Latest / Marketing Automation with Fexingo: HubSpot, Marketo, and Automated Customer Journeys / How Zendesk Built a Bot That Actually Sells
Transcript
- Lucas: When you hear 'marketing automation,' you probably think of HubSpot or Marketo. But there's a dark horse that's been quietly eating the e-commerce space for the last two years, and it started as a customer service tool. Luna: You're talking about Zendesk. Lucas: Exactly. Zendesk's answer bot, which they originally built to deflect support tickets, has evolved into a full-blown marketing automation engine. And the numbers are surprising. Luna: How so? Lucas: A mid-size apparel brand called 'Thread & Hour' — they do about twenty million in annual revenue — started using Zendesk's conversational bot to trigger product recommendations based on browsing behavior. Within sixty days, their average order value went up twelve percent. Luna: Twelve percent is big. But that's just a bot saying 'you might like this,' right? Lucas: It's more subtle than that. The bot doesn't just blast recommendations. It asks a qualifying question first. Like 'are you shopping for yourself or as a gift?' Then it tailors the product carousel based on the answer. That single tweak lifted conversion by twenty-three percent according to their internal a/b test. Luna: That's the trigger logic piece. So the bot is essentially a guided salesperson, not a billboard. Lucas: Right. And the key enabler is how deeply it integrates with Shopify. Zendesk can read real-time cart data, browse history, and even past support conversations. So if a customer once complained about sizing, the bot might say 'we now have extended sizes in that style.' Luna: That's personalization that doesn't feel creepy because it's contextual. Lucas: Exactly. The other surprising finding was about timing. Thread & Hour tested sending the bot prompt immediately versus waiting three minutes after a user landed on a product page. The three-minute delay produced a thirty percent higher click-through rate. Luna: So don't interrupt the browsing — let them engage first. Lucas: Exactly. That's a lesson a lot of email automation gets wrong. They fire off an abandon cart email within thirty minutes, but they don't consider that the customer might still be researching. The bot can wait for behavioral cues. Luna: What about the brands that can't afford a full Zendesk suite? Can they replicate any of this? Lucas: Partially. The trigger logic can be rebuilt with tools like ManyChat or Chatfuel for Shopify. But the key is the data integration. Zendesk's advantage is that it already has the support history. A standalone bot without that context won't be as effective. Luna: So the competitive moat is the CRM data, not the bot itself. Lucas: Precisely. And that's why legacy email-first platforms like Marketo have struggled to catch up. They're trying to bolt on chat features, but they don't have the same native service history. Luna: I've noticed that HubSpot's chatbot is improving, but it still feels like an add-on. Zendesk's bot feels like the core product now. Lucas: It's becoming the entry point. When a new customer signs up for Zendesk, the first thing they see is a setup wizard for the bot. Email automation is secondary. That's a deliberate product strategy. Luna: Let's talk about the actual journey map. What does a triggered conversation look like for a repeat buyer? Lucas: Thread & Hour's flow is worth walking through. A returning customer lands on the site. The bot recognizes them via a cookie or a logged-in session. It says 'welcome back, Sarah! Your favorite denim jacket is back in stock in your size.' Luna: That's triggered by a restock event, not just time-based. Lucas: Right. The restock event fires a webhook into Zendesk, which then triggers the bot to message that specific segment of customers who had that item in their wishlist or who previously bought it. The conversion rate on those messages is about forty percent. Luna: Forty percent is insane. Email restock alerts typically see ten to fifteen percent. Lucas: Exactly. The difference is the immediacy. The bot message appears while the customer is already browsing. They don't have to open a separate email. And the bot can upsell — 'we also have a matching belt now.' Luna: So the bot becomes a revenue driver, not just a support cost-saver. How does Zendesk price this compared to HubSpot? Lucas: Zendesk's Suite Team plan starts at sixty-nine dollars per agent per month, and the bot conversations are included. HubSpot's Marketing Hub Professional, which includes the chatbot, starts at eight hundred dollars per month for up to two thousand contacts. For a small e-commerce brand, Zendesk is significantly cheaper. Luna: But HubSpot includes email automation, CRM, and landing pages in that price. You'd need to add those separately with Zendesk. Lucas: That's true. The trade-off is depth of conversational features versus breadth of marketing tools. Zendesk is betting that the conversation becomes the primary channel. Luna: And for some verticals, that bet is paying off. I saw a case study from a furniture brand that used Zendesk to handle complex configuration questions — like 'will this sofa fit through my doorway?' The bot asked for dimensions and gave a yes/no answer. Lucas: That's a high-intent interaction. If you're asking about dimensions, you're ready to buy. The bot captured that intent and routed to a human agent only when necessary. They reduced support tickets by thirty percent and increased conversion by fifteen percent. Luna: So the bot isn't replacing the salesperson; it's qualifying leads more efficiently. Lucas: Exactly. And that's where the marketing automation industry is heading. The old model was 'send an email, wait for a click, then send another email.' The new model is 'have a conversation, adapt in real time, and close the loop with the CRM.' Luna: What about the risk of over-automation? If every visitor gets a bot message, doesn't that feel intrusive? Lucas: Thread & Hour learned that the hard way. They initially triggered the bot on every page load. Bounce rates went up. So they limited the bot to product pages and cart pages only. They also added a delay of at least ten seconds before the bot speaks. Luna: That makes sense. Give the user time to orient themselves. Lucas: And they allowed users to dismiss the bot permanently with a single click. That actually increased trust. The people who stayed engaged were more likely to convert. Luna: Let's zoom out. Does this mean that email automation is dead for e-commerce? Lucas: Not dead, but demoted. Email is still great for re-engagement campaigns, like 'we haven't seen you in three months.' But for the purchase moment, chat is winning. The data from Thread & Hour shows that chat-driven revenue now accounts for forty-five percent of their online sales. Luna: That's nearly half. So if you're a marketer building a stack for 2026, what's the takeaway? Lucas: Invest in conversational triggers that use real-time behavior. Don't just automate emails. Automate the conversation at the moment of intent. And make sure your bot has access to customer history — whether that's purchase data, support history, or browsing behavior. Luna: And if you're a small brand, start with a simple welcome bot that asks a qualifying question. You don't need the full Zendesk suite. Lucas: Exactly. The core principle is the same across tools: ask before you tell. That's what makes the bot feel helpful, not creepy. Luna: Alright, Lucas, so for our listeners who want to try this: what's the minimum viable setup? Lucas: A Shopify store, a chatbot tool that integrates with Shopify, and a spreadsheet to track your tests. Start with one trigger: cart abandonment after ten minutes. Add a single qualifying question. Measure the impact on average order value. That's it. Luna: And if you see a lift, then you layer in more triggers like restock or browse abandonment. Lucas: Right. Most brands overcomplicate it. They try to build a perfect journey map before they have any data. The winners iterate fast based on real conversations. Luna: So the future of marketing automation might not be more complex workflows — it might be smarter, simpler conversations. Lucas: I think so. And Zendesk just happens to be in the right place at the right time, with the right data. But the principle applies to any platform that can unify service and marketing data. Luna: Any other tools you're watching in this space? Lucas: Intercom has a similar play, but they're more expensive. Drift was early but got acquired. The dark horse I'd keep an eye on is Tidio — they're targeting the SMB market with a freemium model and deep Shopify integration. Luna: Alright, so the lesson: don't sleep on your customer service tool. It might be your next marketing automation engine. Lucas: Exactly. Thanks, Luna. Luna: Thanks, Lucas.