AI Scales Marketing Innovation at Palo Alto Networks

Jeff Mosier:

Hi, I'm Jeff Mosier. I'm the marketing content lead for the Infosys Knowledge Institute and I'm here with Nav.

Navneet Singh:

Hi, I'm Nav. I lead the marketing for network security and AI at Palo Alto Networks.

Jeff Mosier:

Tell us a little bit about where your company is specifically in marketing in AI. What is your maturity level? Where do you stand right now?

Navneet Singh:

I think we would be somewhere in the top few percentile of companies, I believe. That's not just marketing, although we focus more on marketing. When ChatGPT burst onto the scene in 2022, within a few months, just because our exec leadership is very AI forward, we organized leadership meetings which are VPs and above everybody in the room and we brought in people who were experts in AI. For example, people who had worked at IBM Watson, people who were working on Google Gemini, people who were startups focused on AI and they did panel discussions and discussions with us on what are the best use cases for AI. How are they using AI, what are the best practices, what are the things to avoid and we formed governing counsels and so on at the very beginning of 2023.

Today we have a sales copilot. We are doing experiments in marketing, we'll talk more about that, but there has been very AI forward and very optimistic about AI usage in our company.

Jeff Mosier:

How helpful was it to hear what other departments were doing with AI and how they were utilizing it? Did that influence marketing based on how their experiences?

Navneet Singh:

Very helpful. One, some of the departments were very data-driven. For example, IT was talking about how many answers based on the knowledge base were accurate? It gave us information about accuracy of AI in that time. We also heard about how sales is planning to use it in prospecting, et cetera and because marketing and sales work together anyway, it helped us figure out what should marketing do to help sales as they use AI.

Jeff Mosier:

Where have you found AI to provide the biggest value for marketing? Where are you getting the biggest bang for your buck?

Navneet Singh:

I would love to be able to say that ROI for content generation is X and ROI for marketing tech stack or optimization is Y. I don't have data to be able to say where it is the most useful. I think anecdotally, again, it might be different based on who you interview even within marketing at Palo Alto Networks. One thing that really struck me was localization. When we develop content, either a virtual event that we do or a new product launch or a campaign that we do globally, there is always a lag between that launch and when regions are able to consume it in local languages. We've been able to reduce that significantly, so that's one example that has been huge ROI.

Jeff Mosier:

You're confident that in the future you're going to be able to calculate that ROI a little bit better or you're going to have the metrics to understand what value you're getting out of it.

Navneet Singh:

I think so. Although I would say our focus is not as much on metrics and measurement of AI because we believe AI will be there and we have to use the most of AI. For example, I just mentioned the localization. It's just very obvious that instead of 21 days, it's like two days or that you're able to localize in all tier one languages instead of only the top three based on budget and agency constraints. Now we are experimenting with videos. The biggest ROI I see is that many times we work with either internal or external creative services and we present the video concepting to execs, for example and it does happen you get review comments and now what we are able to do with AI, we are able to incorporate those reviews very, very quickly overnight and produce multiple versions very quickly that can be reviewed.

Jeff Mosier:

Are there some areas where you found that AI has kind of disappointed or not lived up to what you had hoped it could do?

Navneet Singh:

Sometimes we just assume AI will do everything and I think when I was watching Tim Altman's interview, he said that it is no longer important who has the best answers because AI will give you all the right answers. What's really important this age is who is able to ask the right questions. I think that it is still very, very critical for us to figure out what is our strategy, what are we trying to do? What is the business strategy? That alignment with business strategy and then creating a marketing campaign and then doing a measurement of that is still critical. AI can accelerate all of that like translations, video creation campaign, but if upfront your messaging or business strategy is not aligned, then you can do everything correctly, but you can still be climbing the wrong hill.

Jeff Mosier:

Have you found many hurdles or barriers to kind of advancing what you want to do with AI?

Navneet Singh:

My team works a lot on product launches. The biggest barrier was being able to use AI in products that have not yet launched. We did not want to leak confidential data, right? Obviously that's difficult for us, so if I can't use AI for that where 80% of my team spends their time on, you can't write a press release for it, you can't write a blog post for it, you can't create videos for it, for products that are not yet launched. That was a big barrier, but now we do have that partnership with Google and we use their tools and we are able to use it for launches as well. That part is removed because our data is no longer used to train the model, so it is private to us. I think that we were able to take care of it very recently.

Jeff Mosier:

What are some other barriers that are still there that you've not been able to get past yet, but you might be able to?

Navneet Singh:

One of the issues that comes up with AI is just data being in different places. AI is only as good as the data it can act upon. If we want an end-to-end view of a campaign, how it performed from social, to press, media outreach to campaign, to get it content to pipeline generated and attribution, then it needs to act on data that is in one place or it needs to be able to access data in different places and I don't know for now whether that'll be possible. Either the standards emerge where all of these vendors have data in one place or they can allow AI to access data in different places or there is one data lake where maybe a platform vendor that can aggregate all of that data in one place.

Jeff Mosier:

Do you need more data than what you have to be able to do what you want to do with AI and marketing? Is there data out there that's just simply not available to you that you need?

Navneet Singh:

If we are able to just get the data that we already have. We have social impressions. We have the media impressions. We have the media outreach. We have the MQLs and SQLs and conversion metrics and we have the attribution to the accounts. Even if we are able to just get the data that we already have and get holistic answers on did this brand campaign affect the pipeline that was generated because it was only focused on brand. We made a decision to ungate more content. How did that affect us? Did it result in more pipeline? More recognition of our brand, our brand recall and how did it affect our MQLs, maybe three months down the line or six months down the line? I think we do have enough data, we just need to be able to ask the right questions and it should have access, the AI should have access to all this different disparate sources of data to be able to give us the answers. Then I'm sure there is some data that we don't have, but I feel that priority one is to get answers from the data you already have.

Jeff Mosier:

How are you addressing the issue of AI skills for marketing and making sure that everyone can use the tools, the capabilities that are coming in, sometimes by the week or by the month?

Navneet Singh:

Two ways really. One is formal learning. For that, we have courses that our CIO and our people team have created on AI and that email, for example, goes out from the CIO to say to everybody, "This is how you use Notebook LLM. This is how you use Google Gemini. These are some of the best practices." These are examples. That's formal learning courses and training and there is informal, best practice sharing. I was mentioning to you that we just did a video creation project using Google Flow. In this team meeting, which is today happening with my team, I've asked the team, the people who actually experiment with the Google Flow, to share best practices with others, to show the videos that they were able to create and what are the limitations? For example, it only creates eight second video and then you need to stitch it together and you need some basic iMovie skills or some other software that you can use to stitch those videos together.

Those kinds of things are informal best practice sharing, which varies from team to team. The good part is we can align those best practices sharing to the mechanism that best works for each team. I do team meetings, but then also have a Slack channel where people can share information and best practices.

Jeff Mosier:

Are there plans in the future to keep adjusting that and keep evolving how you're kind of helping the marketers learn and how to keep their skills up?

Navneet Singh:

Definitely. I think there will be a time when I'm sure in the formal learning there will be surveys on how we have been able to train people and how effective the training was and then I think there might also be a point where we bring in new talent. Our CEO recently talked about, in many cases, we bring in interns and teach them how to do things because they need to learn the corporate world.

Jeff Mosier:

What are some AI capabilities you wish you had that you don't yet?

Navneet Singh:

One is about the performance metrics and KPIs that I just mentioned, that I wish I was able to ask a question in natural language and an AI would just return an answer based on all the data. I see something very similar in terms of AI as well. I'm able to ask it any question about history or the plants and trees going in my backyard or how to talk to my teenage kid and it's able to give me the answers. I'm not able to do that about enterprise data and enterprise environments and enterprise campaigns, enterprise marketing, enterprise content. I wish that that will change and I hope it will change.

Jeff Mosier:

Do you think that that will happen soon because of where we are now? I think after the launch of ChatGPT, that kind of put AI at the top of everyone's agenda and it's AI above all, right? I think every company, even if they weren't that enthusiastic about it, had to at least make an attempt.

Navneet Singh:

I'm very optimistic that once AI has access to all of this data, that AI is so powerful and the capabilities are only growing, that we will be able to ask it questions in natural language and it'll be able to return answers. We might be able to even go a step further with AI agents to say, not only tell me the answer to this, but based on everything that you see, craft me on my next campaign for my next product launch, which is three months down the line and here's the basic ingredients, which is the PRD or product requirements document that has been created by engineering. Just use that and input and all the data about our best performing campaigns in the past, just create me a campaign for something that I'm going to launch in the next three months. I think that there is going to be not only AI answering our questions, but also with AI agents, then autonomously creating a campaign, executing a campaign, learning from the campaign, optimizing it and then giving me the results.

Jeff Mosier:

How far off do you think we are from the promise of Agentic AI? That's something that is still pretty... It's definitely not here yet, but it's starting to creep in there. There's starting to be use cases. How quickly do you think that arrives for marketing?

Navneet Singh:

I'm not sure. I am seeing a lot of promise. I haven't seen that actually in practice. I've seen in ads. I don't know, but based on the pace of change, I am guessing six to 12 months is my guess.

Jeff Mosier:

That's faster than I would've guessed. That's pretty bold.

Navneet Singh:

Yes. AI is changing so quickly and I was just giving you the example of the videos. If I show you some of the videos that we've been able to create, it's mind-boggling. They look real. It is no longer where you can actually make out this is AI generated video. It no longer looks like a deep fake. It is actual human person talking just like us and if you look at Notebook LM, I give it a document which is 30 pages long and it creates a podcast, two people talking to each other explaining what's in the document. Very, very natural. I think things are changing very, very rapidly and I think that in six to 12 months we will be at that stage where agents will do some work for us, maybe not all.

Jeff Mosier:

Are there any areas where you think AI is just not going to be capable of doing some of the things that people think it will be able to or where will it fall short, do you think?

Navneet Singh:

Yeah, so I don't know if it's about falling short. I think it's about really recognizing what AI can do for you. Going back to my earlier response of what is really important for us is to be able to ask the right questions and AI will not ask the questions on your behalf. You will have to ask the right question, which is, what is the business strategy? Is this product that I'm launching, should I focus more on price or value or time to market or something else? What is my competitive advantage? Where's the market headed? What are my goals with this launch? What am I measuring? Do I want to measure adoption or do I want to make money or do I want to do both? Those questions are really critical. AI will not ask you that question, will not ask the right questions. You have to ask the right questions and then align the marketing strategy and then say, AI, do this job for me and AI agents go do this autonomously.

Jeff Mosier:

Yeah. You need strategy, you need planning, you need guidelines, you need all of that to be able to do it. You can't just let it loose. It's not magic. It's a tool.

Navneet Singh:

We also actually did because in cybersecurity, we have seen some of our customers talk about why AI should have content guardrails. As an example, there are customers who are building AI applications like a human-like agent which will answer the questions. In many cases, the agent says, based on all the research that I've done, this competitor's product is actually better for the question that you're asking. You don't want that as a brand to refer a competitor's product, right? You need some kind of content guardrails. There are some other customers who've said, "I never want my AI agent to answer anything about the military or about politics." That's content guardrails, right? Customers are concerned about AI being just completely autonomous or just answering any question that the end user is asking, so there are some guardrails that customers want AI to have as their own AI to their end users.

Jeff Mosier:

That's going to be a lot more important whenever you have more autonomous AI doing a lot of different operations, not just one thing, but even agents directing other agents to do things. That increases the complexity of that, I think.

Navneet Singh:

When agents start talking to other agents, there may be a time when agents say that, "Why am I communicating in English?" It is a language that is meant for humans. Machines to machines, I should invent a new language, which may be a machine language much faster, much more efficient and if they start inventing their language and start communicating that language, then we don't know what they are talking about.

Jeff Mosier:

Well, talk about black box and AI, that's a different level altogether.

Navneet Singh:

Yes, yes. I think there are many things that are still unknown. We are at the very early stages of AI and definitely about AI agents. I think there is a lot to learn.

Jeff Mosier:

Well, thank you so much for your time. I really appreciate it. This was a lot of fun.

Navneet Singh:

Thank you.