Using Machine Learning to Help Lattice Identify the End Market of Over 50,000 Potential Customers

hexpattern-2
hexpattern-2

CLIENT SUCCESS |

USING MACHINE LEARNING TO HELP LATTICE IDENTIFY THE END MARKET OF OVER 50,000 POTENTIAL CUSTOMERS

The Client

Lattice Semiconductor is a leading semiconductor manufacturer of programmable logic devices. Thousands of companies across the communications, commercial, and industrial markets rely on their technology.

To find out more about how InterVision’s machine learning and predictive analytics helped Lattice identify their customers, we reached out to Doug Hunter, Senior Director of Corporate Marketing at Lattice.

Here’s what he had to say about his experience.

The Vision and Challenge

Question: What would you say is the greatest challenge you face, as the head of the corporate marketing team at Lattice?

Hunter: We’re the company’s public megaphone. We need to make sure that potential customers are aware of Lattice, and, most importantly, that they come engage with us and work their way through the lead to revenue funnel.

So, the problem isn’t just getting more traffic—we also need to engage traffic, increase funnel yield, and get more people coming down through the pipe. That means talking to people on their own terms and serving them content they’ll find relevant.

Question: Serving personalized content can be challenging, especially when you have thousands of customers. How did you target leads before InterVision?

H: We had some data from our Marketo database. And, at a base level, we knew that if someone looks at our industrial webpage, they’re probably an industrial customer; if they look at our consumer page, they’re probably a consumer customer. But a lot of our products span multiple markets, and we didn’t have a strong, simple way of identifying our customers’ end markets.

Question: How many leads were you able to identify with this data?

H: We had identified the market segments of maybe one-third of our 77,000 active leads.

Question: Did you notice increased engagement when you began personalizing content for those leads?

H: Absolutely. The first thing we focused on was creating email newsletters personalized for three of our key markets: the communications market, the consumer market, and the industrial market.

When we served communications content to communications people, they engaged at higher rates versus communications content served to industrial people. The more relevant the content, the higher the open rates and click-through rates.

Question: Seeing that success, did you try to identify the remaining two-thirds of your potential customers using alternative methods before collaborating with us?

H: We tried a campaign where we reached out to leads with an email that basically said, ‘Help us better serve you, please tell us what your end market is.’ That, frankly, didn’t work. We only received a few hundred responses from tens of thousands of leads.

Question: So, what did you do?

H: The question became: how to we systematically identify people’s end market if we can’t get them to self-identify? That’s why we had the conversation with InterVision.

We needed a more advanced, more complex methodology for figuring out who these people were. We needed them to help us identify a larger amount of the population based on their behavior.

The Outcome

Question: Did you have any experience with machine learning before working with us?

H: I was very new to the machine learning field. I knew it was hot and could potentially help us, but I didn’t know how it worked. After talking with InterVision, we both realized that my end market identification project was a great fit with high ROI. InterVision did a great job teaching us what was possible and helping us spin up.

Question: Was the process smooth once you did start collaborating?

H: We actually went through a fairly protracted process to come up with a data set, but it wasn’t InterVision’s fault. Not all of the data was necessarily useful or indicative or sent strong signals, so getting a clean data set was challenging.

Question: Did you know what you were looking for in terms of useful data?

H: Yes. We went through an iterative process together where we’d look at the data, get cleaner data, and then have a conversation about what the data meant. We’d send them a data set, and then they’d walk us through what they’d observed and how we could improve it.

On their end, InterVision did whatever it took to get a stronger signal. They kept coming back with great observations on the data and suggestions for how we could collect more meaningful data going forward. We had three rounds of refining.

I really appreciated their collaborative approach.

Question: When you did finally have a solid data set, what was the next step?

H: InterVision created machine learning models. They tried many different types to see what would give us the best results, and then they came back to us and explained what they were seeing and what options we had moving forward.

The collaboration was great, and InterVision went above and beyond. They didn’t take some stock standard model and shove our data through it—they created many different models specifically for us, and they tested the quality of each one.

Question: Would you say you had a custom solution created just for you?

H: Absolutely. Yes, I absolutely felt like they heard us, understood our challenges and worked hard on solving our problem.

Question: After all the fine-tuning it went through, do you feel like the machine learning model you’re using now will help you identify customer market segments?

H: Yes. Let’s say we can get a one percent lift off of 10,000 people—that’s 100 people. An extra 100 people who will engage more deeply and convert.

The project’s not done yet, but what I’ve seen so far is promising and will put us in a better place.

Question: What impact has using machine learning solutions to predict the industry segments of your potential customers had so far?

H: We can now positively identify the market segments of two-thirds of our active leads—a 200% increase. That’s more than we were able to get from any other methodology, and InterVision got us there way faster than we could have on our own.

Question: What have you been able to do with these new insights?

H: We’ve been able to refine our targeted email campaigns, and email open rates are up 2% as a result. That’s significant when you have an open rate running at 10%—that’s potentially thousands of engaging customers.

We’re also working on tailoring the homepage experience. We’ve released different home pages for three of our key markets, and we’ve created a fourth page for visitors we haven’t been able to identify.

Our long-term vision is to be able to customize and personalize each customers’ journey throughout our entire website. We want to understand exactly where they are in the buying process, not just their end market, and give them clear calls to action that take them to the next step. 

Question: The end goal being to convert these leads into sales?

H: Exactly. Digital marketing is a numbers game. The more people I get into the funnel, the more I can target and personalize. The more I do that, the better our yields will be on the back end.

I have a fundamental belief that if you want people to convert, you need to need to present them with content that’s relevant to them. Machine learning helps us meet our customers where they’re at.

Question: So overall, working with us has been a very positive experience?

H: It’s been great. Two things, in particular, stand out. The first is the collaborative process. Machine learning is a very new field. Most people know that it’s hot right now, but very few people realize how many problems it can solve.

Plus, whenever you introduce new technology, it changes the way things are done. You need to overcome certain cultural and philosophical norms.

InterVision did a great job explaining how it worked and what they could bring to the party. They helped us design and execute implementation strategy. Overall, I appreciated their collaborative approach.

Question: And the second thing?

H: The second thing would be I felt like they went above and beyond. We didn’t get some cookie cutter product. They tried a lot of different models. They tested the quality of everything they did. They came back with observations. And they produced custom solutions for us that generated results we could actually use.

I wasn’t expecting to get the broad suite of results that I ended up getting from them.

 

NOTE: This client success story originally appeared before SeyVu was acquired by InterVision. All mentions of “SeyVu” have since been changed to “InterVision” to avoid confusion.