Customer Intent Recognition

Successful companies deploy AI to help their people create more value. Companies that fail deploy AI to avoid paying people to create value. Clients expect AI’s ROI to come from cost reductions, but bigger wins come from turning cost centers into revenue generators. A large airline client expected AI to reduce its customer service costs. We implemented AI to detect customer intent and deliver outcomes faster. Productivity improved, but instead of laying people off, we deployed a sales coach into select agents’ workflow. One model gives every customer a rating based on how likely they are to buy an upgrade and predicts the top upgrades to recommend. A second model generates a personalized pitch for the customer service agent to use. We ran a 3-sided experiment: 1️⃣ One group of customer service agents kept working on the AI intent-outcome augmented workflow. 2️⃣ A second group was given a generic script and discretion to pitch upgrades without the AI coach. 3️⃣ A third group was given the AI sales coach and discretion to decide when to accept its recommendations and which upgrade to pitch. After 3 months, the second group had an 8% upgrade pitch success rate, and the third group had a 31% success rate. In the first month, the second group pitched more upgrades than the third, but that switched in months 2 and 3. People do not immediately trust AI. They need to see it function reliably before they truly integrate it into their workflows and trust its output. Giving customer service agents discretion was critical for adoption. As the initiative scales to the entire customer service team, the airline expects to make significantly more money from upsells than it would have saved with layoffs. We reclaimed time with the AI intent-outcome agent and used the opportunity to create a new revenue stream for customer service. We found that when customers quickly go from “I have a serious problem,” to “Hello, thanks for calling support, how can I help?” to “Wow, that was an easy fix,” they’re more receptive to upsells. Businesses that win with AI are reorchestrating workflows and finding new ways to create value. Others don’t see these opportunities, so their only option is cost-cutting.

🔍 Banks Are Missing Personalisation Clues Hiding in Plain Sight Every payment tells a story — who, what, where, and why. Yet most banks still treat payments as mere transactions, not as rich signals of customer intent. 💡 A cross-border tuition payment? ➡️ Could trigger FX hedging support, student insurance, or tailored credit offers. 💡 A recurring hospital payment overseas? ➡️ Might indicate a need for medical travel insurance or visa advisory. 💡 A supplier payment in a new country? ➡️ Signals expansion — prime time for trade finance onboarding or KYC refresh. 💡 An unexpected incoming payment from a crypto exchange? ➡️ Suggests the need for tailored tax advice or digital asset risk disclosures. 💡 Multiple small value international remittances? ➡️ May reflect a gig worker or freelancer profile — prime for invoice automation or FX cost optimization tools. 💡 A seasonal spike in local utility payments? ➡️ Could hint at vacation rental income — an opportunity to cross-sell SME tools or property-linked products. But here’s the catch: Most of this intelligence is locked in unstructured fields — lost in payment purpose text, truncated in legacy formats, or ignored entirely after settlement. That’s the missed opportunity. 🧠 The future of banking lies in interpreting payment intent, not just processing payments. Banks that build intelligence into their payment stack — with structured ISO 20022 data, AI, and contextual analytics — will be the ones to deliver true personalisation at scale. Ron Shevlin Nth Exception #payments #iso20022 #banking #data #ai #vc #crossborderpayments

Fixing your email copy won't save your outbound campaign if you're emailing the wrong people. I've watched 100+ outbound campaigns fail at ColdIQ (and the problem is almost never the messaging). Here's what I mean: Most companies just grab some list from Apollo, send out like a hundred cold emails, and hope they get a reply. They have no idea if these people actually need their solution. The fix? Intent data. You'll never be 100% sure, but these signals help you: - Target people who are actually in-market - Re-engage old lists when timing makes sense - Stop wasting time on prospects who'll never convert There are several categories of intent signals: 1/ First-Party Intent ↳ Data collected from YOUR business ecosystem. These are people who already know you exist. They're: - Using your product - Visiting your website - Engaging with your content - Subscribing to your newsletter Tools to track this: Website Visitors: Instantly.ai, Clay, Midbound, Vector 👻 Product Usage: Common Room, Pocus, Mixpanel, Amplitude Social Engagement: Teamfluence™, Trigify.io, Clay 2/ Second-Party Intent ↳ Data collected about your company, via partners. That would be prospects that interacted with: - One of your partners who has overlapping customers - Your brand through a partner site - like your G2 page if you're a SaaS Examples of platforms to uncover these signals include: - Champion Tracking: Clay, Common Room - Affinity Signals: Crossbeam, WorkSpan - Review Sites: G2, Capterra 3/ Third-Party Intent ↳ Data collection via external providers. This is public data that shows they might actually need what you're selling. Examples include: Hiring Trends: LoneScale, Mantiks, PredictLeads, Clay Tech Stack Changes: BuiltWith, Similarweb, HG Insights, Clay Funding Events: Crunchbase, PitchBook, Owler - A Meltwater Offering, Clay Custom AI Agents: Relevance AI, Claygent, Apify We use Clay to build most of these workflows: → Filter for buying signals → Enrich contacts in real-time → Combine multiple data sources → Score and segment dynamically Better targeting = better reply rates = better pipeline. P.S: What intent signals are you tracking in your GTM motion right now?

For months, one of our biggest operational challenges was the mandatory human touchpoint needed to route customer interactions. Every new support ticket required a Tier 1 agent to read the description, classify the Intent, judge the Sentiment, and then manually route it to the correct specialist or seniority level. This delay was a drain on agent time and, worse, a source of customer frustration. In the last few days we've successfully implemented an AI-powered system using the Gemini API to solve this problem. We trained a model on our historical data to automatically and accurately classify every incoming interaction in real-time. The Model Now Automatically Determines: 🎯 Intent: Is this a 'General Inquiry,' 'Subscription Cancellation,' or 'Billing Inquiry'? 😠 Sentiment: Is the customer 'Neutral' or 'Critical Negative'? 📈 Priority Score: A dynamic score (1-5) that combines intent and sentiment. The Impact is Immediate and Measurable: Eliminated Triage Bottleneck: Senior agents now spend 100% of their time solving problems, not reading tickets. Faster Crisis Response: Critical issues (Priority Score 5) are routed directly to the L3 team in seconds, not minutes. Improved Customer Satisfaction (CSAT): By routing complex issues immediately, we're cutting down on resolution time and reducing the need for costly agent transfers. This shift is a game-changer for our customer experience and a prime example of how targeted AI tools can drive real operational efficiency.

Two customers owe the same amount. Both have overdue accounts. But when you look closer, the stories couldn’t be more different. One says, “I’m sorry, can we settle?” The other says, “Not my problem. Go ahead.” Same balance. Completely different intent. And this single difference changes how you should recover, how your AI should respond, and how your brand is perceived. The traditional recovery model is outdated. Most organisations still segment overdue accounts by: - Amount - Age of debt - Legal route It’s logical, but incomplete. Because not all overdue behaviour is financial. Much of it is emotional, situational, or perceptual. That’s where intent-aware recovery comes in. Instead of only asking “What’s the risk?” We also ask, “What’s the intent behind the behaviour?” By analysing tone, communication patterns, and history, AI can identify what the customer is really expressing, even without words. For example: 1/. Cooperative: needs a human-first, empathetic follow-up 2/. Hardship: needs flexible payment options 3/. Dispute / wrong party: needs verification, not escalation 4/. Hostile: needs structured, compliant escalation 5/. Silent: needs smart data validation before recontact Why this shift matters At the Puls Biznesu Receivables Conference, Joanna Hirsz said it best: “Recovery is a brand moment.” And 78% of customers judge a brand by how it behaves when things go wrong. When recovery feels human and contextual, customers remember it, even during tough times. Organizations using intent-aware recovery report: 👉 15–20% higher kept PTP (promises to pay) 👉 30% fewer complaints 👉 Lower churn on cured accounts Because the system routes cases based on empathy and psychology, not just financial data. It listens first and acts second. Collections isn’t about pressure anymore. It’s about precision. Understanding intent before acting, that’s where recovery becomes both ethical and effective. If your AI or collections system still scores only by amount or age, it’s time to evolve. Start scoring intent, not just risk because the future of recovery isn’t about chasing payment. It’s about earning trust. Who’s already running intent‑aware recovery at scale?

A $2B+ ARR customer just concluded an intent signal A/B test, and UserGems 💎 signals converted to deals 6-8x higher than their existing providers That's because of our contact-level intent. If you're still using account-level signals alone, you're missing out on a lot of insights. Your account scoring model is probably off. And your ABM motion likely sees low rep adoption. Examples of Account-level signals? Funding, hiring, M&A, tech stack, intent, partners, etc. Contact-level signals? Job changes, contact-level intent, website de-anonymization, etc. This $2B+ customer's been using account-level intent. 'Someone from Montreal Metros is researching your [topic]' is weak signal. It leads to a lot of noise, wasted ad budgets, and reps' time (resulting in reps ignoring these 'intent' signals altogether) 'Shane Hollander, CEO of Montreal Metros, was researching your [topic] 3 times the last 7 days & visited your homepage' -> Now, that's a c̶o̶t̶t̶a̶g̶e̶ home run 🏡 ❗But even contact-level signals alone aren't enough. If you still treat signals as an individual event, you won't be able to cut through the noise. This is why we see 1-2% reply rates on sales outreach. 'I saw you on our website' alone isn't gonna cut it. For signal-based GTM (including ABM & AI outbound) to be effective, you need to COMBINE the Account- & Contact-level signals with... 🎯 Your own first-party data (Closed lost, call transcripts, CRM data, event regs, content downloads, product usage, etc) If a target account has: - new executive just joined - they're hiring for sales & marketing roles - their tech stack is your ICP - their team's been researching & visiting your website - they came to your events a few times L12 months - they evaluated you last year but it got deprioritized -> That's an A account to target. right. now. This is how a $235M+ ARR customer gets 11-16% reply rates on their AI-assisted outbound. It's also how our ABX program consistently converts 10%+ to demo requests. This is the shift I'm seeing in many high-performing teams. They're building a signal system - some call it a 'signal hub', others call it a 'GTM cockpit' or 'AI brain'. At UserGems, we call it the AI Command Center 🧠 Once signals are unified and trustworthy, you can do a lot of fun stuff like: - Reps start their day with the hottest 30 prospects already queued up in their Outreach/Salesloft/Gong - ABM managers know which accounts to focus on - CMOs see the impact of brand programs in moving accounts through buying stages - fewer alignment & herding cat meetings This is where the true unlock of AI for GTM happens. I honestly think this will be a core requirement for every modern B2B stack and process by EOY.

A customer told us the “every form fill is a sales lead” mindset was killing their pipeline. Inbound looked great. But reps were drowning and deals weren’t moving. After they turned on our intent analysis, we showed them the breakdown: Only 42% of form submissions came with intent to buy. The rest were job seekers, partnership requests, and support. So they changed the flow: 1. Use AI to detect intent in real time 2. Qualify and route by intent, not just lead info 3. Only send buying intent + ICP fit to sales They saw the impact fast. A huge chunk of inbound got disqualified automatically, saving hours for the sales team.

If you want your marketing to work, you need to understand customer-intent. Why? You can segment your potential customers into three groups: 1. Those who have made a decision to buy. 2. Those with a problem, looking for a solution. 3. Those yet to figure out they even have a problem. If you try to communicate to each segment in the same way, the results won’t be great. Why would someone care about your muscle-gain supplement’s customer testimonials if they aren’t even looking to gain muscle? Or read a 10-minute article on the benefits of the product they already decided to buy? Now, what kind of content generally works for each stage of the funnel? 1. Top of the funnel → Make prospects problem & solution aware Benefit-focused content, such as: → infographics → blog posts → video ads 2. Middle of the funnel → Convert problem & solution aware into leads Informative content and offers, such as: → Educational resources → Quizzes/surveys → Discounts 3. Bottom of the funnel → Convert leads into customers Conversion-facilitating content, such as: → Customer testimonials → Case studies → Guarantees These are just some examples, but generally quite good rules of thumb. Start with understanding your prospects’ intent, and communicate accordingly. Your results will reflect the change in a positive way. P.S. How do you take customer intent into account in your marketing communication? - - - - - - If you're enjoying content like this, Follow me and hit the 🔔 on my profile: @Verneri Brander #Marketing #CustomerIntent #Ecommerce