For global small business platform Xero, onboarding new subscribers involves bridging a significant knowledge gap. Small business operators, such as tradespeople, contractors, and hospitality owners, often face steep hurdles when establishing their core financial infrastructure.
To address these drop-offs, Xero has shifted away from static, manual lifecycle marketing toward an automated, AI-driven engagement model designed to intervene at the exact moments users stall.
According to Vladka Kazda, Global Executive General Manager and Senior Vice President of Customer Engagement at Xero, speaking at Twilio’s SIGNAL 2026 event in Sydney, this operational transformation centres on moving beyond the constraints of manual workflows.
She says, “Where traditional lifecycle marketing breaks down is pretty much anywhere you have to rely purely on manual processes. It caps how fast you can experiment, limits how deeply you can personalise, and restricts how well you can support customers without simply adding more people to a contact centre. The north star has to be genuine one-to-one personalisation, optimising every single touchpoint, regardless of the surface the customer is on.”
Identifying the critical activation milestone
Xero’s onboarding strategy prioritises early product adoption metrics over superficial engagement. For new subscribers, establishing a direct connection to their financial institution represents the primary threshold for long-term product stickiness.
“Setting up accounting software isn’t something our customers do every day, they’re plumbers, tradies, and café owners. Getting their chart of accounts sorted or connecting bank feeds can be daunting. Attaching a bank feed is the pivotal milestone for us: once a customer does that, they can reconcile and realise the value of Xero early, which directly drives retention in those first 90 days”, says Kazda.
Moving from delayed batches to real-time interventions
To prevent onboarding abandonment, Xero integrated Twilio Segment to operationalise behavioural data directly from the product environment, pairing these signals with predictive models to trigger timely communications.”Real-time capability was a massive reason we chose Twilio Segment. We’ve got around 50 live real-time activations running today, triggered straight off product behavioural cues alongside models like churn propensity and next best action. The payoff is real: simply shifting our abandonment comms from a 24-hour delay to real time delivered an immediate three to six percentage point lift in conversion.”
Turning one-way friction into two-way support
External technical variables, including bank-specific interface requirements, frequently disrupt the onboarding journey. Standard automated abandonment reminders often directed users to self-serve help articles, which frequently resulted in dead ends. To resolve this, Kazda’s team restructured the channel to support direct agent contact.
“We noticed significant friction with certain banks when customers tried attaching feeds. An automated abandonment email nudging them to search a help centre was just a one-way dead end. So our team embedded reply-to capability directly into those emails. If someone gets stuck and abandons the setup, they can instantly reply and connect with an agent right there in the channel they’re already in.”
Constructing a self-optimising content engine
Recognising that conventional A/B testing cycles were too slow to keep pace with subscriber behaviour, Xero developed an autonomous content workflow focused entirely on product adoption.
“As an engagement team, opens and clicks are fine, but what we actually care about is downstream product behaviour, did the customer attach the bank feed? Because manual testing cycles were taking too long, we set ourselves a challenge: build a self-optimising content agent in 90 days. It draws from a single knowledge base, generates variations, pushes them into our email platform, scores them against real product adoption, and loops those learnings straight back in to adapt the next round.”
Measuring incremental impact and managing expenditure
Deploying machine learning and generative tools across mission-critical journeys introduces both analytical and budgetary scrutiny. Kazda emphasises that demonstrating actual business outcomes incrementality is crucial for maintaining organisational buy-in.
“If you’re wondering where to start with AI agents, focus on abandoned journeys and feature adoption, wherever you’ve lost someone and need to win them back. But crucially: partner with your analytics team to test incrementality with a proper control group. If you can’t prove whether a customer would have converted anyway, you’ll never get past the business’s natural objections.”
On the commercial front, Kazda recommends teams focus first on proving operational value, then refine infrastructure efficiency as use cases scale.
Navigating cultural change in marketing teams
The shift toward autonomous and real-time systems has also changed how engagement teams work. Kazda says concerns about job displacement are common across the industry, and that practical demonstrations and a move towards continuous testing helped build confidence in new ways of working. “Adopting AI requires an open mind and letting go of fear of the unknown. People worry about being replaced, but there’s always going to be a role for humans, the role itself just shifts toward work that needs human judgement. Once the team saw what AI made possible and put a few runs on the board, the mindset flipped. We’re permanently, healthily dissatisfied in the best way possible, always looking for what we can test next.”
