How to improve accuracy
Using best practices and optimizing the AI Knowledge base is key to accurate answers.
Achieving high response accuracy depends on the quality of your content and how it's structured.
With a well-prepared knowledge base, the answer accuracy can exceed 90%, but this requires some human attention.
Start small. Upload a few well-prepared examples, test, and iterate based on real usage.
Data Sources
Format
Adding the whole website
When using your website as a data source, it is necessary to only include relevant pages and use the correct CSS selector to filter out unnecessary data (menus, headers, footers, …). AI Knowledge cannot and will not answer accurately when 50% of indexed data is irrelevant or unreadable even by a human — let alone AI.
Always check the data you indexed from your website manually in the AI Knowledge base.Text-Based content
All information should be in text form. The content must remain understandable even if all images, videos, and other media are removed. This is a common problem with PDFs or websites.Tables/charts
When tables or charts are converted to text form, a lot of information is lost due to loss of structure. Information needs to be in the text itself — not in the structure, format, or visual layout. Imagine reading a table on a braille display.Avoid text in the image format
PDFs or documents with text, tables, or any other information saved as images, should be converted into editable text. AI Knowledge base cannot read image-based text.
This text can be copied and AI Knowledge is able to read it easily.

Structure
Avoid redundant explanations
Ensure that each topic is covered only once in the data sources to avoid conflicts. A common mistake is including multiple data source types (e.g., web page + FAQ document) that both contain slightly different answers to the same question (or when the answer on the web page is later updated).Use FAQ format when possible
Write your content as question-and-answer pairs for clarity and easy retrieval. You can include multiple variants of possible user questions. Consider recreating your knowledge base using the Manual QnA option in AI Knowledgebase to achieve best results.Visual separation
Ensure topics are visually distinct by using headings, subheadings, or bullet points.
I want to change email in app. How to change login email to my account?
You can change the email in both the application, or on our website.
In the application, navigate to…
On the website, click…
Account email can be changed in the application or on the website. Open account settings and navigate to…
Copywriting
Restate answers
Answers should provide full-sentence responses, not just “yes” or “no”. This allows the AI to reuse the answers effectively. Also, include full context in every sentence rather than assuming it’s obvious from context or location. This is especially important when not using question-answer pairs.Repeat key information
Don’t rely on context to make important details clear. Repeat the information where necessary.
On Apple devices (iOS): Press X to…
On Android devices: Press Y to…
On Apple devices, press X, or on Android devices, press Y to…
Clarity and simplicity
Use straightforward sentences. Ambiguous phrasing can confuse AI, leading to inconsistent responses.Explain abbreviations
Always define abbreviations within each text block, even if they are used elsewhere in the document. Avoid internal jargon.Consistent tone
Maintain a uniform writing style.
Choose one and use it consistently.
Enter the code…
It is necessary to enter the code…
While LLMs are usually able to process data even without adhering to all these rules, it will ensure more accurate answers in cases when user questions are not clear, contain grammatical mistakes, or speech-to-text transcriptions errors.
We are also strictly limiting our models to answer only when the answer is found in your data sources, to prevent hallucinations — that’s why it’s essential to be able to search your data effectively.