to spend with us. Once we get that commitment, we actually go back and implement. Implement is our next phase.

Most advertising campaigns need some kind of optimization, tweaking, to make sure we are meeting the customer objectives. Optimization is the next phase.

We do some level of up-selling. We tell them, if they put some more money, they can get better outcomes. Maybe they can get more reach, or maybe they can get better conversions, which is people going to their website, right?

Last, but not the least, we want to package all of this stuff in the form of a nice measurement and give them synthesized outcomes that we can go back and have a discussion again, which is back to the plan phase and say, "Hey, we delivered this for you. What can we do more for you? How can we help your business grow?" This process of going from plan, pitch, implement, optimize, upsell, measure, is the lifecycle that we came together after having these conversations internally and externally.

Michael Krigsman: For you, investment decisions follow the customer and the advertiser lifecycle.

Dutta Satadip: Absolutely. We look at, how do we want to operate, what does good look like, and what are the biggest friction points in each of them? Then, we can make a decision, "Hey, this is the place that has the biggest friction and, if you invest something here, this is going to give us the biggest bang for the buck." Obviously, we are running numbers. Obviously, we are running models that are not always perfect, but that's the overall approach we are taking to prioritize our investment decisions.

Michael Krigsman: Now, the customer lifecycle at a company like Pinterest would be quite different from that at Google. Although, I guess, for both Pinterest and Google, you have multiple customers and segments and so, therefore, different customer lifecycles. But, when you were at Google, was this similar? Did investment again follow the customer lifecycle?

Dutta Satadip: Yes. We actually went through this process, a version of this process. It was obviously different because Google has many different products in the portfolio. From the customer perspective, I had the full portfolio of advertising products, which is over 100 products. Yes, we had to do a little bit more segmentation. Also, the growth trajectory of each of these products determined where and how much you would prioritize. But, the effective framework remained the same of how we went about finding what matters to customers, what will drive the best outcomes for customers, and what were the biggest frictions points for us to enable those outcomes.

Michael Krigsman: How do you make even those decisions because, as you're going through this process, you'll have people who raise their hand or raise their voice the loudest? How do you decide? How do you figure it out?

Dutta Satadip: I think that is the reality of these roles and this journey. People feel very, very passionate about their segment of the business or their product or how investments should go into a particular product line because it's growing faster than others, right? This is why I say starting to establish facts from the get-go and using data to drive the decisions is a way to sort of bring this together.

Now, data is a very technical way of doing this. One of the things that I have always believed is really trying to understand what is it that we want to optimize for. A lot of this is understanding what my stakeholders want. Because this is an extremely cross-functional role, one of the things I would say is I spend a lot of time not only at the ground level, but truly understanding what my peers, the different executives for the different functional roles, are looking to accomplish.

Is there a better happy medium? It's not always if the data says X, let's just do X. Sometimes there is qualitative information that needs to be funneled in.

The second thing that I always say is, none of this is locked and loaded, and this is the strategy. We're just going to take six months and execute. I think the best way to look at this is launch and iterate. Try something. If there are two competing points of view and we really don't have a good way to understand, do experiments on both of them with a smaller set of customers and see which one is yielding the best outcomes. Then make the pivots and the investment decisions accordingly.

What I've observed is typically it's not a choice amongst 40 things. We're really debating about two or three things. Being able to pilot something with two or three choices is far more manageable than trying to boil the ocean with, like, a list of 30 or 40 objectives.

Michael Krigsman: Dutta, how do you reconcile the data and transactional view of customers with the marketing view, which is about touchy-feely and empathy? Can data capture that empathy in any way; measure it in any way?

Dutta Satadip: I think, while there is technology to look at things like comments, scale, and do sentiment analysis, et cetera, this is a little bit of an art and a science. Right? I think the data is a good way to shorten the problem or refine or segment the problem. But, ultimately, there is obviously an element of understanding the business, having business intuition, leveraging our own experiences to layer on top of it.

Even for things like marketing outcomes, certain things we know to be true. We do know there is a certain level of awareness that will drive product adoption. We do need to do announcements and things like that for people to be aware of what we provide; what our value proposition is.

They are not necessarily a transaction. You can't take these events and say, "Did I immediately get X, Y, or Z?" They're not necessarily performance marketing objectives. They're probably more longer-term relational metrics that we are looking at.

We do try to balance what is a more long-term needle we're trying to move versus what's a short-term thing we're trying to do. As an example, if you want to acquire more customers in a more performance marketing-oriented way, then the metrics are a lot more tied in and it's more metrics first versus if you're trying to build brand awareness, visibility, relationship, et cetera. It's a little bit more long-term.

Michael Krigsman: Do you use proxies? Do you look at data and say, "Okay, this data, if a customer is doing this, if they're leaving more comments or whatever it might be, that this means we're somehow touching them in an emotional way"?

Dutta Satadip: Absolutely. We definitely look at both qualitative and quantitative, the frequency of interactions, et cetera. We were born in the world of the Web and beyond, Web and mobile, so we also have a lot of metrics we look at from the product perspective. Our products are instrumented to understand how much engagement we have with our platforms and who is spending it.

The beauty about Pinterest is it's just not a passive platform. When people come here to discover stuff, they're looking at images, they're reading stuff, and they're actively organizing them by clicking on save and pinning them onto their boards. That's a very good signal for us to understand what's working and what's not working because we can, overall, not only see how much time they're spending on the platform, but we have a clearer intent that we are seeing on the platform through some of the actions that they are taking within the platform.

Michael Krigsman: You're trying to discern customer intent from what, from the actions they're taking? Is that the right way to describe it or put it?

Dutta Satadip: That is right. That is absolutely right. On the platform, actions like pinning it onto a board, how many boards they have, all of those things are directional indicators of engagement for us from the pinner perspective. Then, from the advertiser perspective, I always like to say money talks. If people love our outcomes of our campaigns, they spend more. If they don't like the outcomes, they shift budgets and they go elsewhere, right? We have a very clear sort of "voting" through a dollar system that kind of works on the advertising side.

When we look at what's happening on the pinner's side, when we look at what's happening on the advertising side, we can combine all of these perspectives. Like I said, I really wish it was as easy as putting everything onto a spreadsheet and just kind of sorting and picking the top two. The best we can do with data is we come to the top three, four, or five recommendations, and then we have discussions around what is the best outcome that we are trying to optimize for.

Michael Krigsman: We have an interesting question from Arsalan Khan on Twitter who asks, "How does or how should internal customer experiences affect external customer experience?"

Dutta Satadip: I think that's an amazing question. I think there's been so much research in the recent years that happy employees, engaged employees, the overall culture of empowerment within an organization leads to amazing experiences. If I take a little bit of a longitudinal view, for the longest time the focus of customer experience had unfortunately transitioned to how to optimize, do more calls, follow a script, and so on and so forth.

Unfortunately, we all have been on the other side of that call or the other side of that engagement or interaction where we've been asked to reboot our computer even though we know that is not the issue, right? I think the evolution is, if we empower our team members, they're often seeing the issues at the ground level. If we ask them, "What is it that is going to help us give the best customer experience?" 95% of the time they have a better, more practical idea than what I can come up with.

Part of our mission was, at least at Pinterest, and it was the same at Google, how do we engage and create that forum to suss out these ideas, engage our team members, and make sure that they are part of the solution? Creating that forum, we tend to get more, happier, engaged internal team members. Then, the discussion is not about, "I need to follow a script," or, "I just need to close the transaction," or, "I just need to answer this and move on forward." The question is, "How do I facilitate the best outcomes?" And, if I need to go out and do something extra, I go and do it.

It's the process of pure empowerment, creating forums as a leadership team to get those ideas, activate those ideas, and creating the culture where people can go above and beyond. I think those are the real dimensions of that internal mobilization that needs to happen to deliver amazing, not just experiences; experiences and outcomes.

Michael Krigsman: Yes, the focus on both experiences and the outcomes. We tend to get lost in the experiences without connecting the outcomes to that, and very often it seems.

When you think about customer experience, clearly the product is a core part of that. How do you break down the silos in order to ensure that the feedback from the customers makes its way back into the product in order to, again, improve that experience?

Dutta Satadip: That's a great question. I've been very blessed, I feel. Both of my recent companies, we've always had the overall values, mission of truly keeping our pinners first, or our users first. As a company, we not only believe it; we act on it. So, this makes my job so much easier because taking the feedback from what we are hearing, whether it is during a sales cycle, whether it's during objection handling, whether it's during actual running of a campaign, we have a team called Product Operations that is responsible for actually looking at all of this data, synthesizing across new opportunities and improvements in our products, truly partnering with engineering, and making sure that the biggest issues get solved.

I think it's always a balance around innovation on one side and improvement on the other side. As a growing platform, that balance is sometimes hard to accomplish because we want to obviously showcase the great innovation and sometimes improvement lags behind. But, by having the data from customers, having the data about what the market is asking for, I believe we are able to arrive at a better balance of innovation versus improvement, as we engage with our product teams.

Michael Krigsman: Speaking of innovation versus improvement, let me bring in efficiency here. How do you think about the balance between the efficiency of customer operations, which is to say saving money, versus the additional investment that's required in order to create a better customer experience? A lot of times those two are diametrically opposed when it comes to a customer service, especially.

Dutta Satadip: That's a great question. Having been in different roles in my past and having done a lot of customer-centric and operation centric work before I came to Pinterest, I think has evolved my personal thinking along the way. I think the traditional approach to a lot of problems around customer experiences, people can solve them. But, we are at a very interesting junction, I think. I think customers are expecting more and more personalization at scale.

Think about our experiences when we engage with a company like Amazon? We're not only getting a very personalized website; we're getting very fast transactions in terms of delivery of service. But also, when we go and reach out, we get extremely personalized outcomes in terms of how they're treating us. This concept of personalization at scale is something that we would see more and more of a trend moving forward. The great thing about being at a company like Pinterest is we're still building everything, so we have an opportunity to look at what's coming in front of us and rethink.

I think my philosophy has significantly evolved, and at least my first way of approaching this is, can we use technology to solve problems first? Let's take a simple thing, something that we are working on at this point in time. A lot of our questions tend to be very basic questions. Now, choice one could be to say, "Hey, we shift costs and look at an option like outsourcing." Cheaper but, effectively, we are shifting costs, right? The second option could be to invest in chatbots that are powered through machine learning that are building intent models using our help center and our previous questions answered and give a very different experience.

I think the approach that I am looking at in my evolution in this journey is, how do we solve the technology first? If there's no option with technology, how are we building the technology and the data to get there? And, if all else fails, then we truly look at adding people, adding processes, et cetera, and solving that. That's the big pivot that I think I have made through my career and journey, not just in customer operations, but looking at all the functions that I have been a part of.

Michael Krigsman: Where are we in the evolution of being able to use technology in order to provide those kinds of better customer experiences without having to add more people? Like with chatbots, sometimes it seems like it's still fairly primitive days.

Dutta Satadip: I think it is very primitive days. We are just getting started. But, the good thing is the underlying, the underpinning technology is there. And, like anything, there is a transition period and then it gets better.

I almost like to compare it to, like, not that long ago, before Siri was acquired by Apple, Siri was like an independent app and it did some very basic things. The evolution from that to the way Siri is in an iPhone or the Google assistant, Alexa, or Cortana, the evolution that we have seen in the past three, four years, I think, has been significantly augmented and different. I think we're going to see the same sort of uptick. A little bit of my role is to look into the future and make investments in a way that we are not loading it today in a way that we need to make completely drastic decisions tomorrow.

Michael Krigsman: What kind of investments are you making? I'm not trying to dig down into the deep secrets of Pinterest. Although, if you want to share the deep secrets of Pinterest, that's good too. But, what kind of investments are you making today? Again, how do you draw the balance?

Dutta Satadip: Absolutely. We are definitely evaluating. At Google, we were able to do a lot of this stuff. At Pinterest, we are also looking at really making sure that we have the right basis of data so that, as we evaluate and introduce technologies based on machine learning, et cetera, we are able to leverage and get the outcomes quickly.

One of my biggest learnings from my past is it's not the algorithms; it's the data that matters. Having a good foundation of data is what will ultimately power this. Just taking some sort of a chatbot solution and implementing right away will have, I think, some benefits, but not really significant benefits because, effectively, what we're trying to do is copying some version of the phone tree into a chat system.

What makes it more powerful is if we have all the related data and then we're building a true intend model based on how we have actually answered questions, how we have actually solved questions, what kind of artifacts we give to customers to get them started, onboard very easily, and then build that into the real-time interaction like a chatbot at this point in time. That's one of the areas that we are exploring. But, what we're really focused on is actually building that foundation of data at this point in time so that, when we start to introduce these technologies, it's not an incremental benefit but a significant step up.

Michael Krigsman: Can you be more specific about the kinds of data that you're trying to aggregate right now? I think that's really interesting.

Dutta Satadip: Absolutely. I think, when we look at data, I sort of see three big buckets. One, there is an operational view of the data. Think about all the transactions, interactions, et cetera.

At the end of the day, things need to be tied in. Everybody is collecting them in their own silos. It's very hard to join the space of data. That's the foundational element of what we want to do, and that's what we're doing right now.

The second, I think, is the customer view of the data. Okay, we are doing these activities. Is it leading to better outcomes? I always say, "Just because I showed up, is that actually good or this was actually going to happen regardless?" Right? We need to understand, through some statistical modeling, what correlation, what kind of services, what kind of interactions can be attributed to better outcomes? That's the second stage that we are working on.

The third stage is really sort of looking at a more business-centric view, which is more of a business decision-making. But, the combination of understanding what's happening in the operations to how our customers are spending, what our win rates are, what our churn rates are, that's the big connection that we are working on building at this point in time.

Michael Krigsman: That type of view enables you to connect the data that previously was isolated in silos. That seems like the magic bullet, ultimately.

Dutta Satadip: Yeah. I really wish it was a magic bullet, but that is, I think, the step in the right direction. [Laughter]

Michael Krigsman: But, it seems like, from what you're saying, that being able to connect data from these multiple perspectives ultimately is the way to really understand the customer and, therefore, build the kind of predictive models that you need in order to have machine learning enabled chatbots that become more than just fancy phone trees.

Dutta Satadip: Exactly. If you think about it, if we can give somebody a very helpful set of tips on what kind of creatives work at Pinterest, knowing that they are somebody who is a specialty retailer, it's a great insight. It's something that we can surface and give it to them right at the beginning versus having lots of conversations and then giving it to them downstream because that's going to have a meaningful impact. It's going to probably stop them from spinning their wheels, doing things that are not going to work. But, we need to unpack and understand what is it that specifically will drive those great outcomes for our customers.

Michael Krigsman: In a case like that where you're talking about content and aesthetic decisions, as well as financial outcomes and correlations or causations relating to, "Okay, yeah, do this type of ad and you get this number of clicks or results, click-throughs," how do you determine the outcomes, especially again on content and aesthetic decisions? How do you even approach that?

Dutta Satadip: The good thing about content is, especially with a platform like Pinterest, you have two very good measures/proxies. One is you click on it and you go somewhere. If it's working, we know somebody will click on it. The second one is, if somebody likes it, hopefully, they will pin it. Those are two very concrete actions in the system that we know, over a period of time, if we collect the data and correlate that, we will be able to give more concrete recommendations.

It's not like we don't have these recommendations. It's making these recommendations and making them extremely relevant when the customer needs it. It's combining; give them the right information, understanding where they are so that we can establish the right time to get that information. Those are the three pillars that we're trying to connect together at this point in time.

Michael Krigsman: You're always looking at the data in relation to the customer, the desired customer outcome: pinning it, higher ad spend, or clickthrough rates. Would it be safe to say that when you're talking about these aesthetic decisions, the machine doesn't understand aesthetics; the machine understands that if you do a thing a particular way that there's a higher likelihood that that item will get pinned, for example?

Dutta Satadip: Absolutely. Absolutely. I think you summarized it very well because machines can't really tell you if something is good or not good - not yet. I'm assuming that a technology is going to come at some point in time, but not yet. It can recognize patterns. It can tell you it's a banana or an apple, but it really can't talk about aesthetic attributes yet.

Then, part of it is, what is beautiful to me is going to be very different than what is beautiful to someone else. Beauty is in the eyes of the beholder, and that's what we are trying to suss out through data. Okay, if it is beautiful, what are you doing with it? That's either a click, a pin, or some kind of an engagement in the platform.

Michael Krigsman: I guess it raises the very philosophical question. Ultimately, can we isolate the qualities of aesthetics into a set of rules if we have enough data? But then, even if you do that, those rules ultimately are always outcome-based in terms of decisions and you leave out the unique, innovative, brilliant combination or invention that somebody might come up with aesthetically that is really great, but nobody ever thought of it before.

Dutta Satadip: I think, when people talk about artificial intelligence, they get this view of Arnold Schwarzenegger coming out and killing people in Terminator. Then the reality is, these tools and technologies are really more assistive. They can help us point in the right direction. They can help us suss out noise, patterns, et cetera. But, at least for now and what I know, it's definitely not at the stage where it can truly create rule-based systems and just automate everything out because aesthetics is so personal, and it is so context-driven. If you're an advertiser like Axe, you are probably okay with something a little bit edgier than probably a brand that's a little bit more conservative. Again, it's like all this dimensionality that is very hard to suss out and just put in an algorithm.

Michael Krigsman: But, regardless of the kind of brand you are or your aesthetics, if you're a brand on Pinterest, you want people to be pinning your materials.

Dutta Satadip: Absolutely.

Michael Krigsman: And that we know. [Laughter]

Dutta Satadip: That we know for sure. Exactly.

Michael Krigsman: Okay. Well, what a very fast little more than 45 minutes this has been. We have been speaking with Dutta Satadip, who is the global head of customer operations at Pinterest. Dutta, thank you so much for taking your valuable time in being with us here today.

Dutta Satadip: Thank you so much, Michael.

Michael Krigsman: Everybody, don't forget to subscribe on YouTube. That's important. We will see you soon. Check out CxOTalk.com for upcoming shows and to watch our amazing group of past shows as well. Have a great day.