Coherence of criteria: before writing the content, we must define a style and treatment that predominates in all the answers and gives coherence to all the attention.
It is essential to understand how we are going to address the user, whether it will be with a colloquial language (typical of an informal and relaxed conversation), or in a formal way (with a correct language and keeping distance). In addition, it is necessary to determine the complexity of the answers, whether they will be highly technical or aimed at an average user, etc.
TIP: an exercise that could be used is to think of a stereotype of customer service personnel appropriate to the company's profile, even if it is not represented visually. Based on this definition, make up the answers as if it were the character answering them.
Brevity and clarity: the answers must be concise, clear and punctual. It is necessary to adapt the available content to the chat format. The EXTENSION and MANNER of writing is not the same as that used for a web page, e-mail or user manual. It is notorious the drop in quality of the answers when using information clippings from other media.
TIP: for the treatment of a long answer (those that require scrolling to read), it is always possible to subdivide it into small, more specific topics and suggest related topics next to each answer that would complete the information available on the topic.
Contextualize: it is recommended to provide a brief introduction that contextualizes the answer, while maintaining the specificity of the information provided. Avoid answers that begin with "Yes" or "No", as they are inadequate for forms of questioning that are not questions. Here is an example:
-Question form 1: Can I change my line to subscription?
-Form of question 2: I need to change my line to subscription
-Wrong answer: Yes, if possible.
-Correct answer: You will be able to change your line to a subscription by going to a branch with your ID card.
Avoid referring the user: keep in mind the premise that the first solution channel is the virtual agent: its objective is to solve the problem. If most of the answers only refer the user to other channels, it becomes an annoying previous step of attention, which harms user satisfaction and the company's cost of attention.
Advice to retain and solve queries:
- Quantity of content: this is the fundamental input. AgentBot cannot provide solutions if it does not have the information to do so. It is required to constantly feed its knowledge base by resolving unresolved questions.
- Keep the answers updated.
- Know all the tools and complements offered by the platform to enrich the answers such as: maps, images, videos, related questions, PDF, carousel, etc.
Advice for linking to internal pages or sections:
- Make a brief introduction of the content of the link to which the user is being redirected.
- If the derivation is due to the fact that the answer requires a whole section, try at least to deliver in a few lines the most important and specific answer to the user's question, giving him the option to expand the information with the link. Many times the user searches only for a specific part of the content and resorts to the Virtual Agent to avoid reading all the information on the page.
- Instead of giving access to the home page of a site, and then the directions to the appropriate section, provide direct access to the solution.
Welcome message: it is not essential, but in case of including a greeting for when the user opens the chat, it is advisable to be brief: welcome, presentation (if it is decided to clarify that it is an automated channel, it is a good time to do it), and be available to the user for queries. In case the agent only deals with a particular issue, it is also good to clarify it quickly to avoid frustrations later on.
TIP: Use the FAQ tool at the beginning, there you can place the main customer queries in the add-ons window.
Content NOT to miss: before launching, it is essential that AgentBot has answers for: greetings, farewells, thank yous, evasions and insults.
TIP: Load more than one response for common or personality messages, as it will make AgentBot more natural. The platform allows you to load more than one response option for the same query or message, which will then be randomly selected for delivery.