Latest / Marketing Automation with Fexingo: HubSpot, Marketo, and Automated Customer Journeys / Why Marketing Automation Ignores Customer Voice Assistant Searches
Transcript
- Lucas: So here's a stat that stopped me this week: over 40 percent of U.S. adults now use voice search at least once a day. And yet, most marketing automation platforms treat every search query as if it was typed. Luna: Right — same text field, same trigger rules, same response. But a spoken query is fundamentally different from a typed one. Lucas: Exactly. When I type 'best noise-cancelling headphones under $200', I'm probably early in research. But if I ask my smart speaker 'Hey, what are the best noise-cancelling headphones under $200?' — there's a good chance I'm ready to buy right now. The conversational phrasing signals higher intent. Luna: And the platforms just flatten that distinction. HubSpot, Marketo, Pardot — they capture the keyword string, but they don't tag whether it came from voice or text. Lucas: That's the core problem. The channel itself carries meaning. A voice query is usually longer, more natural, and often includes words like 'where', 'how', or 'what's the best'. Research from Google shows voice queries are 30 percent more likely to be action-oriented. But if your automation doesn't preserve that signal, you're optimizing for the wrong intent. Luna: I remember a case from last year — a home goods retailer that started tagging voice search clicks separately. Their standard email flow for 'kitchen mixer' was a generic recipe roundup. But for voice searchers, they switched to a 'buy now' discount email. Click-through rates jumped 22 percent. Lucas: That's exactly the kind of win that's hiding in plain sight. And it's not just about intent — it's about context. Voice search often happens in the kitchen, the car, or while multitasking. That implies a different state of mind. A person typing 'roof repair' at 2 a.m. on a desktop is probably panicking about a leak. Someone asking their phone 'find a roofer near me' at 10 a.m. on a Saturday? That's planned maintenance. Luna: So the automation should treat those two signals differently. But most platforms don't ingest the device type or the query source at the trigger level. They just see 'search keyword: roof repair' and fire the same email sequence. Lucas: And that's a missed opportunity that costs real money. I looked at a few large Marketo instances recently — less than 5 percent of them had any custom field for voice versus text. The data is available in analytics tools like Google Analytics 4, but it never makes it into the marketing automation database. Luna: Why is that? Is it a technical limitation or just an oversight? Lucas: I think it's mostly oversight — plus a bit of inertia. Platforms like HubSpot allow custom properties. You could create a dropdown called 'Search Source' with options 'Voice' and 'Text'. Then you'd pass that via UTM parameters or from your CRM. It's not hard, but it requires someone to think about it. And right now, the industry is still treating voice as a novelty, not a core channel. Luna: But the numbers say otherwise. Voice commerce was estimated at $40 billion in 2024 in the US alone. That's real transactions, not just queries. Lucas: And it's growing. Amazon's Alexa, Google Assistant, Apple's Siri — they're all getting better at transactions. So if you're a retailer and your automated journey treats a voice buyer the same as a browser buyer, you're leaving money on the table. Let's talk about one practical fix: intent-based trigger branching. Luna: Okay, walk me through that. Lucas: Simple. In your automation platform, you create a decision node that checks a custom field — say 'query_channel'. If it's 'voice', route to a shorter, purchase-oriented flow. If it's 'text', send the educational nurture. That's literally a conditional branch. It takes ten minutes to set up once the data is flowing. Luna: But the data has to flow first. How do you capture that at the point of search? Lucas: There are a few ways. If you're using a chatbot or a voice assistant on your own site, you control the payload. For external search, you can use UTM parameters with values like 'utm_source=google_voice' if you're running voice-optimized ad campaigns. Or you use Google's reported queries data if you're passing it into a data warehouse. It's not seamless, but it's doable. Luna: And then you have to consider privacy. Voice data feels more intimate. People might not want brands knowing they asked about 'hemorrhoid cream' out loud. Lucas: That's a real concern. And I'd argue you should only use voice data with explicit opt-in. If someone's interacting with your skill or action on a smart speaker, they've already consented to the terms. But for passive tracking? That's a different story. The best practice is to anonymize the audio and only keep the transcribed text. And never, ever store raw voice recordings in your marketing database. Luna: So the brands that do this well are transparent. They say 'we noticed you asked about this via voice — here's a helpful next step'. Not 'we heard you talking'. Lucas: Exactly. Frame it as a service, not surveillance. And if you get it right, the customer feels understood. That's the holy grail of marketing automation — relevance that doesn't feel creepy. Luna: Speaking of feeling understood — one thing I've noticed is that voice queries often include emotional cues. Like, someone might say 'I'm so frustrated with my slow internet'. A text search is just 'slow internet fix'. Lucas: That's a brilliant point. The natural language processing layer that powers voice assistants can extract sentiment. If your automation could ingest that sentiment score, you could route frustrated customers to a retention offer immediately. No forms, no waiting. Just a timely 'We hear you — here's a discount on your next month'. Luna: I've seen a few companies experimenting with that — telecoms, mostly. They use a tool like Dialogflow to analyze voice transcripts and then push a sentiment label into Salesforce. Then Marketing Cloud triggers an email or SMS based on that label. Lucas: Right. And Dialogflow can integrate with most major platforms via APIs. So the architecture is there. It's just a question of prioritization. Most marketing teams are still drowning in basic segmentation — they're not thinking about voice as a distinct signal. Luna: But the early movers are going to have an advantage. As voice search becomes more common, the gap between 'voice-aware' automation and 'voice-blind' automation will widen. Lucas: Totally. And I think we'll see platforms start to build this in natively within the next two years. HubSpot recently acquired a small conversational AI firm — I'd bet they're working on voice intent tagging as a standard field. Marketo's parent company Adobe has the AI infrastructure to do it. It's just a matter of when. Luna: Let's pause on that prediction for a second. If today's conversation gave you a new way to think about your own automation triggers — maybe a practical idea you can test — that's exactly the kind of value we try to deliver on this show. And the reason we can keep doing it without ads is listener support. If this episode sparked something useful, you can help us keep going at buy me a coffee dot com slash fexingo. No pressure, just a simple way to say 'this matters'. Lucas: Yeah, that's exactly right. A few listeners chipping in makes a huge difference — keeps everything independent and focused on the details that actually help you build better automations. Luna: Alright, back to the future of voice. Where do you see the biggest opportunity in the next year? Lucas: I think the low-hanging fruit is in post-purchase flows. After someone buys via voice, send a follow-up that acknowledges the channel. Something like 'Hope you love your new headphones — since you asked about them on your smart speaker, here's a setup guide optimized for voice commands.' That's a tiny personalization that creates a huge brand impression. Luna: And it builds a feedback loop — the customer feels like the brand understands their context. That kind of experience drives repeat purchases. Lucas: Exactly. And it's all based on one simple tweak: treat voice as its own channel in your automation. Not rocket science, just intentional design. Luna: Alright, Lucas — good topic. I think our listeners have a concrete action item today: check if your CRM captures the search source. If not, it's time to add that field. Lucas: Do it. And test a branch. See if voice-sourced customers behave differently. I'm betting they do.