Latest / Marketing Automation with Fexingo: HubSpot, Marketo, and Automated Customer Journeys / How Marketing Automation Workflows Miss Voice of Customer Data
Transcript
- Lucas: If your marketing automation platform has ever sent a 'We miss you' email to someone who just filed a support ticket about a broken feature, you already know the problem I want to talk about today. Luna: Oh I have been on the receiving end of that. It's infuriating. The company clearly has my data — they just aren't connecting the dots. Lucas: That's exactly it. Most marketing automation setups are built around behavioral signals — page views, email opens, form fills. But they're completely blind to what customers are actually saying. voice of customer data — survey responses, support ticket sentiment, call transcripts — it's like a parallel universe that never touches your workflow triggers. Luna: So the automation is optimized for engagement, not for understanding. That feels like a huge blind spot. Lucas: It is. And the thing is, the data is already there. Most companies run Net Promoter Score surveys or have a CRM full of support cases. They just never pipe that intelligence back into the marketing automation engine. Luna: Before we go deeper — and this connects — I love that this show is listener-supported and ad-free. It means we can talk about real gaps without worrying about a sponsor's platform taking offense. A handful of folks chip in monthly at buy me a coffee dot com slash fexingo, and that's what makes these honest conversations possible. Lucas: Exactly. No vendor agenda, no sponsored segment. Just what actually works. And speaking of what works — let me give you a concrete example. I spoke with the marketing ops lead at a B2B SaaS company last month. They had a standard lead nurture flow: download a whitepaper, get three emails, then a sales call. Classic. Luna: Right. And what did they find? Lucas: They started tagging support tickets with a simple sentiment score — positive, neutral, negative — using a basic natural language processing tool. Then they built a workflow in Marketo that triggered when any existing customer submitted a negative-sentiment ticket. The workflow would automatically send a different email series: one that acknowledged the issue and offered a direct support escalation, not a generic upsell. Luna: That's smart. Instead of blasting them with 'Have you tried our premium plan?' you're actually responding to their frustration. Lucas: Exactly. They saw a 40 percent reduction in support ticket escalations within that segment over two months. Because the customer felt heard by marketing, too, not just by the support agent. Luna: Forty percent — that's real. So how did they connect the support system to Marketo? Did they have to build a custom integration? Lucas: Surprisingly simple. They used a middleware tool — Zapier, actually — to watch for new tickets tagged 'negative' in their help desk, then sent a webhook to Marketo that updated a custom field on the contact record. That field became the trigger for the workflow. Luna: So no expensive enterprise integration. Any team with a basic Zapier account and a help desk that supports tags could do this. Lucas: Exactly. And the same logic works for survey data. If you send a quarterly NPS survey, and someone responds with 'Detractor' — that's a trigger. Most companies just let that sit in a spreadsheet. But you can push that score into HubSpot or Marketo and have it change the contact's lifecycle stage. Luna: What would that workflow look like? Someone takes the survey, selects a low score, and then what? Lucas: A few things. Immediate trigger: the contact gets moved to a 'churn risk' list. They receive a different email from the CEO or head of customer success — not marketing, but a personal note. And their lead score might actually get decreased, not increased, because a frustrated customer is not a sales-ready lead. Luna: That's the opposite of what most scoring models do. Most models would just see 'opened email' and bump the score up. Lucas: Right. And that's the problem. Traditional scoring rewards activity, not sentiment. But activity can be misleading. Someone who opens every email because they're annoyed and looking for a cancellation link is not your best prospect. Luna: So how do you start small? I can imagine a team listening to this and thinking, 'We should do that,' but getting overwhelmed by the data pipeline. Lucas: Start with one survey field. Pick the single most predictive question you already ask. For a lot of companies, that's the 'likelihood to recommend' or 'overall satisfaction' score. Add that field to your marketing automation platform. Then create one workflow that triggers when the score falls below a threshold. Luna: One field, one trigger. That's manageable. Lucas: And measure the impact. Compare churn rate or support ticket volume for people who received the sentiment-triggered workflow versus those who didn't. That gives you the business case to add more data sources. Luna: What about call transcripts? That seems harder because it's unstructured audio. Lucas: It's getting easier. There are tools like Gong or Chorus that transcribe and analyze sales and support calls. They already produce sentiment scores and topic tags. Some have APIs that can push that data into a CRM. So if a customer says 'I'm considering competitors' on a call, that could trigger a retention workflow. Luna: That's powerful. But I wonder about the cost. The B2B SaaS company you mentioned — were they a large enterprise? Lucas: About 200 employees, maybe 5,000 active customers. They already had a help desk and a basic NLP tool that cost a few hundred dollars a month. The Zapier plan was the free tier. So the incremental cost was basically zero. Luna: So it's not just for the big players with full MarTech stacks. A startup with SurveyMonkey and HubSpot could do version one of this. Lucas: Absolutely. And I think that's the takeaway. voice of customer data is not some futuristic AI play. It's sitting in your existing tools. The missing piece is just the pipe that connects it to your automation triggers. Luna: What about privacy? If you start piping support ticket sentiment into marketing, are you crossing a line? Lucas: That's a great question. You have to be transparent. Let customers know that feedback may influence the communications they receive. And you should never use the content of the ticket itself in a marketing email — that would be creepy. Use the sentiment score, not the specifics. Luna: So the trigger is 'this person is frustrated,' not 'this person said their printer is broken.' Lucas: Exactly. And give customers an easy way to opt out of the sentiment-based personalization. Most won't mind because the result is a better experience. Luna: I want to circle back to the example of the CEO email. Did that actually perform well? I've seen customers roll their eyes at 'personal' emails that are clearly automated. Lucas: It depends on the authenticity. In that company's case, the CEO recorded a short video message — not a text email — saying they'd reviewed the feedback personally. The video was hosted on a private link, so it felt more genuine. They saw a 15 percent lift in retention among detractors who watched the video. Luna: That's a nice touch. It shows you're not just checking a box. Lucas: Right. And it's exactly the kind of thing that's impossible to do at scale without automation. But the automation is only as smart as the data you feed it. If you're only feeding it behavioral data, you're basically flying blind on customer sentiment. Luna: So what's the one thing you'd say to a marketer listening who wants to start next week? Lucas: Log into your survey tool, export the last batch of responses, and find the column that has a satisfaction score. Import that into your marketing automation as a custom field. Then build one workflow: if score is low, send a different email. That's it. You'll learn more from that one test than from a hundred A/B tests on subject lines. Luna: And if they don't have a survey tool? Lucas: Start one. A single-question survey after a support interaction takes five minutes to set up in Typeform or Google Forms. The data will pay for itself. Luna: Alright — I think we've given listeners something concrete to try. Let's see who actually goes and builds that workflow. Lucas: Would love to hear how it goes. And if you do, you know where to find us — same place where the show stays ad-free.