SumUp deployed Pyto's voice AI across 8 countries for inbound lead qualification. +91% SQL conversion, 82% cost reduction, $57.8M pipeline. Here's what happened.
SumUp is one of the most recognized payments companies in the world. They serve millions of small businesses across Europe, Latin America, and beyond, restaurants, retailers, tradespeople, market stalls. Their product is simple: affordable card readers and payment infrastructure for businesses that can't afford enterprise point-of-sale systems.
Their sales motion is equally direct. Inbound leads come in at high volume, small business owners who've seen an ad, heard about SumUp from a peer, or searched for a card reader solution. The job of the sales team is to reach those leads, qualify them, and book appointments with the right reps.
Simple in theory. At the scale SumUp operates, across 8 countries, in multiple languages, with thousands of inbound leads per month, it became a significant operational challenge.
This is the story of how they solved it.
The challenge: inbound volume that outpaced SDR capacity
SumUp's inbound lead volume was not the problem. Their marketing was working. Leads were coming in.
The problem was what happened next.
Small business owners are hard to reach. They're on their feet during business hours, serving customers, managing staff, handling deliveries. They don't sit at a desk waiting for a callback. The window during which they're reachable, willing to talk, and still thinking about the product they just looked up is narrow.
SumUp's SDR team was doing its best. But doing its best meant working through a lead queue, calling down a list, leaving voicemails, following up, moving to the next. In practice, this meant:
- Significant portions of inbound leads were contacted hours or days after they submitted their information, well past the window of peak intent
- Some leads were never contacted at all, falling through the cracks as the queue grew faster than the team could work through it
- Coverage across 8 markets required local-language SDRs in each country, a headcount model that was expensive to maintain and difficult to scale consistently
- CRM data quality was variable, SDR notes were inconsistent, qualification criteria applied differently across reps, pipeline visibility was limited
The result was a qualification rate that left significant revenue on the table, not because the leads weren't there, but because the process couldn't reach them effectively enough.
The solution: voice AI for first-touch inbound qualification
SumUp deployed Pyto's voice AI platform to handle the first-touch qualification call for every inbound lead.
The setup: the moment a lead submits a form, at any time of day, in any of the 8 markets, Pyto's voice AI agent calls them. Within seconds of CRM entry, not hours. The agent introduces itself, explains why it's calling, and moves through a qualification flow adapted to each market and language.
The qualification conversation covers the core criteria SumUp needs to determine fit: the type of business, the volume of transactions, the current payment setup, and the timeline for a decision. If the lead qualifies, the agent books a meeting with the appropriate AE, directly in their calendar. If the lead doesn't qualify, the call ends cleanly and the outcome is logged.
Every call, qualified or not, answered or voicemail, generates structured data that flows directly into SumUp's CRM: qualification status, objection type, call transcript, AI summary, and recording.
Key elements of the deployment:
- Voice AI agents deployed across 8 countries and languages, with native-sounding voices adapted to each market, not generic text-to-speech
- Qualification flows customized per market to account for local product variations, pricing, and regulatory context
- CRM integration with real-time data sync, every call outcome logged immediately, no manual entry
- A dedicated Conversion Engineer assigned to the account, monitoring performance metrics, running A/B tests on greetings and qualification sequences, and reconfiguring the agent as SumUp's sales motion evolved
- Proactive performance monitoring, when contact rates shifted or qualification rates dipped, the Conversion Engineer identified the issue and resolved it before it became a problem
The results
The outcomes of SumUp's deployment are among the most documented in the voice AI sales category.
+91% improvement in SQL conversion
The headline result: SQL conversion improved by 91% compared to the SDR-only process.
The source of this improvement is important to understand, because it wasn't a better script. It wasn't more sophisticated AI. It was, fundamentally, the fact that every lead was now being reached.
As Sahana Bommanagouder, Senior Business Process Owner at SumUp, described it:
"Since launching the AI SDR, we have seen a 91% improvement in SQL conversion, primarily due to better engagement rates. We are now able to reach prospects that we were previously unable to engage."
Leads that previously fell through the cracks, because the queue was too long, because they submitted at 10pm, because the local SDR team was at capacity, were now being called within seconds of form submission. The improvement in SQL conversion was a direct consequence of improvement in contact rate.
82% reduction in cost per lead processed
Qualifying a lead at SumUp's volume previously required SDR headcount in each market. Headcount with salary, management overhead, ramp time, and the inevitable attrition that comes with SDR roles.
Automating the first-touch qualification call reduced the cost per lead processed by 82%. The economics of the program shifted fundamentally, qualification at scale became viable in markets where the SDR unit economics previously didn't work.
$57.8M in pipeline generated
Across the deployment period and markets, the voice AI program contributed to $57.8M in qualified pipeline. This is a direct revenue impact metric, not a proxy, not a leading indicator. Pipeline that wouldn't have existed without the program.
5.3x average ROI
The program delivered a 5.3x return on investment, meaning every dollar spent on Pyto's platform generated $5.30 in measurable pipeline value.
8 countries, multiple languages, consistent performance
The deployment ran across 8 countries simultaneously, with agents operating in local languages with native-sounding voices. Performance was consistent across markets, not just in the flagship deployment but in markets where SDR coverage had previously been thin or non-existent.
SumUp internally recognized the program as one of its most successful AI projects, building a dedicated internal team around the Pyto partnership.
What made the difference: Conversion Engineering
The results above didn't happen on day one. They compounded over time, and that compounding was not accidental.
Every Pyto deployment includes a dedicated Conversion Engineer: a specialist assigned to the account whose job is to monitor performance continuously, identify what's working and what isn't, and actively optimize the agent's performance.
For SumUp, this meant:
- Greeting optimization, A/B testing the opening line of the qualification call across markets to find the formulation with the highest engagement rate. Small changes in the first 5 seconds of a call had measurable impact on whether prospects stayed on the line.
- Sequence tuning, adjusting call timing, retry cadences, and voicemail messaging based on contact rate data across markets and time zones.
- Qualification flow iteration, refining the qualification questions as SumUp's product evolved, new markets launched, and the team's understanding of their ICP sharpened.
- Proactive monitoring, when metrics shifted, a dip in contact rate on the Spain agent, a change in qualification rate in a new market, the Conversion Engineer flagged it and responded before it affected pipeline.
Tristan Forest, Go-to-market Lead at SumUp, described the partnership this way:
"Thanks to Pyto, we added an automated step to our sales funnel, pre-qualifying all leads. This allowed our sales reps to stop wasting time on unresponsive leads and focus on the qualified ones, calling them at the time they chose. In just a few weeks, this improved our SQL conversion by 91%."
The framing matters: "a few weeks." Not a year-long rollout. Not a complex implementation. The speed of impact was a function of infrastructure that was ready to deploy, and a Conversion Engineer who knew how to optimize it fast.
What SumUp's experience reveals about inbound lead qualification
SumUp's results are striking. But the underlying dynamic they reveal is not unique to SumUp. It's common across any B2B company with high inbound lead volume and a phone-first sales motion:
The leads are there. The process isn't reaching them.
Most inbound lead qualification failures are not qualification failures, they're contact failures. The lead was qualified. It just never got called at the right time.
The implication: the highest-leverage intervention in most inbound sales processes is not a better script, better targeting, or better SDR training. It's closing the gap between "lead created" and "first call made", consistently, at scale, across every lead in the pipeline.
That's a capacity and consistency problem. And capacity and consistency problems at scale are infrastructure problems, not people problems.
Key takeaways for B2B sales leaders
1. Contact rate is the first lever, not the last Before optimizing your qualification script, your AE demo process, or your CRM fields, look at your contact rate. If you're not reaching a significant portion of your inbound leads within 10 minutes, the optimization ceiling on everything downstream is artificially low.
2. Volume creates the need for automation SumUp's SDR team wasn't underperforming. They were doing exactly what human SDRs can do. The ceiling they hit was a volume ceiling, one that hiring more SDRs would have raised incrementally and expensively. Automation raised it structurally.
3. Multilingual, multi-market consistency is a genuine differentiator Maintaining consistent SDR quality across 8 markets and languages is a management challenge that most companies underestimate. Voice AI with native-language capability and consistent performance across markets solves a problem that headcount-based SDR coverage creates.
4. The agent is only as good as the optimization behind it The 91% SQL improvement at SumUp wasn't the result of a great initial deployment. It was the result of continuous optimization, testing, adjusting, monitoring, improving. A voice AI agent set up once and left alone will plateau. One that is actively optimized will compound.
See how Pyto's voice AI performs on real sales calls → Listen to a demo call
Ready to compare it against your current process? → Book a demo
Related reading:
- Inbound lead qualification — the complete guide for VSaaS teams
- What is an AI SDR? Definition, use cases and limits
- The real benefits of automated outbound calling