How Relatio reached 86% CSAT and automated 50% of tickets with Evly (EverHelp AI)

Relatio, a relationship and intimate well-being app used by more than a million people, launched its customer support from nothing and reached an 86% CSAT score within its first year of working with EverHelp.

‍

Relatio came to EverHelp as a startup with no support system in place. Within twelve months, that turned into a full omnichannel operation across email and live chat. Now, they have a well-established support system with 50% of all their incoming tickets resolved automatically by Evly, EverHelp's AI agent.

‍

If your business is launching a product and needs support ready before users start arriving, this case is worth a read. It shows how a startup can establish reliable, high-quality support from scratch without building the entire function in-house.

‍

The quick-take summary

‍

• Company: Relatio
• Industry: Saas application
• Challenge: Building a customer support operation from scratch.
• Solution: We built Relatio's support operation from scratch with omnichannel coverage, a trained team, quality tracking, and Evly-powered automation for routine tickets.
  
• Results:
  → 86% CSAT
  → 50% automated tickets
  → 4-minute FRT
  → 4.4/5 Trustpilot rating

‍

Meet Relatio

‍

Relatio is a relationship and intimate well-being app, designed to make working on your connections feel less overwhelming. It gives users a library of expert-built exercises aimed at closing communication gaps and easing emotional distance. The product has built a sizable, loyal following, with:

‍

• More than 1 million users
• 150+ exercises developed with experts
• 350,000+ five-star reviews

‍

An app handling something this personal is fundamentally built on trust, which raises the stakes for how the users who reach out for help should be treated. As a brand-new product entering the market, Relatio needed that support to work well from day one and that's why they partnered with EverHelp.

‍

Searching for a fix

‍

Every early-stage company that starts thinking about establishing support operations faces the same call: build a team in-house or hand the work to someone who already runs support for a living. For a startup, the in-house route carries more risk than it seems.

‍

Hiring a full internal team means committing to salaries, tools, training, and a layer of management before you know whether your ticket volume will be high enough. You're betting on a level of demand the product hasn't proven yet. If growth comes slower than planned, or the roadmap shifts, you're left paying for an idle department.

‍

Outsourcing, on the other hand, lets a young company launch small professional support quickly and scale the team to match real volume. That flexibility is one of the main reasons companies outsource at all. Deloitte's Global Outsourcing Survey names the following as the leading drivers behind the decision:

‍

• cost reduction
• access to skilled talent
• and agility.

‍

That's roughly where Relatio stood. The company came to EverHelp as a startup that needed a well-organized support system but lacked a functioning customer support team.

‍

{{cta}}

‍

The problem: growing pains in support

‍

Young startup products that quickly grow their user base tend to experience support difficulties from the very start. 

‍

Think about it: As an up-and-coming product, you can't predict how well your offer will land or how big your audience will get. So a lot of founders end up covering support with whatever internal resources they have on hand: product developers, the marketing team, a billing specialist. That leaves the whole support system resting on people who were never trained for it, answering questions and closing tickets instead of building real customer relationships.

‍

As Relatio’s user base climbed, so did the support volume. A growing share were billing disputes: unrecognized charges, failed payments, and refund requests. With no support team and no billing workflow in place, those messages had nowhere to go. 

‍

This is the story of most early-stage products, and the primary instinct is to quickly hire an internal team. But headcount is expensive and slow to train, which is why AI has quickly become a lucrative solution. Who wouldn’t want to just plug in a tool and forget about their support worries?

The AI deployment issues

So, unsurprisingly, 91% of customer service leaders reported pressure from executives to adopt AI in 2026. Yet, at this point, too many business executives have been burned by AI that: 

‍

• Hallucinates
• Mishandles information
• And answers inaccurately.

‍

And the worst part was that the customers could immediately tell the difference, too. In a 2025 Deloitte survey of 5,801 US consumers, 54% said they trust human agents more than AI for recommendations, against 32% who trust AI more.

‍

And though Relatio had all the chances to face the same issue, they didn’t. On the contrary, their story has become an example of how a well-built and properly integrated AI agent can clear the routine workload and keep the customers satisfied with the assistance they get.

‍

Enter Evly (EverHelp AI)

‍

Initially, Relatio had an ambitious goal to automate around 75% of tickets end-to-end. At the time:

‍

• We were already running Relatio's support as its outsourced partner
• We’ve established a trained operations team that knows the product and its customers. 
• Our team had an AI solution ready to be trained and implemented. 

‍

So, rather than sourcing a separate AI tool, we layered Evly onto Relatio’s existing support operation, with the same team training and supervising it.

‍

However, after analyzing the product's nature and support requirements, we found that 75% automation may not be achievable. Most of the requests were either too sensitive or too complex to delegate to AI. 

‍

Example: At first, even the refund process was so intricate that the AI could only fully cover the first two steps, while the rest required a person to review and decide.

‍

So, instead of pushing for the full 75%, we decided, together with Relatio, to stick with the hybrid human + AI model. This allowed them to use Evly to handle high-volume, routine tickets and refocus their human team on sensitive, multi-step requests. 

‍

And we believe that it’s exactly such a setup that has allowed us to automate a larger share of the volume while maintaining a positive customer experience, even as demand has grown.

‍

How it actually worked

‍

At EverHelp, we believe that it's the gradual automation that brings businesses the most results and success. We started narrow, with the requests that were repetitive and simple enough for Evly to resolve in one go, and widened the scope as the system earned trust.

1. Scoping

First, we laid out the requirements. Going through the ticket types, we established:

‍

• Which ones could be automated end-to-end
• How the resolution workflow for each was built
• How many steps would Evly need to resolve them, and which ones?

‍

For Relatio, we started with 3 request types:

‍

• Cancellations, handled end-to-end
• Refunds, but only for the first two steps, with a person finishing the process
• General questions, handled end-to-end

‍

Then we gathered a batch of real examples of those kinds of tickets and used them to teach Evly the resolution pattern.

2. Setting guardrails

Before going live, we set up rules for Evly to follow while processing tickets. The main guardrail was that it should route any ticket it wasn’t trained to handle to the appropriate support agent.

3. Launch

Once the AI was trained and all rules were set up, it started processing tickets. During that time we were closely monitoring: 

‍

• How the AI handled ticket processing
• What it escalated to humans.

‍

We then analyzed the types of cases that kept coming back to human agents and added them to AI’s “typical workflows,” so it could handle them in the future. 

‍

This has allowed us to automate 50% of ticket volume in the first month.

4. Automation improvement

Seeing Evly’s workflows hold up, we expanded the scope and pushed automation to 60% over the next couple of months. 

‍

Now that Relatio's confidence in Evly has grown, they decided to let us set up the AI to run refunds end-to-end, so it could also issue refunds on its own.

‍

Today, the human team mostly handles the cases that don't follow a script or have a repeatable pattern. That mostly covers:

• Tickets involving legal entities
• User complaints that lean on trigger words
• And the occasional unusual or highly technical questions.

‍

{{cta-lm}}

‍

The results

‍

Our partnership with Relatio has allowed to achieve palpable results:

‍

• 60% of tickets are automated at 95% processing quality → thanks to proper AI training and gradual implementation
  
  
• 4 minutes FRT → Evly provides responses to the repetitive task in 30 seconds, while the human agents focus on quickly reacting to complex queries.
  
  
• 92% SLA fulfillment → because we established structured support workflows that allow agents to optimize their work and reach necessary KPIs.
  
  
• 86% CSAT and 4.4/5 Trustpilot rating → thanks to AI and agents working together to deliver the assistance Relatio’s customers seek.

‍

What changed for the team

‍

Before Evly, a large share of our agents’ time was spent on the same handful of requests. Now, as the AI agent handles most of that, they can concentrate more on customer care, building actual trust with the user base. 

‍

For a product built around handling sensitive and more involved topics, that’s exactly what you want your support team to focus its effort on. 

‍

Key takeaways

‍

Relatio’s situation is yet another proof that AI tends to work best when it's layered onto support that's already organized, then introduced gradually instead of all at once.

‍

Thus, we recommend:

‍

• Start with your highest-volume, lowest-complexity tickets that AI can automate in one go.
• Don't chase an automation target for its own sake. Let the product decide which cases must be processed by human agents, since some requests are too sensitive or too complex to hand off. 
• Train the AI and set guardrails so it routes anything it wasn't built for to a person rather than improvising an answer.
• Expand automation gradually, as the system proves it’s working. 

‍

What's Next for Relatio

‍

Now, Relatio has a stable, well-run support operation, which is a strong base for us to build on further. 

‍

Our next steps will mostly focus on: 

‍

• Expanding the automated ticket processing to possibly reach the expected 75%.
• Maintaining SLAs
• Decreasing FRT to optimal 1-2 minutes.

‍

Optimize your support with AI handling the routine

‍

Product support indeed looks different, so there's no one-size answer here. Yet, Relatio is a great example that, when support operations are well organized and workflows are clearly established, AI can be gradually introduced into the ecosystem and bring in actual business results. 

‍

So, if your team is thinking about trying AI solutions for your operations, talk to our team. Maybe Evly is the exact match you were looking for.

‍