The Challenge of Social Commerce: Effort vs. Reward

Despite the clear advantages in conversion rates, selling directly on social media and messaging apps remains an undervalued sales channel. Insights gathered from hundreds of customer conversations and the performance data of thousands of merchants on our platform reveal that social media and messaging apps significantly outperform other sales channels regarding conversion rates.

Selling on social media and messaging apps requires more work for customer engagement and support. As discussed at the Meta Business Messaging Summit in April 2024, businesses must invest extra effort in platforms like Facebook, Messenger, and Instagram, in addition to their existing activities on marketplaces and webstores. Consequently, many businesses prioritize marketplaces where most sales volume originates, limiting investments in social media and messaging apps.

However, this landscape is evolving. More consumers are now purchasing directly through social media and messaging apps, making it increasingly important for businesses to optimize their presence on these channels despite the initial higher engagement cost.

‍

Bringing Back the Genie with Copilot

Selling on social media and messaging apps demands high-touch, proactive efforts in content creation, community engagement, promotions, and paid marketing. Unlike marketplaces, where businesses can participate in large marketing campaigns by paying higher commission fees, social commerce requires more direct interaction. The challenge is to maximize sales on these platforms with limited resources. How can businesses achieve higher sales on social media and messaging apps efficiently?

To help businesses maximize sales on social media and messaging apps with limited resources, we developed ChatGenie Copilot. This AI assistant supports merchants in several key areas:

1. Content Creation: Generates engaging posts for social media and messaging apps and updates product catalogs.
2. Promotion and Upselling: Utilizes current offerings configured in the ChatGenie dashboard to boost sales.
3. Community Engagement Automation: Automatically responds to order-related inquiries in comments and direct messages.
4. Ad Campaign Recommendations: Analyze data from previous campaigns and community interactions to suggest effective ad strategies.

‍

The Sweet Spot for Using Copilot

As an eCommerce enabler within social media and messaging apps, ChatGenie connects businesses' social media assets, like Facebook and Instagram Pages, to our platform. This integration allows us to launch In-App Stores within Messenger and Instagram, leveraging rich contextual data from social media interactions, merchant dashboards, and actual transactions. This comprehensive access enables ChatGenie Copilot to effectively assist merchants in managing and promoting their businesses. Here are some key capabilities of ChatGenie Copilot:

1. Brand Tone and Style: Automatically adapt content creation and automated reply copies to match the business's existing social media posts without requiring manual configuration from the merchant.

2. Purchasable Product Links: Generates purchase flows and links based on the business's post captions and images, making it easier for customers to buy directly from social media.

3. Target Audience Profiles: Creates detailed audience profiles using data from social media engagement, purchase history, product descriptions, tags, and campaigns, helping tailor marketing efforts more effectively.

Our prompt design and parameters continually evolve based on merchant and customer interactions with the preview release of ChatGenie Copilot, ensuring we provide the most relevant and effective support.

‍

Target Audience Profile Agent Design

What sets ChatGenie Copilot apart is its unique implementation design, which uses multiple agents to roleplay the target audience profiles for each merchant on the platform. This approach ensures that the generated content and recommendations are highly relevant and tailored to each merchant's specific customer base.

For instance, let's consider Vans as a sample merchant. Based on Vans’ Facebook and Instagram posts, along with available training data from GPT-4, we’ve identified four main target audience profiles that Vans focuses on for content marketing. Below is a detailed description of each target customer profile:

Enhancing Content Appeal with Target Audience Profile Scores

A key feature of the defined target audience profiles is the ability to measure how well the generated content or responses align with the intended audience through a Target Audience Profile Appeal Score. This score ensures that Copilot’s output resonates with the target audience, reducing the guesswork for merchants in various areas such as ad campaign recommendations, content creation, product description generation, welcome and greeting messages, and purchase flows.

This multi-agent design aids merchants in optimizing their social media and messaging app strategies by providing clear, data-driven insights into the effectiveness of their content. Here’s a sample output of a Target Audience Profile Agent assessing a Target Audience Appeal Score for published content creation.

Reintroducing Chatbot Functionality with ChatGenie Copilot

Since 2020, we’ve shifted our focus from "chatbot" terminology to offering In-App Stores. Despite this, many still associate ChatGenie with chatbot functionality. With the advancements in AI and the increasing demand for automated interactions, reinstating chatbot capabilities is now a priority on the ChatGenie Copilot roadmap.

Our initial tests show promising results using smaller models like Llama3 to manage conversations and identify inquiries with strong purchase intent. While we primarily use GPT-4 for most Gen AI features, Llama3 offers a cost-effective solution for chat-specific use cases. We've evaluated several large language models (LLMs) and decided on Llama3 for its balance of performance and cost when hosted on a GPU-optimized instance.

Expect a detailed blog post soon, where we’ll discuss into our choice of LLMs and the design considerations for enhancing ChatGenie Copilot with robust chatbot functionality.

Our Goal: Transition from Copilot to Autopilot

Currently, all generative AI features on the dashboard require merchant approval before being applied to their In-App Stores. This is due to the inherent unpredictability and varying accuracy of generated outputs. We aim to refine our AI prompts to achieve a 90% acceptance rate for Copilot-generated content. Once we reach this level of reliability, we will begin implementing Autopilot mode, allowing AI recommendations to be automatically applied to each merchant’s In-App Store without the need for manual approval.

‍

What to expect from ChatGenie Copilot Preview Release?

We are just getting started and there are still a lot of Gen AI features currently in development and are being tested privately and publicly. For merchants that already have access to their dashboard, they can use the following features for free while we are in the preview release:

1. Welcome Message Generation

2. Greeting Message Generation

3. Product Description Generation

4. Comment Auto-Reply Auto-Configuration

5. Copilot Recommendation Section

5. Announcement Post Generation

6. Dynamic Comment Reply Generation

We are testing certain use cases that are fitting to the chatbot feature. If your business receives hundreds of inquiries on public comments and direct messages daily reach out to m.me/ChatGenie and we are looking forward to discussing how future releases of ChatGenie Copilot can help you.

‍