Master self-service through product documentation

We want to take you in depth into what’s driving the trends in customer support and the role product documentation plays for businesses financials, customer satisfaction and AI strategy.

As competition between brands has increased, as margins have tightened, pressure to not only implement AI, but show benefits from AI has increased - it’s never been more important to stay ahead of the curve.

Why self-serve is the new standard for SaaS

For the past few years, we’ve seen a rise in customers expecting self-service. This is backed up in a study done by Salesforce in which they find that 61% of customers now prefer self-service resources over talking to a human support agent.

When companies see "lack of support" as a detractor in NPS or CSAT surveys, it usually stems from two issues:

1. A genuine lack of available agents.
2. The inability to self-serve, leading to a friction-filled support experience.

The ability to self-serve for customers to self-serve can decrease these complaints by providing answers when they need it. Whilst self-serve doesn’t involve contacting a human support agent it does leave the customer satisfied with your service.

The more effective your ability to self-serve customers, the more efficient your actual support teams become as they receive less support tickets. This means that self-serve support is both a revenue play and cost reduction activity.

In fact, we strongly believe that self-service is key to scaling efficiently. Though, there a three main barriers to doing so:

• Content Fragmentation
• Lack of documentation optimisation or attribution
• The cost or burden of producing documentation

Transitioning support into a revenue generating machine

Why is self-serve the way? Well, traditionally, customer support is viewed as a "cost center"—an expense that sits below revenue and eats into gross profit. What we’ve seen is an increasing pressure for support centres to make customer support a competitive advantage and get involved in revenue generation.

Although it seems more work, it actually gives support teams two levers to positively contribute:

• Reduce cost through efficiency
• Increase revenue by making support a competitive advantage.

At Knowted, we’ve built our platform around optimisation and attribution. Just as marketing teams track which ads drive sales, support teams should be able to quantify how their product documentation reduces churn and increases product adoption.

For years, we’ve been aware that bad CX leads to churn, and customer dissatisfaction. We expect that the CX becomes more and more of a selling point and competitive advantage for brands.

Knowted believes we should borrow some lessons from revenue teams such as marketing who’ve been at the forefront for years. One feature that is common across all revenue focused teams is their constant experimentation and drive.

That’s why Knowted is built around optimisation and attribution that you really only see in true marketing platforms. By making optimisation front and centre, Support teams can iterate, improve and crucially quantify their value.

Efficiency is more important than cost reduction

Wait a minute. Are you sure? Yes. Being efficient with how documentation is created can create a flywheel of effects:

• Documentation is created automatically
• Less content gaps and higher accuracy leads to higher self-service
• Higher self-service leads to lower requirements for support teams
• Cost efficiency can be captured or redeployed to revenue generating activities.

We have introduced Knowted Agents which are complex agentic AI systems that autonomously explores and documents your product exactly like a user would! This gives you rich, accurate documentation from the users perspective.

Knowted Agents can significantly increase the speed in which you create your documentation but also significantly decrease the total cost of the maintenance of your documentation.

Content for humans and LLMs

In 2025 and beyond it isn’t just humans reading content - it’s AI too. As more customers turn to AI such as ChatGPT, Gemini or Perplexity, appearing positively in AI is crucial. Interestingly, the old idea behind product documentation has become even more important as a solid source of truth is needed.

LLMs.txt

LLMs.txt is a proposal launched by the famous Jeremy Howard to standardise how LLMs consume content.

The idea behind it is quite simple: LLMs require content, but ingesting websites with all its HTML markup is too inefficient.

LLMs.txt seeks to provide a standardised way to present information to LLMs. LLMs.txt are both human and LLM readable and are effectively Markdown files containing your content - allowing for easier processing than a standard website.

To date, LLMs.txt has been the initiative that has gained the most momentum. That is why when you create a Help Centre with Knowted we automatically create both a llms.txt and llms-full.txt for you, located at your-domain/llms.txt. (you can check out Knowted’s own help centre llms.txt here and llms-full.txt here.

Knowted also updates your site’s robots.txt to help direct LLM crawlers to your llm.txt files. This helps your brand get recognised and increase your GEO posture. As noted by SEMRush, product documentation is becoming increasingly important for brand recognition as it’s used to answer pre-purchase questions when users interact with LLMs such as ChatGPT.

Taking documentation to where users are

Documentation is getting wider reach. The closer you can bring help to your users the better. Even better if you can begin to predict what help your users need before they even need it.

The wider reach of content creates a unique challenge for documentation teams as it becomes harder to manage across platforms and silos. This becomes a bit of a snowball effect when outdated content is scattered across tools and dare we say, AI tools.

Knowted is designed to be central to your content operations, but not the central tool. What do we mean by that? We mean that Content is created or synced to Knowted. Our powerful REST APIs allow it to be connected to your own Knowted Help Centre, CRM system such as Intercom, or even your product.

But, by being central, you still get the benefits of Knowted’s machine learning based optimisation so your users get the best documentation every single time. Plus, everything is recorded and attributed back to your content.

New AI channels need grounding

Whether intentional or not, our documentation is being served in new channels and ways. From Google’s generative answers, your own generative answers or even chatbots you may use, it all needs a strong grounding through a source of truth.

This, in most cases, is your help centre. Content gaps or incorrect information becomes even more important as there is little to no human oversight within these support channels.

Take the example of a chatbot, if your content is outdated there’s a high chance your chatbot such as Intercom’s Fin will give incorrect answers without you even knowing. Given that 56% of customers rarely complain about a bad service you may never know that this has happened.

Keeping your product documentation up to date is a hugely time-consuming and costly task no matter the size of your business. Knowted Agents can help hugely in this. By scheduling Knowted Agents to run regularly, Knowted Agents will be able to explore your actual application like a user would and validate the accuracy of your product documentation and update it real-time.

This drastically reduces the total cost of ownership for your documentation, making your teams a lot more efficient.