-May-15-2026-08-50-57-4885-PM.png?width=184&height=45&name=image%20(3)-May-15-2026-08-50-57-4885-PM.png){kind=small}
-May-15-2026-07-59-38-3456-PM.png?width=1080&name=image%20(1)-May-15-2026-07-59-38-3456-PM.png)
The Background
Balancing Tradition and Modernity
Tradeprint's customers don't just buy print. They build their own businesses on it. A reseller in Manchester, a design agency in Glasgow, a small business owner ordering 500 flyers for Saturday's market. Each of them has a customer of their own waiting on the other end of that order. When the print lands on time and looks right, it's a quiet win for everyone in the chain. When it doesn't, it's somebody else's missed deadline, somebody else's disappointed client.
That's the unglamorous truth of the trade-print industry. Tradeprint, founded in Dundee in 1997 and acquired by the Cimpress Group in 2015, has spent nearly three decades getting that chain right: same-day production, next-day delivery, an e-commerce platform built for resellers who operate in hours rather than weeks.
The company sits inside a portfolio that includes Vistaprint, Pixartprinting, Exaprint, and Druckwerkdeal, and it's recognized as one of the leading-edge operators in UK print, both commercially and on sustainability. Its director, Anthony Rowell, chairs an environmental forum for the British Printing Industries Federation and sits on the standards committee for the Independent Printing Industries Association.
But print is a traditional industry. And the modern e-commerce experience customers have come to expect — instant answers, real-time order tracking, and no friction between a question and a reply — was running ahead of what most printers could deliver.
The Challenge
Every Unanswered Chat Was Someone's Missed Deadline
For a reseller, "where is my order?" isn't a casual question. It's a billing call from their own client in twenty minutes. Tradeprint's ~15,000 active customers and ~30,000 in the database lived inside a transactional rhythm where artwork queries, payment glitches, product-line questions, and order-status checks couldn't wait until the next working day. They needed answers in minutes, not hours.
The problem was that Tradeprint couldn't keep up. The legacy chatbot, "Bella," was a decision-tree tool deflecting only about 20% of inbound chats. The rest landed on a small support team already inundated with phone calls about the same low-level admin questions.
Anthony Rowell, Tradeprint's Sales and Customer Success Director, watched the queue and saw the obvious cost: every minute spent triaging "where's my order?" was a minute the team wasn't spending on the customers who needed real, human attention.
A Small Team, By Design
What made the pressure point sharper was how Anthony thought about his team. He doesn't lead a contact-centre operation. He leads a small, tight unit he chose to keep that way. So when he started evaluating AI vendors in 2023, he kept hitting the same pitch (replace headcount with AI) and rejecting it. "I actually found that quite offensive," he says.
"I've got a small tight team where we want to use technology to scale without adding additional headcount. I want to add skills that sit alongside the agent." He was not looking for a tool to replace people. He was looking for a tool to give a few people the reach of many. The vendors that didn't understand that distinction kept disqualifying themselves.
When HubSpot's messaging at its INBOUND-era growth conference shifted to agents alongside human agents, Anthony took notice. Cimpress Group's tech-stack alignment nudged him toward Breeze too. He had already negotiated favourable terms with another AI partner and was, in his own words, "really hesitant" about pivoting. But he made the call.
The Solution
Instant Answers, Without Adding Additional Headcount
The customers calling Tradeprint mostly wanted simple things, fast: order status, payment confirmation, product specifications, artwork guidance. Anthony and his team set out to build a system that could answer those things in seconds, around the clock, while still routing the hard problems to a human who had time to handle them properly.
That system became Breeze Customer Agent inside Service Hub, but not in the way most "rip and replace" AI deployments imagine.
The Triage Layer That Made Everything Else Work
The first instinct was to switch off the legacy bot and let Customer Agent run solo. It didn't work. "We weren't, things weren't being routed in the right way," Anthony recalls. So the team rebuilt Bella as a V2: a triage layer that sits in front of Customer Agent, plotting the route a chat should take before handing off. Bella V2 catches and routes; Breeze Customer Agent resolves.
That single design choice (keep the existing flow, layer Customer Agent into it) moved deflection from ~20% to ~40% almost immediately. It was a counter-narrative to the rip-and-replace pitch, and it worked.
The Six-Month Investment Most Stories Skip
The next jump didn't come from the agent. It came from the unglamorous work of feeding the agent. For about six months, Anthony's team rebuilt Tradeprint's knowledge base, enriched product information, restructured customer-facing content, and most importantly flowed live order data into HubSpot. That was the unlock. The single biggest unanswered question category had always been "where's my job?" Once Customer Agent could see an order, it could answer the question.
Deflection climbed from ~40% to ~60%. The day Anthony checked the dashboard before the interview for this story, it was 59%. His target is 80%, and he's iterating toward it.
The Quiet Wow Moment
The most telling sign of the shift wasn't a dashboard at all. It was the behaviour of his human agents. "The agents like cherry-picking good inbound chats," Anthony says, half-laughing. They'll see a chat from a customer they recognise, or one with a clear opportunity, and jump in to take it themselves, even though Customer Agent could have handled it. For Anthony, that's not a problem to solve. That's the cultural marker he was hoping to see. His team isn't on the back foot anymore. They have the bandwidth to choose their conversations. His CSM treated Bella V2 "like her baby" for three months, holding its hand through the learning curve, because she didn't want a single customer to have a bad experience on the way through.
The Transformation
What Changes for Customers When the Team Has Time to Call Them
The most visible result of all this isn't the deflection number. It's the phone call.
Tradeprint now calls every first-time customer who places an order. Not a templated email. Not an automated sequence. A real conversation, from a real person on the customer success team, asking how the order went and what could be better. That motion didn't exist before. It couldn't exist before, because there was no capacity for it. Customer Agent's deflection is what created the room.
For a first-time reseller running a small agency in the UK, that call lands differently than another marketing email. It signals that the supplier behind their brand actually cares whether their first job went smoothly — which means their own clients are about to get something good.
From Reactive to Proactive: A Compounding Effect
The compounding effect runs further. Customer Agent transcripts are now an input to Tradeprint's product development backlog: payment-provider glitches, product-line issues, and friction in the pre-press flow surface in chat data, get logged, and feed directly into engineering and finance.
The pre-press strategy — historically the biggest source of friction in printing because customers struggle to supply artwork correctly — has been reshaped on the back of what Customer Agent sees every day. New online tools and better self-serve flows are being built from real customer signal, not assumption.
A parallel initiative on churn — embedding RFM scoring and a Customer Score metric into HubSpot to let the sales team make proactive calls to at-risk customers before they leave — has driven roughly a 2% churn reduction. Inbound call volume is down because more customers are getting resolved on the first contact. And the prospecting motion Anthony has built alongside this work is returning revenue the company simply wasn't capturing before.
The Mindset Shift
The fear is gone too. When Anthony's team started, the worry was the one every support team has when AI shows up at the door: is this for us, or is this instead of us? Anthony was explicit from day one that it was for them.
The proof is in the way the team works now. They don't avoid the tool. They lean on it. And when a chat goes sideways, the CSM can tell Anthony exactly why — usually a stale page on the website or a piece of legacy content the agent picked up — because she knows the system that closely.
Anthony has built a 33-slide Breeze AI roadmap for his executive stakeholders, and Cimpress Group leadership is actively watching Tradeprint's results with a view to scaling Customer Agent across the wider group, including newly-acquired Mixam. AI at Tradeprint isn't a side experiment anymore. It's strategy.
For Anthony, the shift is less about the technology than about what the technology gave his team back: the ability to be on the front foot.
-May-15-2026-08-50-57-8415-PM.png?width=120&height=120&name=image%20(5)-May-15-2026-08-50-57-8415-PM.png){kind=avatar}
-May-15-2026-08-50-57-4885-PM.png?width=195&height=48&name=image%20(3)-May-15-2026-08-50-57-4885-PM.png){kind=small}
-May-15-2026-08-50-57-5826-PM.png?width=215&height=50&name=image%20(4)-May-15-2026-08-50-57-5826-PM.png)
-Apr-13-2026-05-55-20-9731-PM.png?width=215&height=50&name=image%20(1)-Apr-13-2026-05-55-20-9731-PM.png)
