Case Study

Improving Self-Service Search in Salesforce with Coveo

"When we switched over to Coveo, it felt like a caveman getting a Maserati—the experience was just not even close."
- Alex VanFosson, Knowledge Engineer at OSIsoft

OSIsoft makes the PI System, the market-leading data management platform for industrial operations.

Solutions

Industries

High-Tech

With Salesforce as their self-service platform, OSIsoft still needed intelligent search to serve personalized experiences to their customers. Coveo was that solution.

OSIsoft is known for its technical athletics, having built an operational intelligence platform to provide asset operators, and onsite analysts with real-time, high-fidelity operations data for informed decision-making. So not surprisingly, OSIsoft’s self-service journey started with a homegrown experience that tapped the in-house expertise.

Connecting disparate systems with custom search

OSIsoft built a highly customized solution to search across various systems, including their knowledge base wiki and YouTube. They also had multiple target audiences, composed of internal users, authenticated customers, and external/public audiences. While the search worked fairly well considering the amount of customizations they did, users continued to struggle. OSIsoft received complaints ranging from the typical  “couldn’t find what they were looking for” to the alarming “research was pointless.”
“Ultimately, because of this poor search experience, many of our customers, and even some of our own internal support engineers, would use the public Google search to find answers on our website.” Darragh Perrow, Business Analyst at OSIsoft

Additionally, a core function of self-service is to assist customers struggling with an issue or error. However, OSIsoft’s search tool was sensitive to syntax, and required queries to be entered in a very specific way. This was a problem because product error numbers began with a negative symbol (-). The search tool interpreted the symbol as the Boolean operator - and actually  excluded anything that appeared after the symbol. As a result, if users typed the full error number including the symbol, they either received unrelated results or didn’t get any results at all.

OSIsoft’s support team tried to guide users by adding messaging to the UI to and creating documentation on how to successfully search. They also made additional customizations to the search results, building a faceted search based on product and content type. Despite working fairly well, they continued to to receive negative feedback around the self-service experience and users continued to struggle with finding needed content.

Because the self-service experience was so heavily customized and content was disjointed, it was difficult to capture the analytics necessary to measure its effectiveness. The only reporting they were able to gather were top ten lists, which didn’t give them full insight into search trends or where they had content gaps.

Ultimately, the team at OSIsoft reached a point where they realized it wasn't feasible to continue maintaining the custom search solution. And with a corporate initiative to move to a more customer-centric model, the team at OSIsoft was determined to transform their entire self-service experience.


Seeking one source of truth

As key players in their self-service transformation, Alex VanFosson, Knowledge Engineer at OSIsoft, and Darragh Perrow, Business Analyst at OSIsoft, had a few objectives in mind:

  • A single view of customer data—one source of truth
  • A low maintenance and relevant search
  • Content delivery personalized to multiple audiences

After having spent so much time developing and maintaining search in their previous initiative, OSIsoft wanted to make sure that if they were going to undergo this massive transformation, they had to get it right. This meant removing old tools and processes to make room for a new and modern experience.

The goal was to have Salesforce become the platform for everything, from internal knowledge base to external facing community experience. This would dramatically reduce the number of tools people had to use — and learn.

Further, having been burned by a highly customized experience, they wanted to stick to only the out-of-the-box capabilities provided by Salesforce. Because of this, they were committed to eliminating features they’d become used to in their custom system. 

They also made some changes to their processes, rolling out Knowledge Centered Service (KCS) within the support team. The KCS model required relevant search results in order to be effective.

Another challenge was that Salesforce Einstein didn’t search their Salesforce Knowledge content. They needed to deploy an intelligent search solution that would not only unify all of their content from outside sources, but also provide the most relevant information, personalized to each user’s needs. 

So we had a process problem and a search problem, and we solved both at the same time by adopting Salesforce for a new knowledge base, Coveo as a new search tool, and KCS as a new process. Alex VanFosson, Knowledge Engineer at OSIsoft

Enhancing knowledge management with Relevance 

It was during a knowledge management conference when Alex learned about Coveo. There were a few things that stuck out to him that made him bring this solution to the attention of Darragh and team. The first was the AI capabilities and hearing that Coveo would be able to start learning immediately after implementation, and they could start seeing results within two weeks to a month. 

They said we could start seeing results within two weeks to a month. At first I thought they’re pulling my leg. No, actually, that was about how long it took.” 
Alex VanFosson, Knowledge Engineer at OSIsoft
Additionally, by switching to Coveo, most of the customizations they previously created in their old search were now out of the box. This meant the team wouldn’t have to spend more development time building their desired experience and instead could focus their efforts on scaling the business. 

 

OSIsoft

Maintenance was easier too. They used to manage a long list of synonym rules for their old search. Because Coveo innately learns and understands those rules based on text preprocessing and usage data, now they only maintain a short list in a Word document. Everything else is handled through machine learning. 

Lastly, Coveo’s multiple connectors and robust integration with Salesforce made it easier to implement both systems at the same time. It was important to them to use Coveo early so they could get improvement insights sooner.  

Improving self-service, with evidence

Coveo is capable of capturing analytics across the entire customer journey, and not just in Salesforce. OSIsoft was able to identify content gaps and focus on creating the content needed to support all audiences in their preferred channels. It didn’t take long for OSIsoft to see results:

  • Click-through Rate increased to 85%
  • Link Rate increased to 80%
  • No-result queries decreased to less than 1%

As their knowledge base grew, the time to resolution also dramatically improved by decreasing 56%. The team answered questions faster internally, more consistently, and enabled customers self-service more efficiently.


Looking to the future

Because Coveo continues to be effective from a technical support perspective, Darragh and team are now looking at how they can leverage it as a company-wide internal search, as the Coveo search on their marketing website just rolled out.  

They also want to improve the self-service experience by developing a taxonomy that will help better organize the content. This will allow users to narrow down the search region and browse for content based on topic or product. 
Right now, we're very focused on support and services. But I feel like there's a big potential for more sharing throughout the company. Darragh Perrow

Inscrivez-vous pour obtenir l’étude de cas

En soumettant ce formulaire, vous acceptez le traitement de vos données personnelles par Coveo Solutions Inc. et sociétés affiliées conformément à notre politique de confidentialité.

Merci

Get the Comprehensive Case Study

drift close

Hey 👋! Any questions? I can have a teammate jump in on chat right now!

drift bot
1