Cannot wait to chat!
Skip to content

Deploy AI Chatbot

Juji AI chatbots can be deployed to a website or to a Facebook Messenger on a Facebook page.

Deploy to Website

When deploying a chatbot to a website, one can customize the web cover page. As shown below, one can edit the welcome message and configure the fields to be shown on the cover page:

  • First Name (required)
  • Last Name (optional)
  • Email (optional)

Web Deployment

Activity Tracking

An optional activity tracking setting is provided for page-owners who want to track their visitors' chat activities in their Google Analytics. In case you are not interested in adding activity tracking, you can safely skip this section and go directly to Generate URL.

Web Deployment Activity Tracking

To enable activity tracking, you will need to fill in all three fields listed in Step 2: Domain, Web URL and ID.

  • Domain: the domain of your page without the scheme or path, e.g., mozilla.org
  • Web URL: the origin of your page, this will typically be scheme + :// + domain, e.g., https://mozilla.org
  • ID: the Google Analytics tracking ID. You can get this ID on your Google Analytics page, go to Admin > Property Settings > Tracking Info > Tracking Code > Tracking ID. The ID will have format similar to UA-123456789-1.

Once the activity tracking is enabled, your deployed Juji web chat page will be updated with your tracking info.

Web Widget Activity Tracking: If you are using Juji web widget on your page, you simply insert your web plugin code as instructed in the Deploy as Web Widget section below. The code gets updated automatically with your activity tracking setting, so if you change your setting here, make sure you use the updated web plugin code.

Web URL Activity Tracking: If you are using web URL directly and you want to have cross-domain tracking between your site and your Juji chat, you will need to perform the following two actions in order to ensure the chat activities appear appropriately inside your Google Analytics.

  1. Copy the GA SECTION from the web plugin code and insert it directly after the opening tag on each page where you want the linking happen. However, if you already have Google Analytics linker set up, just add the domains to your existing linkers.
  2. Adding "juji.ai" to your Google Analytics referral exclusion list at Admin > Property Settings > Tracking Info > Referral Exclusion List

After everything is set up correctly, wait for a day for the Google Analytics change kicks in. Then you will see pageviews similar to /chat/606c1558-d290-4387-b870-8462ae3e3ee5. Such pageviews tell you those visitors have chatted with your Juji chatbot.

Quick Start with the Chat Activity Info

A good way to use the extra piece of chat activity info is to create two segments in your Google Analytics View - "chatted" and "not chatted", and use them to evaluate your chatbot's influence on your page.

To create segments, you go to Admin > View Settings > Segments > + NEW SEGMENT. For "chatted" segment, you edit the Seqences as shown below:

Configure chatted google analytics segment

For "not chatted" segment, you change the filter type from "Include" to "Exclude":

Configure not-chatted google analytics segment

Now, you can add the new segments in our metrics tables (Goals, Site Content etc.), and see how they perform.

Google Analytics site content all pages with chatted vs not-chatted segments

Activity Tracking with Google Tag Manager

The activity tracking setting also works with Google Tag Manager, but it will require some setup in Google Tag Manager. Please contact support at(@) juji.io if you need assistance on setting up activity tracking with your Google Tag Manager.

Generate URL

Use the Generate URL button to generate a web URL. This URL can then be embedded into an email or a website for target audience.

generate URL

Two URLs are generated:

  • Test Link Give this URL to testers so that the chat data can be identified as test data easily.

  • Web Link Use this URL for target audience.

generated two web URLs

Customize Web URL with External Data

In some cases, you may want to customize the generated chatbot URL with additional information for various purposes. For example, if you use a chatbot to conduct surveys or onboard customers, you may want to append a unique user id to the chatbot URL and send this URL to qualified participants or customers. You can also append additional information, such as source or session information, to track where the users obtained your chatbot URL.

It's super easy to pass such external information to Juji. You just need to append the information to the chatbot URL. For example, the following Juji URL is appended with ?source=email to identify that this chatbot URL will be emailed to the target audience:

https://juji.ai/pre-chat/5df866e6-e911-4924-b9e1-7440038825c6?source=email

In contrast, the following indicates the URL will be posted on Linkedin:

https://juji.ai/pre-chat/5df866e6-e911-4924-b9e1-7440038825c6?source=linkedin

Juji will automatically capture such external information in its reports. For example, two entries below from the CSV report file show the captured source information (see how to download a CSV report).

capture external information

Deploy as Web Widget

If you don't want your visitors to be redirected to another page, you can make your chatbot live on your page as a web widget. It will look similar to the Juji chatbot at the bottom right of this documentation page.

To add such widget, after you go through the web deployment process, instead of copying the URL, you copy the web plugin code that appears below the URL. Then, simply insert it directly after the opening <body> tag on each page where you want the widget to appear.

install web plugin

The web plugin code gets updated automatically with your activity tracking in step 2 of the web deployment process. If you have activity tracking enabled, you will see a "GA SECTION" in the web plugin code. The section is required for linking your site and the Juji's chat page. However, if you are already setting Google Analytics linkers, pleaser remove the GA section and add the linker domains into your existing linker setting.

Deploy to Facebook Page

One can deploy a Juji chatbot to a Facebook Messenger associated with a Facebook page.

Facebook Deployment

Similar to deploying to a website, one can also edit the weclome message shown up in a Facebook Messenger. Click on the green Checklist to make a Facebook page ready to host a chatbot.

Use the blue Connect with Facebook button to deploy a chatbot. Select the page(s) you want to deploy to. Then click on the green Deploy button to complete the deployment.

Select facebook page to deploy

If there is no page listed, it means that you have not given Juji permissions to any of your Facebook pages to host your chatbots. In that case, click on the green Connect More button to select Facebook pages and give permissions. The permitted pages should then show up in the list and you can then select one or more to deploy a chatbot to.

Update Deployed Chatbot

It's possible that a chatbot needs to be updated after its deployment. For example, you may want to add a new chatbot question or update the wording of an existing chatbot request. To push the new updates to the deployment, Juji supports two types of update:

  • Update Existing Release Use the Update button on the Facebook deployment page (see below) or the one under the Manage button to push the updates to the existing chatbot. The updated chatbot will restart and reflect the changes made.

Update Facebook Chatbot

After
successful facebook deployment


Update Web Chatbot

Update Web AI chatbot

  • New Release Use the Connect with Facebook button or New Release button under Manage to release a new version of the chatbot. A new release version will be created.

Best Practices

How do you decide whether to update a chatbot vs. to make a new release or version of the chatbot? While the decision is completely up to you, we recommend that you normally just update a chatbot without making a new release/version of it. If you make substantial changes, such as adding questions or attributes, to a chatbot, we recommend that you make a new release of the chatbot (basically with a new version number). This is because data fields (e.g., chatbot questions or custom attributes) could be very different between two releases and you don't wish to mix the old and new data together. Making a new release will also facilitate easier comparison of audience behavior under different chatbot versions.

Note that the current Reports page shows the audience information for the most recent release/version of chatbot. If you make a new release, currently you may not be able to access the audience information for previous releases.

What's Next

Once a chatbot is deployed, you can monitor its status and the associated audience information. Please check out Report Dashboard to deploy your AI chatbot.