Oh shift. You're listening to the O shift podcast.
When something big like AI is coming, we need to reset. We need to reset culture. We need to reset strategy. We need to reset how we build products. Authenticity in general in today's world is like a superpower. I feel that not many people are really authentic. And when you think about agents and LLM, how do you get an agent to trust a brand? We are lucky to lead the biggest transformation of our lifetime.
Welcome to the Oshift podcast where AI meets SaaS leaders. The show is about how real software and e-commerce leaders are making the shift to AI and the ups and downs in that process. I am pleased to be joined here with a legend in e-commerce, Tomer Tagrin, co-founder of Yotpo. The platform is obviously known for UGC reviews, loyalty and more that power growth for direct-to-consumer and omni-channel brands. We hope to get under the covers. First of all, thanks for joining us here at our offices.
Thank you for having us. By the way, big fan of your voice and your important voice in e-com. So thank you, very excited to be here.
Yeah, excited to have you here. I think from anyone on the outside not in the Yotpo day-to-day, things must seem normal for you to change so much so quickly. From the outside we see your posts, you're like we're resetting everything. It almost goes back to the pandemic for Tobi who's like we're going to delete our roadmap, we're going to reset everything. Does it feel that way to you in the last couple of years? How do you think about it?
Probably for Tobi everything is planned. He has his crystal ball. So definitely not Tobi. But yes, I think that's the entrepreneur side of me. I think definitely when something big like AI is coming, the only way to make sure that companies like Yotpo that were pre-AI, we need to reset. We need to reset culture. We need to reset strategy. We need to reset how we build products. And that's what we've been trying to do. I think life is too short not to try to go after opportunities that you strongly believe can make a difference. That also means having a lot of humility and saying where we were wrong and where we need to adapt and where we need to change in order to be again a company of tomorrow. So definitely a hell of a ride. Not everything is planned, doing a lot of mistakes, doing some things right. But I can tell you, Yotpo, at least as long as I'm here and since I've been here, will be a company that will always be willing to change versus reality in the future.
Yeah. No, it's interesting. I like how you laid it out where it's like culture, strategy, how you build product. I'm assuming go-to-market as well. I think that's a great format for us. Let's take those one by one. Culture, pre-AI culture, post-AI culture, is it different? Is it the same? I imagine there's some.
Yeah, there is. I don't think AI changes that people are people. People problems are usually the toughest problem. I do think in the companies of the future you need less people. You have much more automation. We now have teams that are building agents from go-to-market to engineering. The job of an engineer is changing dramatically. In culture for us, Yotpo, and again very openly, was a company that got too big. There was COVID, the zero interest rate. I think coming back to the offices, coming back to more startup mode or founder mode as some call it, has been really crucial to transform. The biggest transformation is culture and how we build products because that matters the most. It also means I need to change, our leadership needs to change. It's a lot, but that's what gets me excited: building a culture that is humble enough to understand where we made mistakes and have the courage to actually make the changes, not just cruise very slowly because we're already a big company.
You're like a big small company, which is actually a dangerous place to be.
100%. If you look, there are bigger companies in the world. Yes. 100%. I think we are lucky, as I told my leadership, to lead a business in the biggest transformation of our lifetime. No one knows, no one has a playbook. We just need to be humble, as close to the customers and as close to the technology as we can.
In terms of the customers, you built some new products, some of them you're spinning out into new ones, some you gave to partners. The strategy has shifted, not just because of AI, but post-COVID. I can't think of a SaaS company that didn't do multiple rounds of layoffs post-COVID. The strategy is a little bit different, being leaner, more focused. Is that part of it?
Definitely. We weren't able to execute on multiple products. We had email and SMS that we decided to sell to partners. There were a few forces happening simultaneously. One, we couldn't execute at the level we wanted, and that affected our reviews and loyalty business, which is the core where Yotpo is truly best-in-class. Second force is AI. We really think AI is changing reviews. You're going to see us continually developing more capabilities because the way people write reviews, the way they consume reviews, the importance of reviews in an AI-first world where everything is fake. What does authenticity mean? In loyalty, we've seen a lot of success from bigger merchants going all-in on omni-channel. With tariffs being introduced, loyalty is becoming more top of mind. Lastly, AI and agent commerce. Users are already doing discovery on LLMs. With recent ChatGPT and Google's protocol, it's insane. I don't know if it's going to be 30%, but definitely it's not going to be 0.5% of commerce moving to LLMs. That's a massive opportunity and challenge for brands. We found that our return on investment can be much more impactful on new markets and opportunities versus trying to innovate in existing markets where we weren't successful.
You said a key word: authenticity. From Yotpo admitting that this is where we're going to succeed, this is where we're not. How did you come to those decisions?
I had a few health issues. When I came back to the company, it was very clear that we need to change. I changed as a human being and said we're stopping. That was almost three years ago when we started what we call Yotpo 2.0, a shift in everything. Part of that was being more founder-led on social, which I'm still not comfortable doing, but you're doing a great job. I think it's important to put people out there because we are people trying to do our best. Authenticity for us as a company has always been uber important. Authenticity in today's world is a superpower. I feel not many people are really authentic. We try to be the most authentic company we can, where we get things right and where we get things wrong. In our product reviews, it's about how we bring authenticity. When you think about agents and LLMs, how do you get an agent to trust a brand? If I say my brand has the best running shoes for men over 50 with flat feet, why should they believe me? Everyone says the same thing. One thing that doesn't lie is what customers are saying at scale. So we ask how we can leverage GenAI and LLM to increase authenticity, write more quality content, and distribute it, but always make sure the content is authenticated.
I think it makes sense. There are so many people underestimating this as the next gold rush. Even if it's 25% of traffic, how many people are trying to spam this? Endless.
Billions of people. Every company has an army of agents trying to do that. We spend a lot of time thinking about how to help brands be unique and authentic. There are two worries on LLMs: how to be discoverable and how to control what the LLM says about you. That requires context and trust. That's the mental framework we're using to help brands with both context and trust.
Context and trust are two super important words. They're also important for the transformation of Yotpo itself. When you came back and reset the direction, what did you tell your team about the trust factor? Like, we're going to make it.
I don't think I said we're going to make it. I said, look, we did a deep process of analysis, saying where we got things wrong. We collectively agreed on them. It wasn't easy discussions. Then we said, if that's the case, we need to change, and this is our new plan. For certain people, it was too much, and that's fine. Part of that process was saying goodbye to many people who had been with us a long time. Trust is built when you're truly vulnerable and authentic. I think Yotpo was a company that grew too fast, chasing short-term growth and compromising long-term growth. We wanted to turn it around and say we'd rather compromise short-term growth for long-term growth. We want to build a profitable, growing business. That takes time and being authentic about it. Once people connected to that and saw we truly understand the challenges and are willing to change, it worked. For example, on product, I read Marty Cagan's book 'Transformed' and couldn't put it down. It was like he was writing about Yotpo. I spammed him, got on the phone, and he became our coach. He's coming to Israel to do a workshop. We're flying everyone from across the world to transform how we build products. We're bringing outside help from the best in the world. If we become the best product company, that's it. I don't know what the business outcome will be, but I'm sure if we do that, all the ducks will align. Tobi and Shopify are a great example of that.
That's a great transition into the next pillar: how do you build product? Let's talk about dimensions from user feedback to requirements to prototyping to shipping to iteration. Where does AI cut across those areas?
AI can speed up certain things and not others. Code generation: writing code today is not a job anymore. Agents can do that. An engineer or architect now—that's a big statement, but I don't think writing code will be a job if that's your entire role. You'll need great PLDs and just write the code, but there won't be much room. The job of an engineer or architect will be like a squad. The designer writes the best design and design system. The PM writes the best PLD saying what the product should do and the value to the customer. The engineer or architect says what's the best architecture and design, then lets agents build 80-90% and do code reviews. The craftsmanship of product is changing. We're heavy users of Cursor. The amount of code in production written by AI is significant. If a designer can finish with Figma and build directly into the design system and front-end code, or if you want to analyze customer feedback at scale, it's really easy. AI won't replace you; people leveraging AI will replace you. You need to know how to use the technology. Prototyping is the easiest use case. The question is how to do user research and ensure you're building something that adds value. There's a great book, 'Good Strategy Bad Strategy,' which says in a world where AI software is easy to build, you need to keep building, fall in love with the problem again and again. For example, in loyalty, the big question is incrementality: did you buy because of loyalty? Are you incenting the wrong behavior? That's a tough problem. We're working on analytics to show true incrementality. In reviews, it's different. Getting the most reviews is valuable, but in an agent-first world, 'great product' or 'loved it' is not so useful. We go deep on generating quality content and knowing what quality is. We also have a team working on a new product geared towards agents driving commerce, which I'm very excited about. It will launch in the next few months.
That's super exciting. If you think about the two pillars of Yotpo, reviews and loyalty, one is very easy to get into an LLM and represent. The other seems much harder because it's tied up in why someone is a fan of the brand and what keeps them coming back.
Yes. We try to leverage our data. We have 350-400 million loyalty members. AI can give recommendations based on our customer base. Loyalty is a lot of experimentation, trial and error. It's a breathing organism, not set it and forget it. We're trying to take it to an agent-first world and get to one-to-one loyalty, which may be the end state, but we're not there yet.
Diving into your tools, you mentioned Cursor is a big one. I'm guessing OpenAI is another. What are some tools you love that maybe everyone has, or if you're in the know, you know them all?
