How to Set Up AI-Powered Employees to Proactively Follow Up with Customers?

I. Function Introduction

Once proactive follow-up is enabled, if a customer does not reply to the AI's last message within a set time, the AI employee can proactively send a follow-up message based on the current conversation content to continue following up with the customer.

The same callback rule can be triggered at most once in each HelpKnow session, and each callback will consume 1 message quota. If multiple rules are met in the same session, each rule can be triggered once.

This article's link: "How to Set Up AI-Powered Employees to Proactively Follow Up with Customers?"

II. Enter "Proactively Follow Up with Customers"

Enter the AI employee that needs to be configured, find "Proactive Customer Follow-up" in the "Skills" area of the "Configuration" page, and click to enter the settings page.

Click "Create Follow-up Rule" to start setting up a new follow-up rule.

III. Creating Follow-up Rules

1. Set the rule name and callback time.

Enter the "Rule Name", which can be up to 100 characters long.

It is recommended to name them according to their specific purpose to easily distinguish different rules, for example:

Follow-up call for unanswered product inquiries
Follow-up call after sending the quote
After-sales follow-up

Enter the waiting time in "Follow-up Time" and select minutes, hours, or days.

The system starts calculating the time from the AI's last reply. If the customer still hasn't replied after the set time has elapsed, and meets the follow-up user criteria set in the rule, the system will initiate a follow-up call.

If the customer replies while waiting, the timer based on the previous AI reply will be canceled; after the AI replies to the customer again, the system will recalculate the time from the latest AI reply.

2. Fill out the "AI Follow-up Request" form.

The "AI Follow-up Requirements" section outlines the goals and important considerations for this follow-up visit. AI staff will then generate a follow-up message based on these requirements and recent conversation content.

This field is optional. If left blank, the AI employee will automatically generate follow-up content based on the recent conversation context.

For example, in a follow-up call scenario for after-sales issues, you can fill in:

When a customer inquires about product malfunctions, installation, or usage issues, and the AI has already provided a solution but the customer hasn't responded, the system proactively inquires whether the problem has been resolved. A follow-up call is made based on recent conversation content; if the problem persists, the customer is guided to provide necessary information or contact human customer service.

In addition to generating follow-up messages, you can also ask the AI to perform pre-configured operations when initiating a follow-up call, such as transferring to a human agent, updating customer tags, or updating attributes.

For example:

Reply to customer: Your issue requires further processing by a human operator. I will now transfer you to a human operator.

Politely ask the customer if they still need help, and update the conversation attribute "follow-up status" to "followed up".

Reply to customer: We are upgrading your service. At the same time, mark the user as "Pending Follow-up".

Before using these operations, you need to complete the corresponding configuration. For example, to switch to human assistance, you need to set the human assistance method in advance; when updating tags or attributes, you need to create the corresponding tags or attributes first.

These operations are executed synchronously when the follow-up message is triggered, without waiting for the customer's reply before making a judgment. For example, "send follow-up message and transfer to human agent" will directly transfer the conversation to a human agent; "send follow-up message and tag" will directly add a tag to the customer who meets the criteria.

The "AI Follow-up Requirements" are used to guide the AI in generating follow-up content; they are not a filter for determining which users need to be followed up. If you need to restrict the follow-up recipients, please select "Specified Users" in "Follow-up Users" and set the filter criteria.

The actual content sent during the callback is generated by AI based on the conversation and may not be exactly the same as the requirements or examples filled in.

3. Select users to visit again

The "Returning Users" feature supports the following two configuration methods:

"All Users": Initiate a follow-up visit for all users whose follow-up visit time has arrived and for whom the rule has not yet been triggered in the current session.
"Specified Users": Only initiate follow-up visits to users who meet the filtering criteria.

After selecting "Specific Users," click "New" to add criteria. The following filter types are available:

Custom Attributes : Select existing user attributes or session attributes as return visit filtering criteria. User attributes also include customer tags. Optional content is based on what's already in the current project. For creating or managing related attributes, refer to "How to Set Up AI Employees to Update Customer Tags?" for customer tags, and "How to Set Up AI Employees to Collect Customer Information?" for other user attributes and session attributes.
Contact Attributes : Use existing contact attributes from SaleSmartly as return visit filters, such as name, email, phone number, language, social media channels, country/region, etc. These attributes are preset by SaleSmartly and do not need to be created separately in HelpKnow.
System attribute : "AI Response Rounds". This variable counts the number of response rounds of the AI employee in the current session.

All conditions in the same condition group must be met; after adding multiple condition groups, it is sufficient to meet any one of the condition groups.

4. Save and enable the rules.

After completing the settings, click "OK".

After returning to the rule list, confirm that the switch in the upper right corner of the rule is turned on. When the switch is off, the rule will not trigger a revisit.

Once the follow-up rules are created, modified, and enabled, they will take effect immediately without needing to click "Publish" in the upper right corner of the AI Employee Configuration page.

IV. Management Follow-up Rules

On the "Proactive Customer Follow-up" page, you can view the rules that have been created, as well as the users, timing, and requirements for follow-up.

You can do the following:

Turn off the switch to suspend the use of a certain rule.
Click "Edit" to modify the rule content.
Click "Delete" to remove rules that are no longer in use.
Click "Create Return Visit Rule" to add other rules.

Multiple callback rules can be triggered within the same session. When creating multiple rules, it is recommended to avoid callback times and content being too close together to prevent sending multiple messages to the customer consecutively.

V. Test and review the follow-up results

Proactive follow-up calls cannot be triggered in the "Q&A Test" on the right side of the AI employee configuration page.

For testing purposes, please initiate a real conversation through websites or social media channels integrated with SaleSmartly. After the AI employee replies, ensure the customer does not send any more messages and wait for the set callback time. For detailed testing methods, please refer to: "How to Test the Actual Reception Effectiveness of AI Employees?" .

After the follow-up message is sent, you can view the record by going to the "Analysis > Messages" page of the corresponding AI employee.

Since there is no corresponding user question for the proactive follow-up message, the "Question" field may display as "-"; the "Basis for Answer" will display as "Trigger Follow-up Rule". Hovering the mouse over this icon will show the name of the follow-up rule used for this message.

VI. Applicable Scenarios

Proactive follow-up is suitable for scenarios where customers have not yet responded but further follow-up is still needed, such as:

After inquiring about the product, ask the customer if they need to know anything about the price, specifications, or purchase method.
After sending a quote or product information, ask the customer if they have already viewed it.
After providing after-sales service methods, inquire whether the problem has been resolved.

VII. Frequently Asked Questions

1. Why wasn't the proactive follow-up call triggered?

Please check the following:

Has the follow-up visit rule been enabled?
Whether the test is conducted through websites or social media channels already integrated with SaleSmartly. Proactive follow-ups are not supported in "Q&A Tests".
Has the set follow-up time been reached?
Does the current user or session meet the filter criteria in "Specified User"?
Has this rule already been triggered in the current session? The same rule can be triggered at most once per session.
In the current session, has the AI employee actually sent a message? The countdown for proactive follow-up begins from the AI's last response; if the session was initially handled by a human customer service representative and then reassigned to the AI, but the AI has not yet sent a message, the follow-up countdown will not begin.

Once the follow-up visit is triggered, you can go to "Analysis > Messages" to view the relevant records.

2. Why do I receive multiple follow-up calls from the same customer?

Each callback rule can be triggered at most once within the same HelpKnow session. If multiple rules apply to the same session, each rule can be triggered separately, so the customer may receive multiple callback messages.

In addition, manually ending a session in SaleSmartly will not cancel proactive follow-up rules that have not yet been triggered; follow-up messages may still be sent when the follow-up time is reached and the filtering conditions are met.

If a customer sends a message again after ending a session in SaleSmartly, HelpKnow will create a new session. In the new HelpKnow session, previously triggered callback rules can still be triggered again.

It is recommended to check the enabled callback rules, the users to be called back, the callback time, and the content to avoid overlapping of the applicable scope of different rules.

3. How is the follow-up time calculated after the customer replies?

The callback time is always calculated from the last AI response in the current session.

If the customer sends a new message while waiting, the original follow-up time will be canceled; after the AI replies to the customer again, the system will recalculate the follow-up time from this latest AI reply.

If, after the set time limit is reached, the customer does not continue to respond, still meets the filtering criteria of the rule, and the rule has not been triggered in the current session, the system will still initiate a follow-up call. If the rule has already been triggered, it will not be triggered again.

4. Will the "AI Follow-up Requirement" determine which users receive a follow-up call?

Won't.

"AI Follow-up Requirements" are used to guide the AI in generating follow-up content and are not used as a filtering condition for follow-up rules. When you need to restrict the follow-up recipients, please select "Specify Users" and set the filtering scope through system properties, HelpKnow user properties or session properties, and SaleSmartly contact properties.

5. Will proactively following up affect the "automatic stop replying" setting?

Yes. If "Automatically stop replying" uses "AI reply rounds (single session)" as the stop condition, proactive follow-up messages will also be counted in the current session's AI reply rounds.

For example, if the stopping condition is set to "AI response rounds (single session) greater than or equal to 3", after actively calling back to make the current session's AI response rounds reach 3, the customer will trigger the stop reply when sending a message again.

Additionally, triggering automatic stop reply will not cancel any proactive follow-up rules that have not yet been executed. Proactive follow-up messages may still be sent if the follow-up time has arrived and the filtering criteria are still met. For detailed setup instructions, please refer to: "How to set up AI employees to automatically stop replying?"