LinkedIn · professional social network
Use AI to draft profile summaries, but sell the help, not the AI
What they did
- Stage
- scale
In their words
Value-Driven Growth: LinkedIn's Billion-Dollar Subscription Strategy — Ora Levit, LinkedIn
David BarnardI would imagine a lot of that's driven by machine learning, but I'd love to hear, because it's such a hot topic these days, but have generative LLMs been useful at all in that or is it primarily driven by more machine learning model, algorithms on the backend?
Ora LevitWe have both. We do use machine learning and AI quite extensively. In terms of LLMs, we offer value prop such as, "Let us help you edit your profile." For example, you want to write a summary, you didn't know how to write. A big thing members tell us is, "I have this writer's block. I look at, it's something, I don't know what to say about myself. How would you describe me?" And so we're like, "No problem. Here's three options of how we would describe you based on your profile data and where you work and you can edit it. We would like you to edit it to sound in your voice." We allow you to make it shorter, longer, more professional, less professional. There's different options, but that is an example of LLMs coming for help to solve a problem like that.
David BarnardSo you're using LLMs more on the consumer-facing side to help the customers and I'm still trying to sort out in my own mind all the different things LLMs are actually helpful for, but as this magic box you throw stuff into and get stuff out, I just was wondering if there was any specific things in personalization and otherwise that it is helpful for or if you are still primarily using machine learning for those kind of personalizations.
David Barnard… But I don't feel like I see AI, magic little AI icon. A lot of companies seem to be positioning it that way of like, "Work faster with AI." So maybe I just haven't seen it because I'm not in the segment where you're doing that, but I'm curious if you are specifically promoting it as AI or just as helpful tools. Because it sounds like maybe most of these things are just more positioned as, "Move faster, better product," versus like, "Magic, AI."
Ora LevitYou don't come to LinkedIn for AI and the magic of AI is a bit ubiquitous, right? What is AI really? And so you come to LinkedIn to keep in touch with your network, to gain professional knowledge, to understand how you can do your job better, keep in touch with the news and the trends and what's happening in your industry and in your local and we want to facilitate that. And AI helps us do that. And if you're a small business owner, it helps us give you the tools to grow your business. If you are a job seeker, it helps us help you find a job, but you come for that. You don't come for the AI. And so the goal is not to sprinkle AI everywhere and say, "We're here. AI is us." What does that really mean for a human, right? And so, "I'm here to first help you with what you're trying to do and AI is a way for me to do that when it's applicable and helpful."
David BarnardThat's such a great perspective and I think a great perspective for a lot of folks trying to figure out how to sell it, how to actually integrate it into their own products of keep that north star, the north star. I love that. People aren't coming to LinkedIn for AI. They're coming to LinkedIn for a very specific job to be done or range of jobs to be done and focus your AI efforts around that versus just sprinkling the magic everywhere and …
Get tactics like this every Monday
New tactics from the week's founder interviews, each linked to where it was said.
More from this episode
- Keep adding paid features, with a team as big as the growth team20+ new benefits in two years; guest credits it for acquisition and retention growth
- Keep the core free; charge for extras like rotating cover photos and a website linkNo numbers; guest says the free base feeds paid growth
- Let paying job seekers mark up to 3 jobs a month as their top choiceTop-choice applicants 46% more likely to hear back from the recruiter
- Judge tests on 5-year modeled revenue, not just first-year bookingsNo numbers; they turn down revenue that would hurt the funnel long term
- Run 1,000+ A/B tests a year, each product manager owning one areaGuest says testing beats internal debate; no specific numbers
- Ask people why they want Premium, then show plans and content that fitGuest says more tailoring always added value; no downside found
- Tell subscribers what their benefits are and how to use themMessaging changes alone improved retention; no numbers given
- Offer a two-person plan at 20% off, plus a big annual-plan discountBoth offers retain better; no retention figures given
- Show past subscribers their own messages and, sometimes, win-back offersReturning ex-subscribers grew into a large share of subscribers
- Offer a new free trial to past subscribers after 12 monthsStill in testing; no results shared
- Give subscribers free months of partner apps like Notion, Spotify and HeadspacePerk users tend to stay longer; no numbers given
Tags: ai-features, llm, ai-positioning