Introduction: The Rise of AI-Powered Brand Monitoring
In the digital age, social media has become a critical battleground for brands. Monitoring brand perception, understanding customer sentiment, and identifying potential crises in real-time are essential for maintaining a positive brand image and driving business success. However, the sheer volume of social media data, particularly images and videos, makes manual monitoring impractical. This is where Artificial Intelligence (AI) steps in, offering powerful tools for automated image and video analysis. This article provides a comprehensive guide to building scalable AI-powered image and video analysis pipelines for social media brand monitoring, enabling businesses to gain valuable insights and stay ahead of the curve.
As the capabilities of AI language models evolve beyond simple text generation, their application in analyzing visual content for brand monitoring becomes increasingly sophisticated. Techniques initially developed for understanding natural language, such as transformer networks, are now being adapted for AI image analysis and AI video analysis. For example, models trained on massive datasets of images and videos can identify subtle cues in visual content that might indicate a shift in brand sentiment. This goes beyond simple object detection to encompass nuanced understanding of context and emotion, mirroring the advancements seen in areas like machine learning in weather prediction, where models are learning to interpret complex atmospheric patterns beyond traditional forecasting methods.
The ability to discern these subtle signals is crucial for proactive crisis management and maintaining a competitive edge. Cloud-based AI solutions are revolutionizing social media brand monitoring by providing the scalability and computational power needed to process vast amounts of visual data in real-time. Platforms like AWS, Google Cloud, and Azure offer pre-trained AI models and customizable tools that enable businesses to build robust AI image analysis and AI video analysis pipelines without the need for extensive in-house AI expertise.
According to a recent report by Forrester, adoption of cloud-based AI for brand monitoring is expected to increase by 40% in the next two years, driven by the need for faster insights and reduced operational costs. This shift towards cloud-based AI is also enabling smaller businesses to leverage the power of AI, democratizing access to advanced brand monitoring capabilities. The application of sentiment analysis to visual content represents a significant leap forward in social media brand monitoring.
By combining AI image analysis and AI video analysis with sentiment scoring algorithms, businesses can gain a more holistic understanding of how their brand is perceived online. For instance, an AI model might identify a trending meme featuring a company’s product and then analyze the comments and reactions associated with that meme to determine whether the overall sentiment is positive, negative, or neutral. This granular level of insight allows brands to tailor their marketing strategies and respond effectively to emerging trends and potential crises. As AI language models continue to evolve, their ability to understand and interpret the nuances of human emotion in visual content will become even more critical for effective brand management.
Defining Key Performance Indicators (KPIs) for Brand Monitoring
Before diving into the technical aspects of building scalable AI-powered image and video analysis pipelines, it’s crucial to define the Key Performance Indicators (KPIs) that will drive the social media brand monitoring strategy. These KPIs should align with the overall business objectives and provide actionable insights into brand perception, market trends, and competitive positioning. Selecting the right KPIs ensures that the subsequent AI image analysis and AI video analysis efforts are focused and yield measurable results.
For instance, if a primary business objective is to increase brand awareness among a younger demographic, the KPIs should reflect this goal, perhaps by tracking brand mentions and sentiment within specific age groups or on platforms popular with younger users. Without clearly defined KPIs, the entire brand monitoring initiative risks becoming a data swamp, yielding information that is interesting but ultimately lacks strategic value. This is particularly relevant in fields like weather prediction, where poorly defined metrics can lead to misinterpretations of model performance and ultimately, inaccurate forecasts.
Therefore, a robust KPI framework is the bedrock of any successful AI-driven brand monitoring system. Sentiment analysis is a cornerstone KPI, gauging the overall sentiment (positive, negative, neutral) associated with the brand across social media platforms. This is often achieved through Natural Language Processing (NLP) techniques applied to text accompanying images and videos. However, modern approaches extend beyond simple keyword analysis. AI language models, evolving beyond the capabilities of even advanced systems like ChatGPT and Claude, can now discern nuanced emotional tones, sarcasm, and cultural contexts that would previously be missed.
For example, an AI model might recognize that a seemingly positive statement like “Their new product is surprisingly good!” actually conveys a degree of initial skepticism, indicating a potential area for improvement in pre-launch marketing. Furthermore, sentiment analysis can be enhanced by integrating visual cues; an image of a frowning face alongside a brand mention can provide additional context to the textual sentiment. This holistic approach to sentiment analysis provides a more accurate and actionable understanding of brand perception.
Logo detection is another essential KPI, identifying instances of the brand’s logo appearing in images and videos, even when not explicitly mentioned in the text. Object detection goes a step further, recognizing specific products or elements associated with the brand in visual content. These capabilities are crucial for identifying unauthorized use of brand assets, tracking product placement in user-generated content, and understanding how the brand is visually represented across different social media channels. Consider a scenario where a competitor subtly incorporates elements of your brand’s visual identity into their marketing materials.
AI-powered logo and object detection can quickly identify these instances, allowing for swift action to protect brand integrity. This proactive approach is far more effective than relying on manual monitoring or waiting for customer complaints. Moreover, advancements in neural network evolution, moving beyond large language models, are enabling more robust and accurate object and logo detection, even in challenging conditions such as poor lighting or partial occlusion. Competitor analysis extends the scope of brand monitoring to the broader market landscape.
By monitoring competitor brands’ mentions, sentiment, and visual presence, businesses can understand market trends, identify competitive threats and opportunities, and benchmark their own performance. Contextual understanding is also vital, identifying the context in which the brand is mentioned, including events, campaigns, or trending topics. For instance, a sudden spike in negative sentiment following a competitor’s product launch might indicate an opportunity to highlight the strengths of your own offering. Finally, image quality assessment determines the quality and relevance of images and videos associated with the brand, helping filter out irrelevant or low-quality content that could skew sentiment analysis or other KPIs. This ensures that the data used for analysis is reliable and representative of genuine brand perception. The implementation of cloud-based AI solutions is crucial for managing the computational demands of these analyses, providing the scalability and processing power needed to handle vast amounts of social media data in real-time.
Selecting Appropriate AI Models for Image and Video Analysis Tasks
Selecting the right AI models is critical for achieving accurate and reliable results in social media brand monitoring. Several AI models are suitable for image and video analysis tasks, each with its strengths and weaknesses. Object Detection models, such as YOLO (You Only Look Once), SSD (Single Shot MultiBox Detector), and Faster R-CNN, are effective for identifying objects within images and videos. YOLO prioritizes speed, making it suitable for real-time analysis, while Faster R-CNN often provides higher accuracy, albeit at a computational cost.
Facial Recognition, powered by models like FaceNet and DeepFace, can identify and verify faces, crucial for understanding demographics and identifying key individuals featured in brand-related content. Scene Understanding models, including Places365 and SceneParseNet, classify the environment depicted, adding context to the visual data. Image Classification models like ResNet, Inception, and EfficientNet categorize images into predefined classes, while Action Recognition models such as TimeSformer and SlowFast Networks recognize actions and events within videos. Sentiment Analysis, traditionally used for text, can also be applied to analyze facial expressions and visual cues in videos to infer emotional responses, adding another layer to social media brand monitoring.
When comparing models for AI image analysis and AI video analysis, accuracy and performance are paramount. Accuracy should be evaluated on relevant datasets; for instance, the COCO dataset is a standard benchmark for object detection, while specialized datasets exist for facial recognition and action recognition. Performance considerations include inference speed and computational requirements, which are crucial for cloud-based AI deployments designed to handle the high volume of social media data. The choice often involves a trade-off: highly accurate models may require more computational resources, potentially increasing costs and latency.
Fine-tuning pre-trained models on custom datasets is often necessary to optimize performance for specific brand monitoring tasks. Transfer learning, leveraging knowledge gained from large general datasets, can significantly reduce training time and improve accuracy on smaller, more specialized datasets relevant to a particular brand or industry. The evolution of neural networks has significantly impacted the landscape of AI models applicable to social media brand monitoring. Transformer-based models, initially developed for natural language processing, are increasingly being adapted for image and video analysis, demonstrating state-of-the-art performance in various tasks.
For instance, vision transformers (ViT) have achieved remarkable results in image classification, rivaling convolutional neural networks (CNNs). Furthermore, techniques like self-supervised learning are gaining traction, allowing models to learn from unlabeled data, which is particularly valuable given the vast amount of unannotated images and videos on social media. This approach enables the development of more robust and generalizable models, capable of handling the diverse and ever-changing content found on social media platforms. The ability to leverage these advancements through cloud-based AI services ensures scalability and accessibility for brand monitoring applications.
Beyond the model architecture, the choice of training data and mitigation of biases are critical considerations. AI models are only as good as the data they are trained on, and biases in the training data can lead to skewed or unfair outcomes. For example, if a facial recognition model is trained primarily on images of one demographic group, it may perform poorly on others. Therefore, careful attention must be paid to data diversity and fairness. Techniques like adversarial training and data augmentation can be used to mitigate biases and improve the robustness of AI models. Moreover, explainable AI (XAI) methods are becoming increasingly important, allowing businesses to understand and interpret the decisions made by AI models, which is crucial for building trust and ensuring ethical use of AI in social media brand monitoring.
Designing a Scalable Architecture Using Cloud-Based Services
To effectively manage the escalating volume of social media data, particularly images and videos, a scalable cloud-based architecture is not just beneficial, but essential for any serious social media brand monitoring endeavor. Cloud providers like AWS, Google Cloud, and Azure offer a comprehensive suite of services that facilitate the construction of robust and cost-effective pipelines for AI image analysis and AI video analysis. These pipelines enable real-time data ingestion, scalable storage, powerful data processing, and streamlined AI model deployment, all crucial for extracting actionable insights.
Selecting the appropriate cloud provider often depends on existing infrastructure, budget, and specific AI model requirements, but all three offer viable solutions. Data ingestion forms the critical first step, capturing the torrent of social media posts as they occur. Services such as AWS Kinesis, Google Cloud Pub/Sub, and Azure Event Hubs are designed to handle high-velocity, high-volume data streams from diverse sources. These services not only ingest the data but also provide mechanisms for buffering and routing it to downstream processing components.
Furthermore, the integration with AI language models allows for preliminary filtering based on keywords or hashtags, focusing subsequent AI image analysis and AI video analysis on the most relevant content. Beyond simple keyword filtering, sophisticated sentiment analysis, pre-processing can be conducted at the ingestion layer, enabling a more nuanced understanding of the incoming data stream. Once ingested, the raw image and video data requires scalable and durable storage. Cloud storage services like AWS S3, Google Cloud Storage, and Azure Blob Storage offer virtually unlimited storage capacity at a relatively low cost.
These services also provide features like versioning, lifecycle management, and access control, ensuring data integrity and security. Moreover, the seamless integration with cloud-based AI services simplifies the process of accessing and processing the stored data. For example, AWS S3 integrates directly with SageMaker, allowing AI models to be trained and deployed on images and videos stored in S3 without the need for complex data transfer operations. These storage solutions are not merely repositories; they are active components in the AI pipeline.
Finally, the heart of the pipeline lies in data processing and AI model deployment. Cloud-based compute services like AWS EC2, Google Cloud Compute Engine, and Azure Virtual Machines provide the computational power needed to run complex AI models and perform data preprocessing tasks. Services like AWS SageMaker, Google Cloud AI Platform, and Azure Machine Learning offer specialized tools for deploying and managing AI models at scale, including features like model versioning, A/B testing, and auto-scaling. These platforms also provide pre-built AI models for common tasks like object detection, facial recognition, and sentiment analysis, reducing the development effort required to build a custom solution. The evolution of neural networks beyond large language models, such as convolutional neural networks (CNNs) for image processing and recurrent neural networks (RNNs) for video analysis, are constantly being refined and made more accessible through these cloud platforms, driving innovation in social media brand monitoring.
Implementing Data Preprocessing Techniques for Improving AI Model Accuracy
Data preprocessing is a crucial step in improving the accuracy of AI models designed for social media brand monitoring. Social media data, particularly images and videos, is inherently noisy and inconsistent, which can negatively impact model performance. This noise stems from various sources, including low-resolution images, poor lighting conditions, irrelevant backgrounds, and superimposed text or graphics. Addressing these issues through meticulous preprocessing is paramount to ensuring that AI models can effectively extract meaningful insights and perform tasks like sentiment analysis and brand logo detection with high precision.
Without proper preprocessing, even the most sophisticated AI algorithms will struggle to deliver reliable results, leading to inaccurate brand perception analysis and potentially flawed decision-making. Common preprocessing techniques include noise reduction, image enhancement, resizing and cropping, data augmentation, and normalization. Noise reduction techniques, such as Gaussian blur or median filtering, can smooth out images and videos, removing irrelevant details that might confuse the AI model. Image enhancement techniques, like contrast stretching or histogram equalization, improve the clarity and visibility of important features.
Resizing and cropping ensure that all images and videos are of a consistent size and aspect ratio, which is often a requirement for many AI models. Data augmentation generates additional training data by applying transformations like rotations, flips, and zooms, which helps the model generalize better to unseen data. Normalization scales pixel values to a consistent range, which can improve the stability and convergence of the AI model during training. These techniques are particularly relevant in the context of AI Language Models: Neural Network Evolution Beyond Large Language Models, as they directly impact the quality of the input data used to train and refine these advanced models.
For instance, a neural network trained on preprocessed images is more likely to accurately identify subtle brand cues or emotional expressions. Beyond these standard techniques, more advanced preprocessing methods are emerging, particularly in the realm of video analysis. These include frame stabilization to compensate for shaky camera footage, object tracking to isolate and follow specific elements within a video, and scene detection to segment videos into meaningful units. Furthermore, the evolution of AI language models is impacting preprocessing itself.
Generative models can now be used for intelligent image inpainting to remove unwanted objects or fill in missing regions, and for super-resolution to enhance the resolution of low-quality images. Consider a scenario where an AI model is used to detect unauthorized use of a brand’s logo in user-generated content. Preprocessing steps might involve automatically cropping the image to focus on the area of interest, enhancing the contrast to make the logo more visible, and then using a generative model to remove any distracting background elements.
Here’s an example using Python with OpenCV:
python
import cv2 # Load the image
image = cv2.imread(‘image.jpg’) # Convert to grayscale
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY) # Apply Gaussian blur
blurred = cv2.GaussianBlur(gray, (5, 5), 0) # Enhance contrast
clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
enhanced = clahe.apply(blurred) # Display the processed image
cv2.imshow(‘Enhanced Image’, enhanced)
cv2.waitKey(0)
cv2.destroyAllWindows() This code snippet demonstrates a basic image enhancement pipeline using OpenCV. It loads an image, converts it to grayscale, applies Gaussian blur for noise reduction, enhances contrast using Contrast Limited Adaptive Histogram Equalization (CLAHE), and then displays the processed image. This is a foundational example, and more sophisticated preprocessing pipelines can be built using a combination of OpenCV functions and custom AI models.
Addressing Ethical Considerations and Bias Mitigation in AI-Driven Social Media Analysis
AI-driven social media analysis raises profound ethical concerns, particularly regarding privacy, bias, and transparency. It’s essential to address these proactively, moving beyond mere compliance to establish genuine ethical frameworks. Privacy considerations demand rigorous adherence to regulations like GDPR and CCPA, but also necessitate a ‘privacy-by-design’ approach. This means embedding privacy safeguards into the very architecture of our AI systems for social media brand monitoring, including anonymization techniques and differential privacy to protect user identities while still gleaning valuable insights.
Obtaining explicit consent, when legally required, is just the starting point; transparent communication about data usage is paramount to building trust. Bias mitigation is arguably the most complex challenge. AI models, especially those used in AI image analysis and AI video analysis, can inherit biases from skewed training datasets, leading to unfair or discriminatory outcomes. For example, sentiment analysis models trained primarily on Western datasets may misinterpret or negatively misclassify expressions of emotion from other cultures.
To combat this, we must employ techniques like adversarial debiasing, data augmentation with diverse datasets, and fairness-aware machine learning algorithms. Furthermore, we need to actively monitor model performance across different demographic groups, using fairness metrics like equal opportunity and demographic parity to identify and rectify biases. The evolution of neural networks, moving beyond large language models, offers potential solutions through architectures designed for inherent fairness. Transparency demands clarity about how AI is used to analyze social media data.
This includes explaining the purpose of the analysis, the types of data collected, and the AI models employed. Black-box AI systems are unacceptable; explainable AI (XAI) techniques are crucial for understanding model decisions and identifying potential biases. For instance, providing visual explanations for why an AI image analysis model flagged a particular image as negative can build trust and facilitate human oversight. Accountability requires establishing clear lines of responsibility for the development, deployment, and monitoring of AI systems in social media brand monitoring.
Regular audits of AI systems are essential to ensure they operate ethically and responsibly, conforming to predefined ethical guidelines and performance benchmarks. Implementing fairness metrics during model training is a concrete step, allowing us to detect and mitigate bias in sentiment analysis proactively. Cloud-based AI solutions offer scalability, but also necessitate careful consideration of data security and governance. The future of ethical AI in social media brand monitoring hinges on a multi-faceted approach encompassing technical innovation, ethical frameworks, and ongoing monitoring.
Showcasing Real-World Examples and Case Studies of Successful Brand Monitoring Implementations
Several companies have successfully implemented AI-powered brand monitoring solutions, demonstrating the tangible benefits of these advanced technologies. These real-world examples showcase how businesses are leveraging AI to gain a competitive edge in understanding and managing their brand perception. Consider the case of an automotive brand that integrated AI image analysis into its social media brand monitoring strategy. By training a custom model on images of vehicle accidents and reckless driving, they could proactively identify user-generated content depicting their vehicles in unsafe situations.
This allowed them to engage with users, offer safety advice, and even alert authorities when necessary, mitigating potential legal and reputational risks. This proactive approach goes beyond simple keyword monitoring and delves into understanding the context surrounding brand mentions, a capability far exceeding that of traditional methods. Another compelling example comes from the fashion retail sector. A leading fashion retailer utilizes AI video analysis to identify emerging trends on platforms like TikTok and Instagram Reels.
Their AI models are trained to recognize subtle style cues, color palettes, and garment features that are gaining traction among influencers and early adopters. This allows them to anticipate future demand, adjust their inventory accordingly, and even collaborate with designers to create new products that align with the latest trends. The speed and accuracy of this AI-driven trend forecasting provide a significant advantage over competitors relying on traditional market research methods, showcasing the power of AI in adapting to rapidly changing consumer preferences.
This application moves beyond basic sentiment analysis, employing sophisticated computer vision techniques to extract actionable insights from visual data. A global beverage company provides a robust case study on the power of cloud-based AI for social media brand monitoring. This company implemented a comprehensive, cloud-based AI pipeline to analyze millions of social media images and videos daily. Their system incorporates sentiment analysis, object detection (identifying their products in images and videos), and logo recognition. During a controversial marketing campaign, the system detected a significant spike in negative sentiment across various social media platforms.
Furthermore, AI image analysis revealed that consumers were altering images of their products to express their dissatisfaction. By quickly identifying the root cause of the negative feedback and adjusting their marketing strategy, the company was able to mitigate potential damage to their brand reputation and prevent a significant drop in sales. This demonstrates the importance of a scalable and adaptable AI solution that can handle large volumes of data and provide real-time insights. These examples highlight the evolution beyond simple keyword-based monitoring.
The integration of AI language models, particularly those evolving beyond the capabilities of standard large language models, allows for nuanced understanding of context and sentiment. Furthermore, the application of machine learning in identifying patterns and predicting trends goes beyond traditional forecasting methods. The ability to analyze images and videos adds a critical visual dimension to brand monitoring, providing insights that would be impossible to obtain through text analysis alone. As AI models continue to evolve and cloud-based AI platforms become more accessible, we can expect to see even more innovative and impactful applications of AI in social media brand monitoring.
Conclusion: The Future of AI in Social Media Brand Monitoring
Building scalable AI-powered image and video analysis pipelines for social media brand monitoring is a complex but rewarding endeavor. By carefully defining KPIs, selecting appropriate AI models – moving beyond the limitations of standard LLMs like ChatGPT and Claude to specialized neural network architectures – designing a scalable cloud architecture, implementing data preprocessing techniques, and addressing ethical considerations, businesses can gain valuable insights into their brand perception and stay ahead of the competition. As AI technology continues to evolve, the possibilities for social media brand monitoring will only continue to expand, mirroring advancements seen in fields like machine learning for weather prediction, where sophisticated models are constantly refining forecasting accuracy.
The key is to leverage the right tools and strategies for the specific challenges presented by visual data. The future of AI in social media brand monitoring lies in the evolution of neural network architectures tailored for visual content. While large language models excel at text-based sentiment analysis, AI image analysis and AI video analysis require specialized models capable of discerning nuanced visual cues. For instance, convolutional neural networks (CNNs) can identify brand logos and product placements within images and videos, while recurrent neural networks (RNNs) can analyze video sequences to detect changes in sentiment over time.
Cloud-based AI platforms are crucial for deploying these models at scale, providing the computational resources necessary to process vast amounts of social media data in real-time. This mirrors the infrastructure used in complex weather models, where massive datasets are processed to predict future conditions. Consider a case study where a beverage company utilizes AI to monitor social media for mentions of their brand alongside images of competitors’ products. By training a custom object detection model, they can identify instances where consumers are choosing alternatives and analyze the context of those choices through sentiment analysis.
This allows them to understand the reasons behind customer preferences and tailor their marketing strategies accordingly. Furthermore, advancements in generative AI could enable the creation of synthetic data for training these models, mitigating bias and improving accuracy, similar to how synthetic data is used to augment training datasets in weather prediction models. The convergence of these technologies promises a new era of precision and insight in social media brand monitoring. Ethical considerations are paramount as AI-driven social media analysis becomes more sophisticated.
Ensuring fairness and transparency in AI models is crucial to avoid perpetuating biases present in the training data. Techniques such as adversarial training and explainable AI (XAI) can help mitigate these risks. Moreover, adhering to privacy regulations and obtaining informed consent from users are essential for maintaining trust and ethical standards. The responsible deployment of AI in social media brand monitoring requires a commitment to ethical principles and a proactive approach to addressing potential biases and privacy concerns. This mirrors the ethical considerations in fields like weather prediction, where accurate and unbiased forecasts are critical for public safety.
Looking Ahead: Future Trends and Advancements
The convergence of AI, cloud computing, and social media analytics is revolutionizing how brands understand and manage their image. By embracing these technologies, companies can unlock powerful insights, mitigate risks, and drive business success in the ever-evolving digital landscape. As AI models become more sophisticated and cloud services become more accessible, the barrier to entry for AI-powered brand monitoring will continue to decrease, making it an essential tool for businesses of all sizes. Future advancements will likely include more sophisticated sentiment analysis, enhanced contextual understanding, and improved bias mitigation techniques, further enhancing the value of AI in social media brand monitoring.
Looking ahead, the evolution of AI language models beyond the current capabilities of systems like ChatGPT and Claude promises to significantly impact social media brand monitoring. Innovations in neural network architectures, such as attention mechanisms and transformers, are paving the way for AI to understand nuanced language and context with greater accuracy. For instance, future AI systems will be able to discern sarcasm, detect subtle shifts in sentiment, and even predict emerging trends based on linguistic patterns in social media conversations.
This enhanced understanding will enable brands to proactively address potential crises and tailor their messaging to resonate with specific audiences, leading to more effective brand management strategies. The integration of these advanced AI language models will undoubtedly refine the precision and depth of sentiment analysis in social media brand monitoring. Furthermore, machine learning techniques initially developed for weather prediction are finding new applications in forecasting brand sentiment and identifying potential reputational risks. Time series analysis, a core component of weather forecasting, can be adapted to analyze the temporal patterns of social media conversations and predict future trends in brand perception.
By identifying leading indicators and anomalies in social media data, brands can anticipate potential crises and take proactive measures to mitigate their impact. For example, if an AI system detects a sudden increase in negative sentiment related to a specific product feature, the brand can quickly investigate the issue and implement corrective actions before it escalates into a full-blown crisis. This predictive capability, borrowed from the field of weather forecasting, represents a significant advancement in the proactive management of brand reputation.
Finally, the ongoing evolution of neural networks, extending beyond the large language models that dominate current AI applications, holds immense potential for enhancing AI image analysis and AI video analysis in social media brand monitoring. New architectures, such as graph neural networks, are particularly well-suited for analyzing the complex relationships between objects and individuals in images and videos. This capability enables AI systems to identify brand logos, detect product placements, and even recognize specific individuals associated with a brand. Moreover, advancements in unsupervised learning are enabling AI systems to automatically discover patterns and anomalies in visual data, without requiring extensive human labeling. This is particularly valuable for identifying emerging trends and potential risks that might be missed by traditional monitoring methods. As these neural network innovations continue to mature, they will empower brands to gain deeper insights into their visual presence on social media and proactively manage their brand image.