Trustpilot reduces their service backlog by 99% with Agentforce

Legacy systems, lengthy backlog

Trustpilot is one of the world's largest trust platforms, serving millions of consumers and businesses globally. As the company continued to expand, one thing became clear: Delivering on their promise of trust at this new global scale meant modernizing support to serve B2B and B2C customers faster and more efficiently.

Legacy systems, disconnected support channels, and expanding coverage generated unnecessary tickets — leaving 20 reps to manage a 15,000-case backlog. Many cases were automatically created without capturing who the sender was or what company they belonged to, forcing reps to spend valuable time gathering context before solving the issue.

This missing identification information was compounded by an email inbox filled with spam, misdirected tickets, and automated responses. Without an efficient way to identify legitimate requests, reps couldn't get to customers quickly.

Trustpilot knew that automating this repetitive work was key to delivering fast, high-quality service to every customer.

An agentic first point of contact

To make it easier for customers to get help, Trustpilot launched Sam, an agentic service experience powered by Agentforce. Sam is every customer’s first point of contact across Trustpilot’s Help Center and B2B portal, ensuring users quickly find what they need.

Once authenticated through a verification code or by logging into the portal, customers can interact with Sam in one of eight languages: English, Danish, German, French, Italian, Spanish, Swedish, or Dutch. Sam answers common questions — like how to use a feature or create a dynamic link — and surfaces resources in real time.

“We offer customers self-help and point them to articles, ensuring they get answers straight away,” said Darren Taylor, Senior Director of Customer Experience and Support. “Our team is on hand for nuanced cases, but when self-help is more efficient, getting an answer straight away beats waiting even 10 minutes.”

Customers on paid plans can be connected directly to a support rep, and more complex requests — like custom API integrations, widget implementations, or purchase inquiries — are automatically escalated. For every escalation, Sam creates a case, summarizes the conversation, and routes it to the appropriate support or sales team through Agentforce Service. Reps receive the full interaction history and recommended next steps, allowing them to pick up where the agent left off.

Authentication followed by action

Behind every conversation with Sam is Trustpilot’s connected data and service architecture, powered by Data 360. When a customer asks a question, Sam uses Data 360 as its system of context to assemble the information it needs to deliver an accurate response.

Knowledge articles are ingested through the Salesforce Knowledge connector, while customer data remains in Google BigQuery and is accessed through zero-copy integration. Together, this creates a unified view of each customer, combining account information with trusted support content so Sam understands who it’s helping, what products they use, and the guidance that's most relevant to the customer’s issue.

From backlog to breakthrough

Just eight weeks after purchase, Sam was live and helping customers. Within three months, Sam helped Trustpilot reduce their support backlog by 90%, reaching a 99% reduction within six months.

“The team sees the value day-to-day in the case reduction and making sure the right tickets get to them,” said Taylor. At the same time, Sam resolved many customer issues through self-service, deflecting 28% of support cases overall and 40% of inquiries in the B2B admin portal.

By automatically gathering context and routing issues to the right team, Sam also helped reduce average handle time for escalated cases by 45%, lightening administrative work for specialists.

Customer feedback is helping shape what's next. With 60% of Sam's interactions receiving high-quality ratings, Trustpilot uses Agentforce Observability to monitor live conversations and understand where Sam is delivering the most value, informing the team's next wave of agentic AI use cases.

“Today, Sam deals with a large volume of support requests,” said Jack Vlissidis, Senior Head of GTM Platform. “But we see it evolving into a one-stop-shop where you open it up, and it figures out where you need to go.”

Salesforce Professional Services and engineering support

Trustpilot fast-tracked implementation by partnering with Salesforce Professional Services. Professional Services conducted listening sessions with frontline workers, built the foundational architecture, and launched the first version of Sam. Forward-deployed engineers refined the implementation and helped expand Sam's capabilities — laying the groundwork for future agents.

Structured, governed coding tools

With Agentforce Builder low-code and no-code tools, Trustpilot built and refined agents faster than traditional development allowed. The team could define agent behavior in plain language instead of translating business requirements into technical specifications, letting them focus on context over code.

On-demand learning in Trailhead

Trailhead helped Trustpilot’s teams build the skills necessary to monitor, refine, and test Sam through hands-on learning and sandbox environments where they could practice while they learned. “If you find Salesforce confusing, get on Trailhead,” said Roz Maher, Head of Marketing Technology. “The modules in there will break everything down for you and make it easily digestible.”