KINSMCA chatbot can quote the wrong shipping cutoff even when the new one is on your website. The older cutoff may still be sitting in a help article or in a file someone uploaded months ago. Amio lets a store bring in knowledge from sources such as web pages, documents and product feeds. The first job is to decide which source should carry each answer and remove competing versions from the material the bot can use.
Pick one approved source for each public policy
Make a short list of questions your store receives every week. Start with general information such as shipping coverage, dispatch timing and how customers contact support. For each topic, write down the page that someone on your team would send to a customer. Assign an owner who can confirm it reflects the current policy. If the storefront and an old FAQ disagree, resolve the underlying difference before asking the chatbot to choose between them.
A narrow set of sources is easier to maintain than a crawl of every page. Look for old promotions, regional versions and downloadable guides that quote terms with a limited time or audience. Keep the correct regional version where a question depends on location. If you cannot tell which version applies, have the support owner resolve it first. Product feeds are useful for product attributes, while an approved shipping page is a better place to describe how a destination or dispatch rule works.
Keep the rule and its exceptions together
Use headings a customer would recognize, such as Dispatch times or Where we ship. Write short sections that make sense when read on their own. A hypothetical store might say it dispatches standard orders on business days and that made-to-order items follow the schedule on their product page. Keep those two sentences together. If the general statement appears in one section and the exception only at the bottom of another document, an answer can sound correct while missing the condition the shopper needs.
Separate promises from estimates. If a number represents the time before a parcel leaves the warehouse, label it as dispatch time. If it represents estimated travel time after dispatch, label it that way. Keep units, time zone and the affected market with the rule wherever those details matter. Have someone who did not write the policy read the section and explain it back. Their questions can reveal assumptions that will also be hard for the bot to resolve.
Connect the clean source and check what changed
Amio documents a Knowledge Base area for files, web content and feeds. Use the relevant source type for the policy you maintain. If the public page is where your team makes every approved change, using that page can reduce the number of copies to update. If you upload a separate file, put its source page and update owner in your own editorial record so the next change reaches both places. Check the synchronization or update options available for your setup before relying on an automatic refresh.
Once the updated source is available, try three sample questions. Ask for the normal rule, ask about the exception and ask for a detail the source leaves unresolved. For example, ask when an ordinary item ships, how a made-to-order item differs and whether a specific parcel will arrive by a date you have not promised. Compare the answer with the approved text. If it repeats an old rule, inspect the connected pages and files for the exact outdated wording. If the policy is silent, the bot should avoid inventing a promise and point the shopper to your approved support route.
Give every policy change an owner and a retest
Record the date, source and sample questions when you make a policy change. Keep one example that has tripped the bot up before so you can rerun it after the next update. If one person owns policy copy and another manages the bot, define who removes superseded material and who confirms the new answers. A small change to a cutoff or an exception should have a clear path from the approved page to the answer a shopper sees.
Public policies are a useful place to start. Order-specific details need a separate access and handoff review. Our online store chatbot test guide covers that part with sample records and checks for when a customer needs a person.