From searching Slack threads to a reliable knowledge base: How a European energy retailer built frontline support for their call center agents

In this case study, we learn how a European energy retailer's customer operations team went from searching through Slack threads to a private knowledge base serving hundreds of agents across five countries, and how a handful of features they never thought to put on a requirements list ended up mattering a lot.

At-a-glance summary:

• Industry: Energy retail

• Company: Part of a multinational energy group operating in six European countries, serving roughly 2 million customers

• Knowledge base access: Private

• Primary purpose: Internal knowledge base for customer-facing agents: procedures, product and billing information, and incident updates, all in one place

• Team using KnowledgeOwl: Customer operations, now expanding into telesales, supporting hundreds of customer-facing agents across five countries

• Login type used: Single sign-on (SAML) with reader groups for private content

• User workflow: A small team in customer operations writes and publishes; hundreds of agents read. The telesales team is building its own separate knowledge base.

• Key features: Single sign-on and reader groups, blog-style categories with callouts, topic articles, AI chatbot with Custom Response Rules, reader and chatbot analytics, glossary with auto-highlighting, search phrases, webhooks and Slack notifications, custom theme.

• KnowledgeOwl customer since: November 2025

About the company and their project

Background

Loads of knowledge (yay!), impossible to find (boo!)

Before KnowledgeOwl, this team didn't have a proper knowledge base. They had Slack, emails, and a few very busy senior teammates who knew where everything was. Product updates, billing procedures, incident notifications, and troubleshooting steps were scattered across different channels.

That worked, more or less, for a smaller team. Then the business grew, and the company started going digital, including moving its customer operations onto Kraken, a platform from Kraken Technologies that pulls a utility's customer-facing systems into one place. That's good news for agents and for customers, but a new system doesn't come with the "okay, so what do we actually say and do about this?" part. That still needed a home.

The stakes were pretty high. We're talking about hundreds of customer-facing agents, regular turnover, and people spread across several countries who needed fast, reliable answers in the middle of live calls. Agents were flummoxed when they couldn't find consistent information. Customers were frustrated when different agents told them different things. And the more experienced folks were continually answering the same questions from newer hires.

At their first Success meeting with KnowledgeOwl, the team put it perfectly: Slack "requires knowing how to search," and those searches didn't always turn up what they needed. When your whole job is giving customers accurate information, that lack of confidence doesn't feel great.

In November 2025, the operations team signed on with an ambitious goal: get a real knowledge base live before the next cohort of new agents started training. This way the new agents would start with good habits of looking first in the knowledge base for answers. They clearly care about their customers and their staff, and they could see how much a knowledge base would help both. 

The process at a glance: 

Before: No real knowledge base. Answers lived in Slack, emails, and the heads of a few very busy senior teammates. New hires asked the same questions again and again, and agents couldn't always answer customers with confidence in the middle of a call.

During: Weekly 1-hour sessions with KnowledgeOwl and the customer until the knowledge base was fully launched and adopted by their team. Meetings continued after launch, eventually dropping to twice a month for maintenance and new feature adoption. 

After: A branded knowledge base with an incident hub, an AI chatbot that only answers from the company's own content, and Slack alerts in Spanish. It went live four weeks after our first working session.

By mid-2026, hundreds of customer-facing agents across several countries were using the knowledge base in their daily work. With that high volume and a lot of agent turnover, a searchable, trustworthy source of truth stops being a nice-to-have and starts being core infrastructure.

When one team's win spreads

What started as a single operations knowledge base didn't stay that way for long. By summer 2026, the company's telesales team began working on its own separate knowledge base for training and product reference documents to support agents on sales calls. We gave it the same brand treatment as the original site.

That wasn't part of the original plan. It happened because other teams saw what customer operations had built and wanted the same thing, which felt pretty darn nice.

I can't say enough good things about the leadership at this company. They understood that time spent building something great now saves countless hours (and a lot of stress) later. They used the help we offered, collaborated with us, and built something that's truly useful in the real world. In an industry that isn't always known for caring about customers, they've shown a dedication to their staff and their customers that I really admire.

Software requirements and KnowledgeOwl fit

This team didn't show up with a requirements document, and that's okay! They knew they had a gap. The company was going digital, the team needed help with the transition to Kraken, and they wanted to better serve their customers along the way.

What was needed: The company was going digital and growing fast. With hundreds of agents rotating in and out across several countries, the team needed one trustworthy place to find answers in real time.

That's a great place to start, but it's also true that the features that end up mattering most are often those that nobody puts on a list. Here are some things this team didn't know they needed, but did:

• Single sign-on and reader groups: With hundreds of agents and lots of turnover, nobody wants to manage hundreds of separate passwords. Agents log in with the credentials they already use through SAML single sign-on, while reader groups keep internal content where it belongs.

• Blog-style categories with callouts: Updates, incidents, and important notices don't belong mixed in with evergreen procedures. Different category types let the team keep news-style content in its own feed, and New callouts help it stand out.

• Topic articles: The incident status table lives in one article and shows up at the top of the Incidents category, so the team keeps it up to date in a single place. That's a topic article at work.

• AI chatbot with Custom Response Rules: The AI chatbot only answers from the company's own content, and Custom Response Rules let admins make sure certain questions always get the exact, approved answer.

• Reader and chatbot analytics: When you're trying to change how hundreds of agents look for information, you need to know who's using the knowledge base and what they're asking the chatbot.

• Glossary with auto-highlighting: Energy and billing have plenty of jargon, and new agents can't be expected to know it all on day one. The glossary explains terms right where agents run into them, and it can be switched off for terms that show up so often they'd clutter every page.

• Search phrases: Agents don't always search using the "official" word. Search phrases help them land on the right article anyway.

• Webhooks and Slack notifications: Slack notifications and webhooks keep distributed teams in the loop when something new is published, and they're also what made the Spanish-language Slack alerts (below) possible.

• A custom theme: Not a requirement, but a knowledge base that looks like it belongs to the company gets used more. (More on that below, too.)

Project rollout: beyond the docs

Rolling out a new knowledge base is one of the places we see customers struggle, and it's one of the places this team really shone. A few things worth sharing:

Four weeks from first call to launch

Our first working session was January 20, 2026, and the target launch date was February 16, so the newest cohort of agents would have a knowledge base from day one. (No pressure!)

To make that happen, I set up something new for this account: a fixed weekly working session for three months, instead of the usual as-requested onboarding calls. I arrived at these working sessions without a plan of what to show them, and instead focused on listening to their experiences and seeing how best to problem-solve on the fly. Together we built out the branding, set up single sign-on so agents could log in with the credentials they already use, imported and organized their readers, and published the first set of real content.

All of our sessions were in Spanish, which was a first for me. Learning a whole new piece of software is hard enough without doing it in a second language, so it felt important to meet the team where they were. The weekly rhythm worked so well that we've since made it the standard for all new Enterprise customers at KnowledgeOwl.

By early February, the knowledge base was already winning people over. We built a custom theme based on the design preferences from their marketing site. One team member called the result "beautiful," and others said how nice it was to work in a space that felt like their own.

Pairing live sessions with recorded training and a written guide

Before the weekly sessions even started, the team got a recorded training video and a basic written guide, so they could revisit the basics any time. Both were lo-fi: a screen recording, narrated by yours truly in Spanish, covering the basics of authoring in KnowledgeOwl, and some key articles from our Support Docs translated with DeepL. Later, a "learn how to use KnowledgeOwl" video was added right inside the knowledge base, so new agents can find it without having to ask anyone. Live sessions plus something people can come back to on their own is a combination that works.

Inviting other teams into the room

As the project grew, the team invited new teammates and colleagues who work in other knowledge bases (like telesales) to our sessions. That meant different teams could swap ideas about what worked, and nobody had to figure things out in a silo.

Clever use of features: Updates, Incidents, and Important

After launch, the real customization began, and almost all of it came from what the team asked for in our weekly sessions.

The first ask was for a quick way to see what's new. We added an "Updates" link in the top navigation that turns green and shows a little orange counter whenever there's new content. Then they asked for the same thing for two more categories, Incidents and Important, for stuff that didn't belong in a general updates feed. That meant rebuilding the logic so each category counts its own flagged articles (and doesn't inflate the number every time something gets published). We also built a color-coded table at the top of Incidents, so an agent can see at a glance which incidents are still active and which are resolved.

Here's what that treatment looks like in action, with one known issue and two alerts for new items: 

One admin summed up what they liked best: it's "a really visual number so people know what's new."

Best practice: Slack alerts in the team's own language

Their Slack workspace runs in Spanish, but KnowledgeOwl's built-in Slack integration only speaks English. So we built a small relay just for this team that translates the notifications before they land in their channel. We wrote the setup guides in English and Spanish so their IT team could get it running without having to think in two languages.

AI answers you can trust

A few months in, the team turned on the KnowledgeOwl AI chatbot. As with every KnowledgeOwl chatbot, it only answers from the company's own knowledge base content, with no random web knowledge mixed in, so every answer traces back to their documentation. The team's first reaction was that it's fast, practical, and easy to trust. They're also planning an internal vote to give the bot a name and personality of its own, which I fully support. #TeamChatkira

As usage grew, they asked for a way to head off certain questions before the bot goes searching: canned answers or links to be pulled up automatically when readers ask common questions. We shipped Custom Response Rules, which let admins set trigger phrases (or meanings) that always get the exact, pre-approved answer.

They also asked us to switch on reader-level analytics. As one admin explained, changing longtime agents' habits is hard, and they want to recognize the people using the tool well instead of letting everyone keep asking their colleagues. Real data on their own team's adoption matters more here than it would on a public help site. From there, we kept building based on their feedback, including chatbot question exports by day, week, or month, plus behavior and traffic reporting.

Lessons learned

• Start with the gap, not the spec. You don't need a full list of requirements to begin. This team started with "we need somewhere for this knowledge to live," and let the weekly sessions turn real needs into real features.

• Pair people with documentation. Live sessions, a recorded training video, and a written guide each do a different job. Together, they covered everyone.

• Test single sign-on and reader groups together early. If you use both, try group assignment with a few real people before you roll out to hundreds. It's much easier to sort out early.

What's next

The team's plan for the rest of the year includes connecting the knowledge base to a new internal sales platform, so agents can find documentation without leaving their usual tools. What started as an effort to replace scattered Slack threads with something searchable is now something other teams across the company want, too.

Wrap-up

This European energy retailer used KnowledgeOwl to give hundreds of frontline agents one trustworthy place to find answers while the whole company went digital. It's a great example of starting with a gap instead of a spec, and letting a knowledge base grow along with the team that uses it. We're proud to have played a part, and we can't wait to see what they build next.

If you take one thing from this

Thinking about a knowledge base for your customer team? Let this be your nudge. You don't need to know exactly what you want. You just need to know that finding information should be easier. And if your company is rolling out new systems or changing how it works, the know-how needs a home, too. Waiting just means your team keeps piecing it together from Slack threads and memory.

Get something out the door, then keep improving it. And if your software vendor offers regular sessions with a knowledge-management expert, use them! There are probably solutions out there you didn't even know existed.

If you'd like to give KnowledgeOwl a try, start a free trial or book a demo with us. We offer 1:1 onboarding on every plan, and we really do love helping our customers.