Competitor Analysis

Competitor Analysis Using GenAI - Fortune 500 - CleverCompete

Codemonk applies its competitior analysis solution to the FMCG sector to study the performance of an SKU (beverage) against competition

Karan Kariappa

• 9 min read
Competitor Analysis Using GenAI - Fortune 500 - CleverCompete

To further our efforts in showcasing viable products and services, Codemonks, today, would like to showcase the applicability of our CleverCompete product suite in the FMCG sector, where staying ahead of the competition either makes or breaks your business. CleverCompete is GenAI-driven competitor analysis solution that leverages advanced machine learning techniques to provide actionable insights into competitor performance. This white paper details the customisation, deployment, and application of CleverCompete, demonstrating how it transforms publicly available data into strategic intelligence.

Introduction

The FMCG sector is characterized by rapid product cycles, diverse SKU portfolios, and intense competition. Traditional methods of competitor analysis often fall short in providing timely, comprehensive insights. CleverCompete fill in this gap by harnessing the power of large language models (LLMs) and natural language processing (NLP) to analyze vast amounts of public data and generate nuanced, actionable intelligence.

In summary, CleverCompete is an end-to-end solution that:

  1. Collects and processes data from various public sources
  2. Utilizes a fine-tuned LLM to analyze this data, wherein, refinement of LLM happens prior to the data analysis
  3. Generates detailed reports on competitor performance and market trends

CleverCompete is designed to be user-friendly, allowing marketing professionals and executives to feed in specific queries and receive tailored insights without needing technical expertise.

Lets look at how the solution does this when applied to a specific use case within the FMCG sector, as mentioned above. CleverCompete is trained on publicly available sources (such as websites and open repositories) and APIs (provided by the company using CleverCompete) that provide access to privately stored information if any. To do this, it categorises the types of information into the below categories.

Data Collection and Processing

Data Sources

CleverCompete draws from a wide range of public data sources, including:

  • Financial filings (e.g., 10-K, 10-Q reports)
  • Press releases and company announcements
  • News articles and media coverage
  • Social media posts and trends
  • Product launch information
  • Market research reports
  • Patent filings
  • Consumer reviews and feedback

The manner in which the data aggregation happens is split into 4 specific pipelines, as agreed upon in the initial deliberation with the client.

Data Collection Methodology

CleverCompete employs a multi-pronged approach to data collection:

  1. Web Scraping: Custom-built scrapers regularly collect data from predetermined websites and databases. Each of these sources can be configured or passed as queries when seeking specific insights
  2. API Integration: We integrate with various APIs (e.g., financial data providers, news aggregators) to ensure real-time data access. Privatised APIs can also be configured upon request and adherence to compliance laws.
  3. Natural Language Processing (NLP) Pipelines: These extract relevant information from unstructured text data, such as news articles and social media posts. The data obtained from the above channels will be parsed, indexed and contextualised as necessary for competitor analysis.
  4. Optical Character Recognition (OCR): This technology extracts text from images and PDFs, thereby collecting non-text based information stored across the above channels, ensuring no valuable information is missed.

All of the aggregated information is cleansed, and formatted in accordance with the LLM chain or model developed. In this case, a model built on 8 billion parameters is further fine tuned to the aforementioned use case and applicability. This process in tandem with data processing and preparation operations as mentioned below.

Data Preprocessing

Raw data undergoes several preprocessing steps:

  1. Cleaning: Removal of duplicates, irrelevant information, and noise.
  2. Structuring: Conversion of unstructured data into structured formats.
  3. Normalization: Standardization of data formats and units for consistency.
  4. Entity Recognition: Identification and tagging of relevant entities (e.g., company names, products, key personnel).
  5. Sentiment Analysis: Determination of sentiment in textual data.

Now all the information needed to run CleverCompete is ready. We can now look at fine tuning the LLM.

LLM Training and Fine-tuning

The fine tuning of LLM can be split into three categories of contemporary workflows that run simultaneously to ensure progressive, corrective actions to improve the overall product suite prior to delivery. On one stream, while we select the model and apply fine tuning methodologies, we create a library of prompts that act as a reference to train the LLM a per competitor analysis queries that could be passed by a user in the future. Here, the most important step to note is the inclusion of adaptive pre-training, supervised fine-tuning and reinforcement learning methodologies, in accordance with the prompt library created. Lets break down each of these steps:

1. Base Model Selection

We start with a state-of-the-art, pre-trained language model (e.g., GPT-3 or a similar architecture) as our foundation. This base model already possesses a broad understanding of language and general knowledge.

2. Domain-Specific Fine-tuning

To customise the model for FMCG competitor analysis, we fine-tune it using a carefully curated dataset. This dataset includes:

  • Historical competitor analysis reports
  • FMCG industry whitepapers and research papers
  • Annotated financial statements from FMCG companies
  • Expert-written summaries of market trends and competitive landscapes

The fine-tuning process involves:

  1. Adaptive Pre-training: Further pre-training on a large corpus of FMCG-specific texts to familiarize the model with domain-specific language and concepts.
  2. Supervised Fine-tuning: Training on expert-annotated examples of competitor analysis to teach the model to generate insights in the desired format and style.
  3. Reinforcement Learning: Optimizing the model's outputs based on feedback from FMCG experts, ensuring the generated insights are not just accurate but also valuable and actionable.

3. Prompt Engineering Library

We develop a library of carefully crafted prompts that guide the LLM to produce specific types of insights. These prompts are designed to:

  • Extract key performance indicators (KPIs) from financial data
  • Identify emerging trends in product development
  • Analyze marketing strategies across different channels
  • Compare pricing strategies among competitors

Example prompt: "Analyze the Q3 financial reports of competitors A and B, focusing on their beverage segment. Identify key differences in revenue growth, market share, and profitability. Highlight any new product launches or marketing initiatives that may have contributed to these differences."

Here is a sample output produced by the pretrained model based on the above conditions: (Click on the below toggle to expand the report)

CleverCompete Report - Fortune 500 - Beverages Segment

This analysis compares the Q3 performance of Competitor A and Competitor B in their beverage segments, focusing on revenue growth, market share, and profitability. Key findings include:

  • Competitor A showed stronger revenue growth (8.5% YoY) compared to Competitor B (3.2% YoY)
  • Competitor A increased market share to 28.5% (+1.5 points), while Competitor B's share declined to 24.8% (-0.7 points)
  • Competitor B maintained higher profitability (22% operating margin) compared to Competitor A (18% operating margin)
  • Competitor A's success is largely attributed to its new product launch and innovative marketing campaign

Detailed Analysis1. Revenue GrowthCompetitorQ3 RevenueYoY GrowthA$3.2B8.5%B$2.9B3.2%

Competitor A significantly outperformed Competitor B in revenue growth. This difference can be attributed to:

  1. Successful launch of Competitor A's new low-calorie, high-protein smoothie line "VitaBoost"
  2. Competitor A's expansion into emerging markets, particularly in Southeast Asia
  3. Competitor B's slower adaptation to changing consumer preferences towards healthier options

2. Market ShareCompetitorCurrent Market ShareChange (YoY)A28.5%+1.5 pointsB24.8%-0.7 points

Competitor A has made significant gains in market share, while Competitor B has seen a slight decline. Factors influencing this shift include:

  1. Competitor A's aggressive marketing campaigns, particularly in digital and social media channels
  2. Competitor B's loss of shelf space in key retail chains due to underperforming traditional soda products
  3. Competitor A's successful partnerships with popular fitness influencers, boosting brand visibility among health-conscious consumers

3. ProfitabilityCompetitorOperating MarginChange (YoY)A18%-1 pointB22%+0.5 points

Despite lower revenue growth, Competitor B maintains higher profitability. Key factors:

  1. Competitor B's ongoing cost optimization initiatives, including supply chain efficiencies
  2. Higher production costs associated with Competitor A's new product line
  3. Competitor A's increased marketing spend, impacting short-term profitability but potentially driving long-term growth

4. New Product Launches and Marketing InitiativesCompetitor A

  1. VitaBoost Smoothie Line
    • Launched mid-Q3
    • Targets health-conscious millennials and Gen Z
    • Contributes to 15% of Q3 beverage segment growth
  3. "Refresh Your Way" Campaign
    • Multi-channel marketing initiative focusing on personalization and lifestyle integration
    • 50% increase in social media engagement compared to previous campaigns
    • Successful collaborations with fitness influencers (combined reach of 50M+ followers)

Competitor B

  1. Relaunch of Classic Cola with Natural Sweeteners
    • Aimed at addressing health concerns while maintaining traditional flavor
    • Mixed consumer reception; 5% sales increase in health-focused retail channels, 2% decrease in traditional channels
  3. "Taste of Nostalgia" Marketing Campaign
    • Focused on brand heritage and emotional connection
    • 10% increase in brand sentiment among consumers aged 35+
    • Limited impact on younger demographics

Market Trends and Future Outlook

  1. Health and Wellness: The success of Competitor A's VitaBoost line indicates a growing consumer preference for functional beverages. Both competitors should consider expanding their healthier options.
  2. Sustainability: Neither competitor has significantly addressed sustainability in their Q3 initiatives. This represents a potential area for differentiation in future quarters.
  3. Digital Engagement: Competitor A's success with digital marketing suggests that Competitor B may need to reevaluate and enhance its digital strategy to regain market share.
  4. Emerging Markets: Competitor A's growth in Southeast Asia highlights the importance of strategic international expansion. Competitor B should assess opportunities in high-growth markets.

Recommendations

  1. Competitor A should focus on improving operational efficiency to boost profitability without sacrificing growth momentum.
  2. Competitor B needs to accelerate product innovation, potentially through acquisitions or partnerships with startups in the functional beverage space.
  3. Both competitors should invest in sustainability initiatives to meet growing consumer expectations and potentially unlock new market segments.
  4. Competitor B should revamp its digital marketing strategy, possibly by adopting more personalized and interactive campaigns similar to Competitor A's approach.

This analysis is based on publicly available Q3 financial reports and market data. For a more comprehensive understanding, it is recommended to correlate these findings with point-of-sale data, consumer surveys, and regional market analyses.

System Architecture and Workflow

To arrive at the above, report, based on the dataset and prompt provided at the start as guidance functions, we need to understand the inner workings of CleverCompete.

The GenAI-based solution operates this wireframe at an architectural level:

  1. Data Ingestion: Fresh data is continuously collected and preprocessed.
  2. Query Input: Users input their specific analysis requirements through a user-friendly interface.
  3. Context Preparation: Relevant data is selected and formatted based on the user's query.
  4. LLM Processing: The fine-tuned LLM receives the user's query and the prepared context data as input.
  5. Insight Generation: The LLM processes the inputs and generates a detailed analysis.
  6. Post-processing: The system applies additional formatting, fact-checking against the source data, and generates visualizations to enhance the insights.
  7. Delivery: The final report is presented to the user through an interactive dashboard.

At the end of this wireframe, CleverCompete generates comprehensive reports that include:

  • Executive summaries of competitor performance
  • Detailed breakdowns of financial metrics and their implications
  • Analysis of product portfolio changes and their market impact
  • Insights into marketing strategies and their effectiveness
  • Predictions of future market trends and competitor moves
💡
Example of specific Insights: "Competitor A's beverage segment showed a 15% YoY growth, outpacing Competitor B's 8%. This can be attributed to A's successful launch of a new low-calorie energy drink line, which captured 5% market share within its first quarter. A's aggressive social media marketing campaign, particularly on TikTok, played a crucial role in this success, suggesting a shift in effective marketing channels for the 18-25 demographic."

Continuous Improvement

CleverCompete is designed to evolve and improve over time:

  • Feedback Loop: Users can rate the usefulness of insights, which feeds back into the training process.
  • Automated Evaluation: The system regularly compares its predictions against actual market outcomes to refine its models.
  • Expert Review: FMCG analysts, entrusted internally, periodically review and validate the system's outputs, providing guidance for further improvements.

Privacy and Ethical Considerations

CleverCompete is built with a strong commitment to ethical AI practices:

  • It only uses publicly available data, ensuring no breach of confidentiality.
  • The system is regularly audited for biases and adjusted to ensure fair representation.
  • All insights are presented with confidence levels and source citations, promoting transparency.

Case Study: Exploring a specific SKU for FMCG - Navigating the Carbonated Drinks Market

To illustrate the power of CleverCompete, consider how it helped a Fortune 500 FMCG company navigate a challenging period in the carbonated drinks market.

In Q2 2023, their main competitor launched a new line of natural, low-sugar sodas that quickly gained traction. CleverCompete analyzed social media trends, sales data, and consumer reviews to provide the following insights:

  1. The competitor's product was particularly popular among health-conscious millennials.
  2. Its success was largely driven by an influencer marketing campaign on Instagram and TikTok.
  3. Despite initial popularity, there were emerging concerns about the aftertaste of the natural sweeteners used.

Armed with these insights, the Fortune 500 FMCG company was able to:

  1. Fast-track the development of their own line of low-sugar sodas.
  2. Design a marketing campaign targeting the same demographic but addressing the identified taste concerns.
  3. Partner with food scientists to develop a proprietary natural sweetener blend with improved taste.

As a result, when they launched their competing product line in Q4 2023, they were able to quickly capture market share, with their sales exceeding the competitor's by 20% in the first month of launch.

Key Takeaways

Observing the adoption of CleverCompete, we firmly believe that our GenAI capabilities act as rocket fuel that ensures that our clients, such as the fortune 500 company, could achieve a significant leap forward just using competitor analysis. Refining CleverCompete Regularly, and also introducing multiple sources of information, Codemonk provides unparalleled insights that drive strategic decision-making for any potential adopter.

For more information or to schedule a demo, please contact us at https://codemonk.io/contact