Skip to main content

Contextual Messaging - Getting Started

Understand the implementation steps for Contextual Messaging

Updated yesterday

Overview

Jacquard Contextual Messaging aim to solve for true one-to-one message personalisation. To implement Contextual Messaging, there are a few required steps:

  1. Configuring your data source (i.e. providing Jacquard with your product details and general customer information)

  2. Setting up your campaign in the Jacquard platform

  3. Connecting our Contextual Messaging API to your customer engagement platform (CEP)

  4. Reviewing and approving new contextualised language

How Jacquard generates language

The key to successfully delivering contextualised content is providing targeted, relevant language where and when it matters.

Week after week, our system runs evaluations of campaign metrics to understand where contextualisation makes the biggest impact. Learning from your product, audience, and engagement data, Jacquard crafts messages that resonate with your specific customers while staying true to your brand voice.

This iterative process enables real-time monitoring of generated content and audience engagement analytics.

Data configuration

Before Jacquard can start generating language, it must have at least a basic understanding of your products and customer attributes. Navigate to the Admin > Data sources section in our application to start providing this crucial data.

Product catalogue

Jacquard requires a structured product database to generate language. Each entry in the product database must contain a unique product_identifier, accompanied by a comprehensive description of your product and up to four categorical classifications. You can provide this data via CSV import.

Customer attributes

We've designed a flexible system that respects your customers' privacy while capturing what makes them unique. Customer attributes must be designated with the profile_ prefix to ensure proper system recognition.

For instance, customer segmentation such as loyalty status might be formatted as profile_loyalty = gold. Jacquard will capture each unique value via your API requests and store them independently of any customer identifier.

Each attribute requires corresponding descriptive documentation to facilitate accurate language generation. You're able to provide these descriptions in the Data source section via CSV import.

1. Create a customer attribute catalogue

Before setting up a contextualised campaign, users must first establish a record to store their customer attributes.

Navigate to the Customer attributes tab and click Create record.

Ensure you choose the proper privacy region from the options configured in your account.

You can either generate a unique record for each campaign or use an existing one.

Once the record is created, attributes will be progressively added as the API receives data labelled with the profile_ prefix.

2. Manage attribute descriptions

Once customer attributes are populated in a record, users can manage attribute descriptions. Providing detailed and well-structured descriptions will enable the Contextual Messaging system to generate language that resonates with the specified audience.

How to write an effective attribute description

1. Describe the attribute

  • Clearly explain what the attribute represents for your brand

  • Example:

    • Audience Segment: This attribute categorises users based on shared characteristics and demographics.

2. Define each value

  • Provide a clear description of each attribute value, specifying its criteria.

  • Examples:

    • Young Professionals: Users aged 25-35, primarily in urban areas with professional occupations.

    • High Engagement: Users who log in daily and interact with at least five pieces of content per week.

3. Highlight special cases

  • Note any exceptions or unique scenarios that may impact classification.

  • Example:

    • Excludes users aged 25-35 who are students or unemployed.

4. Keep it simple and clear

  • Use concise language and consistent formatting to enhance readability.

Did this answer your question?