David Ly Khim on content automation in B2B and growing People.ai

Back in 2015, I was working at an agency and got a job at HubSpot, and I’ve always had this idea that I wanted to start my own business. I didn’t know what kind of business, but I learned that I could get paid writing articles. And eventually that turned into consulting for marketing campaigns. But that was not scalable.

I was selling my time, and it was a lot to manage because my role at HubSpot was also growing. That’s when I met my friend Alex, who also worked at HubSpot. And it was one of those things where you meet and you just get along right away. Eventually we were both at the Growth Marketing Conference that Vasil Azarov was hosting. We met someone who told us how much they were paying an agency for content marketing and that’s when we figured we could do it better. We have backgrounds in SEO and content. We work at HubSpot. We know the playbook. HubSpot teaches everything. But most people don’t know how to put it into practice. And that’s when the idea of partnering up came together. We do what people in B2B software companies need help with, which is content strategy.

We got a couple clients to start, and then eventually we brought on Allie as our Head of Content, and she’s a kick ass content person. She is one of the smartest people that we’ve ever worked with. She started building out an editorial process to scale up how we build content strategy and deliver high quality content to clients.

The joke is, you start an agency because you’re a masochist. But we have a couple of principles. The first is, “this should be fun.” We’re building a business on the side in addition to our demanding day jobs. We should enjoy working with our clients. We should enjoy working with each other. We should enjoy the work and the process of building a company. And if at some point we don’t, we should ask ourselves why that’s the case.

We call it the content maturity model. So if HubSpot were restarting today as a new business, the strategy that they used 15 years ago wouldn’t work because the martech industry is such a saturated market. When you look at basic terms like “what is content marketing” or “how to do SEO” or “how to do email marketing”, there are dozens or hundreds of companies writing the same thing.

Whenever someone wants to rank for a keyword, they go to Google, look what ranks for that keyword and write something similar and maybe make it a little longer, maybe a little better. But it’s not differentiated. The big companies that have high domain authority and have been in the game for a while can write anything and start ranking position one, two or three the next day. That’s what happens at HubSpot. They have the benefit of all the work they did for 15 years.

For smaller startups without the same domain authority, the goal is to put out more thought leadership content. One of the trends with companies big and small is moving more towards thought leadership, and you can combine it with an SEO strategy.

We call it the SEO sandwich. You look for keywords that have search volume, then write content based on your unique point of view on the industry and your products. And then at the end do some optimization, but the SEO part doesn’t dictate what you write. If you hear a company saying something different than what everyone says and it resonates with you, you’re probably going to remember and work with that company. This technique requires small companies to put more time and resources into high quality content that stands out, not just churning out cheap content.

We actually don’t sell them on that. If we feel like we need to prove the value of content, then we’re not going to work with that person because they’re probably months or years away from understanding how content marketing works. We talk to people who say: “we’ve tried to do content marketing for a year now and are not seeing any results; but we know that it’s what we’re supposed to do.” Or someone who has done content marketing programs in the past and came to a new company and wants to set it up again.

We know that there are a lot of people just getting started with content, but those are people that we would have to spend a lot of time trying to convince to do content and wait six months to see results.

For folks we work with who want to see results more quickly, we recommend starting with middle and bottom of the funnel content with low-to-moderate search volume. Going directly for keywords with the most traffic and high competition doesn’t work short term because of how competitive it is.

Interestingly enough we are working with Jasper.ai, a company moving towards writing entire books with AI. It helps us get basic content ideas and actually achieves impressive results. I believe there’s a time and place for that sort of tool. There will definitely always be a space for high quality content which humans will have to write.

If you want to create basic content, like the cheaper SEO content that’s written in large volumes, automation helps. Especially if you’re a team of one and you need to get content out fast. Now, in terms of thinking about content with a unique thought leadership and a voice, AI is not going to be able to write that.

If we talk about content process automation, that’s a whole different thing. We just set up some jobs that turn a couple rows in a spreadsheet into a content brief in a second. Whereas before it would take 30 to 45 minutes to create that same brief. So when we start talking about automation – it is mostly helpful in automating the process of content creation.

Tommy Walker, who used to edit at Shopify and Intuit, has a whole Content Operations course where he shows all the automations he’s doing using Airtable – it’s a really interesting time for content marketers and program managers for content right now because so much of it can be automated. You still need the human touch to it, but it’s much less manual.

For instance with Airtable you can flip a switch so that when you assign a document to review, the client automatically receives the notification that it’s ready to review. And there’s even automations to go from a Google Doc to publish in WordPress and insert images magically. If you have websites or have done marketing, uploading a piece of content from a Google Doc to the CMS and formatting and uploading everything takes way too much time – that’s one of the big wins in content operations.

I can share two experiments. The first one was about using social proof to improve the conversion rate on the sign up flow. We tested three different types of social proof so it was a control and three different variants. One variant had G2 badges, the second variant had a testimonial from Shopify; the third had actual reviews with stars.

The results we got were unexpected: none of the social proof had any improvement on the conversion rate of sign up. We thought this went against the foundations of social psychology that social proof improves trust. So we took a look at the sign up flow and realised the code hadn’t been touched in two years.

It also turned out there were two parts in the sign-up flow and the second part sometimes didn’t load, so people couldn’t finish signing up.

And sometimes the signup flow took 10 seconds to load. We realized we jumped to conclusions about hacky solutions and went back to traditional product work.The real problem with the sign up flow was that it just didn’t work. It’s a perfect example of how thinking about short term tactics has to be balanced with broader picture to understand the underlying causes for specific behaviours on your website.

Another one at HubSpot. The CMS solution started at $300 a month, which is really expensive. You can also get WordPress set up for $3 a month. Not surprisingly, many people using WordPress were implementing HubSpot onto their website. And they were actually pretty valuable, with a higher retention rate, and timely payments – even though for small amounts. Based on this data I proposed to rebuild our WordPress plugin.

There was a bit of pushback internally but eventually we got it done and within a couple of months we saw a jump in the amount of users. This allowed me to get more resources to keep adding features to the plugin and turn it into a new acquisition channel. By the time I left, we were finding significant traction through the plugin.

It was an experiment in the sense that we had a hypothesis. We validated it was worth working on it through small steps. And at some point we decided, okay, this is worth spinning up a whole new team.

We help companies with opportunity and account management. That’s all powered by a smart data layer. The way to think about it is, if you’re a sales person selling to enterprise companies, you have a long sale process. A lot happens in those 12-month long deal cycles. People might change roles or leave the company. So you need to keep track of the evolution of an account – who is involved in the deal, who influences it, who’s a champion, who’s the decision maker.

The way the People.ai data layer helps sales teams is it keeps all that information up to date. Say maybe the decision maker in a deal changes. What it does is it updates our people graph to say this person has now moved to this company. And it also informs other people’s Salesforce instances as well. It allows the sales team to have more inputs and be more accurate in the people they are targeting.

So if the sales team is working on five different opportunities, they can actually see the health of each opportunity and who they need to engage with more, who they need to be calling or contacting or who they haven’t gotten in touch with over the last couple of weeks. And it also creates benchmarks showing that most deals that close have these specific attributes. It may have an average of seven calls, ten emails, five meetings. And so it becomes more of a scorecard to help increase the likelihood of closing your deals.

And lately we’ve been working on a new product called PeopleGlass. It’s a spreadsheet-like app to update Salesforce records, so sales teams can easily automate the boring stuff. We’re building it out with a bottoms-up motion so people can sign up and start using the product and, if they’re interested, they can purchase our other products later on.

That’s a good question. And I hate to admit it, but guides, white papers, events, these things still work. I think as people who have been marketing or marketing for a while, we can think this stuff is so boring and say “Who reads Ebooks or PDFs anymore?” But people still do it.

During the last years we have seen the consumerization of enterprise technology. HubSpot is clearly a B2B software company, but the way they go to market looks very B2C. The ads that they’re running and the scale at which they’re getting people to their website and using the product is different from what happened in the past. And same thing for companies like Trello.

I think it’s normal now for these enterprise companies to have a free product or a free trial, because all those webinars, the ebooks, the lead Gen forms – their goal is to get people to the product. So why not just let them get into your product for free instead of having this funnel where they go to your website, download an ebook, maybe sign up for your product, and realize they don’t like your product later on. The freemium strategy really changed how companies are attracting enterprise customers.

The way I think about it is on a bell curve. There’s a group of us that are thinking about what’s the next thing, what’s the future, what’s the new technology and automation. And there’s this whole group of people that still aren’t as familiar or savvy with the technology yet and how it works. So they’re going to get there. But it’s a large group of people. One thing we learned at HubSpot is when we were building our new CRM, for instance, we thought we wanted to take people away from Zoho or Pipedrive.

That and we realized, wait, 80% of businesses are not on a CRM. We should go after them instead because they’re using spreadsheets and post it notes and notebooks. We need to go after that market first.

There isn’t one tactic that works for everything, but what we found is very personalised, persistent outreach is still your best shot. The best salespeople that I’ve met are the ones that will keep emailing you and contacting you on different platforms, whether it’s Twitter, email or Instagram until they get a response. Maybe it’s a no, but at least they get a response and also provide value. So it’s not just, hey, get on a call with me. Have you read this? I found this thing that might be helpful and also inviting them to things that will help them also continue to learn.

One thing that we’ve learned is posting little group events. Maybe it’s over Zoom, where three different leaders from other companies meet, and it’s a round table for them to share learning or for us to teach something, and that helps us become a trusted advisor to them as well. And it’s one of the reasons we do the podcast in the long term, maybe some of these people we speak to become clients, but in the short term, it’s more of a conversation where we learn from them, maybe from us.

But what we now know is we’ve built your relationship, they know we’re smart, we know they’re smart, we trust each other, and we can have a line of communication going as well. So it’s playing the Long Game. It’s not trying to jump on a call and sell them immediately. It’s building a relationship because your net worth is your network, and relationships are what end up leading to sales in the future.

So on a piece of growing traffic, that’s just a long term thing you got to commit to, producing content and driving people to your website. On running experiments to significance? This is something I go back and forth on. People talk about statistical significance, and I say, so long as you’re not curing cancer, you can use a much smaller sample and not need 99% significance. Maybe it’s 80%. Who cares? But if you have the data that says that’s something worth moving forward on, trust it.

Most decisions are reversible anyway. So if you make the wrong decision or your data is false, guess what, change it back, revert it with a click. You’re not doing brain surgery. You’re changing something on a website or the product. The worst case scenario I warn people about is relying too much on the data or on a process and not making a decision.

That’s when you get stuck and things start going badly.