How Sentiment Analysis Informs Brand Perception
Introduction
In today’s hyperconnected world, a brand’s identity is no longer defined solely by what it says but by what people feel and share about it. Public dialogue, social sentiment, and digital conversations have become powerful forces that shape User & Market Branding Perception.
For brands seeking premium positioning in Dubai or anywhere the challenge goes far beyond managing logos, slogans, or marketing campaigns. It’s about decoding the emotional undercurrent beneath public opinion: the feelings, trust, and expectations that truly define brand equity.
That’s where brand sentiment analysis comes in. Once considered a niche capability, it has evolved into a strategic necessity. It captures the emotional pulse of audiences, translating data from reviews, comments, and conversations into actionable insights about how a brand is genuinely perceived, not just how it intends to be seen.
For a branding agency in Dubai, where reputation, prestige, and stakeholder confidence are core to long-term success, sentiment analysis offers a game-changing advantage. It enables agencies to guide clients with precision revealing hidden drivers of loyalty, advocacy, and trust.
This article explores the mechanics, frameworks, and high-impact applications of brand sentiment analysis. Along the way, it highlights regional nuances that matter in the UAE market and concludes with a vision for how your agency can lead clients in mastering User & Market Branding Perception through emotionally intelligent strategy.
The Strategic Imperative of Brand Sentiment Analysis
Beyond Mentions: From Buzz to Belief
Many organizations still measure success through brand awareness or share of voice essentially, how often the brand is mentioned compared to competitors. But that metric only scratches the surface. The real insight comes from understanding how people talk about the brand, the emotional tone, whether positive, negative, or somewhere in between and why those emotions arise.
As one branding expert aptly said, “Your brand isn’t what you say it is, it’s what they say it is.”
In today’s social-first world, a single post or tweet can ignite conversations that spread across entire networks in hours. Sentiment, therefore, becomes more than a metric; it’s an early warning system. It reveals potential crises, reputational risks, or hidden opportunities long before they show up in quarterly reports.
According to Statista, 60% of marketers now use sentiment data from social platforms to continually monitor how their brands are perceived online. Meanwhile, 44% of CMOs view sentiment analytics as a core component of predictive modeling, a way to anticipate shifts in brand perception and consumer behavior before they happen.
And the payoff is clear: studies show that brands actively using sentiment monitoring report up to 15% higher customer retention, largely because they can respond faster to negative feedback and reinforce positive conversations.
Yet, despite this momentum, a 2024 Global Comms Report revealed a critical gap: only 22% of communications leaders rated their ability to track public sentiment as “excellent,” and just 25% said the same about their understanding of brand reputation drivers. This is precisely where agency leadership can make a tangible difference especially in reputation-sensitive markets like the UAE and GCC, where perception can directly influence partnerships, investment, and consumer trust.
Why It Matters in a Dubai & GCC Context
Dubai’s brand ecosystem is unlike any other. The city itself operates as a multi-dimensional brand competing globally as a business hub, tourism destination, and lifestyle icon. The success of Expo 2020, for instance, lifted Dubai’s brand strength by 1.4 points and added an estimated US$6.6 billion in brand value by 2023. That achievement wasn’t luck; it was the result of intentional strategy, audience understanding, and continuous sentiment calibration.
Across the GCC, sentiment insights have become especially valuable in sectors built on trust banking, real estate, and tourism. A UAE–Saudi banking sentiment survey found that negative themes such as unethical behavior and slow response times were key drivers behind customers switching banks. Interestingly, the only institution that sustained a net positive sentiment was Emirates NBD, thanks largely to how consumers perceived its digital innovation and customer-first approach.
Even at a micro level, sentiment analysis is reshaping how industries think about feedback. In one study of Dubai restaurant reviews, researchers categorized emotions across eight sentiment types and mapped them to restaurant formats fine dining, local eateries, and international chains. This revealed “hidden sentiment” layers that went far beyond star ratings, offering data-driven insights for experience design and service improvement.
For branding agencies in Dubai especially those advising clients across hospitality, fintech, real estate, and government sectors this is more than an analytical skill. Developing sentiment analysis capabilities means mastering modern reputation engineering. It’s how agencies translate emotion into strategy and transform User & Market Branding Perception into measurable brand advantage.
Anatomy of Brand Sentiment Analysis: From Data to Diagnosis
How Brand Sentiment Analysis Works in Practice
Brand Sentiment Analysis (BSA) isn’t just a data exercise, it’s a living system that blends technology, linguistics, and human judgment. To truly capture how people feel about a brand, you need more than algorithms; you need a framework that listens deeply and interprets meaning in context. Here’s how a robust sentiment analysis architecture typically comes together
1. Define Objective & Scope
It all starts with defining your objective and scope. Before diving into data, it’s essential to be crystal clear on what you’re measuring. Are you tracking overall brand health over time, gauging the response to a new campaign, or monitoring how you stack up against competitors? A focused objective acts like a compass; it shapes your data taxonomy, sets your analysis thresholds, and determines how often you measure and refine insights.
2. Map Sentiment Data Sources
Next comes mapping your data sources, because sentiment doesn’t live in one place. It’s scattered across the social and digital landscape Twitter, Instagram, LinkedIn, TikTok, Google Reviews, Trustpilot, industry forums, news outlets, even customer service logs and internal feedback loops. Each channel captures a different emotional tone and perspective. The more diverse your input, the more credible and representative your sentiment model becomes especially in multicultural markets like the GCC, where language and expression vary widely.
3. Preprocessing & Noise Filtering
Once data is collected, preprocessing and noise filtering turn chaos into clarity. Raw text can be messy full of HTML tags, emojis, and irrelevant chatter. So the first step is to clean it: normalize the text, remove stop words and punctuation, and ensure that every mention is actually about your brand. In regions like the UAE, this often means building custom disambiguation models to separate genuine brand mentions from unrelated uses particularly when brand names double as common words or appear in multiple languages.
4. Sentiment Classification & Scoring
Then comes the analytical heart of the system: sentiment classification and scoring. Every mention is tagged as positive, negative, or neutral but the sophistication of this step varies. Some teams use lexicon-based models (essentially word dictionaries with sentiment scores), while others rely on supervised machine learning methods like Naïve Bayes or SVMs. More advanced setups employ neural networks and transformer models such as BERT or GPT, fine-tuned for contextual understanding. The frontier today is aspect-based sentiment analysis, which detects not just the overall emotion but what it’s about for example, distinguishing praise for “customer service” from frustration about “pricing.”
Cutting-edge research pushes even further. Techniques like the Brand-Topic Model use factorization and adversarial learning to map sentiment continuously rather than in discrete bins, while the Semantic Brand Score (SBS) blends three elements prevalence (how often your brand is mentioned), diversity (the range of contexts it appears in), and connectivity (its network centrality in discussions). Together, they form a holistic metric for understanding brand importance in the real world. Many agencies in the region adopt hybrid systems combining a baseline classifier with localized, bilingual models tailored for Arabic-English sentiment nuances.
5. Aggregation, Trend Analysis & Segmentation
After classification, it’s time for aggregation, trend analysis, and segmentation. Here, individual scores evolve into insights. You calculate metrics such as net sentiment (the difference between positive and negative mentions), sentiment share (the proportion of conversations by tone), and time-based trends to track changes during campaigns or major events. Segmentation adds depth revealing differences by geography, language, or audience segment while driver analysis helps pinpoint the factors most influencing sentiment shifts, such as product quality, service speed, or brand design. Statistical tools like regression and time series modeling make these relationships visible and actionable.
6. Interpretation, Prioritization & Action
Still, data without interpretation is inert. The real value comes from connecting the numbers to narratives. Analysts must read sample comments, identify story arcs, and understand the cultural or emotional triggers behind shifts in tone. This is where human insight meets machine intelligence turning sentiment signals into strategic actions for brand messaging, customer experience, crisis management, or even product innovation.
7. Continuous Learning & Model Governance
Finally, every great system needs continuous learning and governance. Sentiment models are dynamic; language evolves constantly. New slang, memes, and cultural references emerge overnight. To stay relevant, models must be audited regularly ideally every quarter or after key campaigns. This involves refreshing lexicons, checking accuracy against human-labeled data, and pruning any drift that might distort results. Governance ensures reliability: flagging anomalies, setting escalation thresholds, and feeding learnings back into brand and communication teams.
In short, brand sentiment analysis is both art and science. It captures not just what people are saying, but how they feel and why that matters. For agencies in Dubai and across the GCC, mastering this craft is key to leading in User & Market Branding Perception, where emotion, trust, and cultural context define competitive advantage.
How Sentiment Informs Brand Perception: Use Cases & Strategic Levers
When done right, sentiment analysis doesn’t just measure emotion it guides decisions. It helps agencies move from reactive communication to proactive reputation design. Below are several high-impact ways it can inform User & Market Branding Perception and sharpen agency-level interventions.
1. Reputation Monitoring & Early Warning
At its core, sentiment analysis is an early-warning radar for brand health. A small uptick in negative emotion, spotted early, can save brands from full-scale crises.
Take a real example from the GCC: an airline noticed a sudden cluster of online complaints about flight delays and baggage issues. Thanks to automated sentiment alerts, the operations team intervened quickly, fixing process gaps before the issue spread to mainstream media.
Similarly, during Dubai’s Expo 2020, AI models tracking public discourse identified subtle sentiment shifts around sustainability, governance, and visitor experience. This allowed decision-makers to adjust narratives in real time proof that sentiment can act as a living pulse of public mood.
For agencies serving government, tourism, or real-estate clients, offering a sentiment-based reputation dashboard isn’t just an add-on, it’s a differentiator. It positions the agency as a partner in risk intelligence, not just brand storytelling.
2. Benchmarking & Competitive Positioning
Most agencies already benchmark awareness, engagement, or creative recall. Sentiment analysis adds a new and more human layer emotional preference.
By comparing net sentiment across competitors or even city brands (for instance, Dubai vs Abu Dhabi vs Doha), agencies can uncover perceptual gaps and emotional whitespace. A 2012 study of Brand Dubai found largely positive buzz compared with peers like Abu Dhabi and Qatar yet also revealed short-term dips tied to the global debt crisis.
That kind of insight can anchor a compelling pitch: “We aim to raise your brand’s net sentiment score by +15 points above your closest competitor within 12 months.” Tracking that sentiment delta over time gives tangible ROI to what is often seen as an abstract brand promise.
3. Campaign Validation & Real-Time Calibration
Sentiment analysis doesn’t just validate a campaign after it ends it can steer it while it’s live.
Imagine your agency launches a major brand film across GCC markets. Within days, you can see how audiences react not just in numbers, but in feelings: Do they connect with the narrative? The visuals? The message tone? Are conversations turning positive around diversity or innovation, or negative around perceived pretension?
If certain themes spark resistance, creative teams can pivot mid-flight tweaking messaging, visuals, or audience targeting. Campaigns that adapt this way are proven to perform better: sentiment-aware ads see roughly 2.3× higher engagement than tone-neutral ones.
4. Experience Design & Touchpoint Optimization
Sentiment data can also become a design tool. When emotion clusters around frustration say, with app usability or after-sales service that’s a cue to revisit the customer journey. Instead of relying solely on focus groups, brands can use real-time emotional feedback to find friction points and fix them fast.
This matters even more in multilingual markets. In the UAE, Arabic and English mentions may carry very different emotional tones. Disaggregating sentiment by language often reveals where localization and cultural nuance are missing, turning what could have been a PR challenge into an opportunity for empathy-based design.
5. Narrative & Content Strategy Refinement
Every brand builds its story sustainability, innovation, community, heritage and hopes it resonates. Sentiment analysis closes the loop by showing which parts of that story actually land.
For example, if a brand leans on the theme “heritage meets innovation,” real-world feedback might reveal that audiences celebrate “innovation” but view “heritage” as contrived. Those nuances help content teams fine-tune editorial pillars, tone, and storytelling frameworks. It’s the difference between telling a story and living one your audience believes in.
6. Crisis Simulation & Scenario Planning
Finally, leading agencies are embedding sentiment modeling into crisis preparedness. By simulating how sentiment could evolve in response to a hypothetical issue, a product defect, regulatory change, or social backlash agencies can pre-plan response thresholds.
This foresight allows communications teams to act with clarity, not panic. Instead of scrambling after sentiment turns negative, they already know when and how to engage, what tone to adopt, and which channels to activate.
The Takeaway
Sentiment analysis isn’t just about data dashboards, it’s about empathy at scale. It helps agencies in Dubai and across the GCC read the emotional undercurrents of markets that are fast-moving and culturally diverse.
When agencies combine data science with human insight, they don’t just monitor perception, they shape it. And that’s where the future of User & Market Branding Perception truly begins.
A Framework: The “SENTI-GATE” Model for Brand Sentiment Strategy
To move from raw sentiment data to meaningful strategy, agencies need more than tools they need a clear operating system. The SENTI-GATE model offers exactly that: a structured nine-phase roadmap that helps transform emotional signals into strategic intelligence. It’s a way of embedding empathy, data, and action into every layer of brand management.
1. Scope & Stakeholder Alignment
Every great analysis begins with clarity of purpose. In this first phase, the agency works closely with the client to align on overarching goals, key sentiment KPIs, and priority audiences. Together, they decide which dimensions of perception truly matter: trust, innovation, reliability, prestige and define the boundaries for data capture, analysis cadence, and escalation thresholds. This shared clarity ensures that what’s measured directly connects to what matters most.
2. Estimate Channels & Volume
Next comes mapping the ecosystem of conversation. This involves identifying every channel where brand sentiment lives from social media and review platforms to customer support transcripts, news coverage, and even internal surveys. The agency forecasts mention volume and secures API connections or data pipelines to guarantee steady, compliant data inflow. The goal here is inclusivity: to capture the full emotional spectrum of how audiences talk, feel, and share.
3. Normalize & Filter Noise
Raw data is rarely clean. This stage focuses on refining it so that what remains is accurate and meaningful. Natural Language Processing (NLP) techniques remove spam, irrelevant chatter, and false positives, while advanced filters distinguish between real brand mentions and unrelated uses a must in multilingual markets like the GCC, where brand names often overlap with common words. The outcome: a reliable foundation for all subsequent analysis.
4. Train & Calibrate Models
No model is one-size-fits-all. Here, agencies either fine-tune existing sentiment models or develop custom ones tailored to the client’s domain, tone, and market. Calibration might involve handling bilingual text, such as Arabic-English code-switching, or adapting sentiment scales for cultural nuances in expression. Human validation remains central reviewing model outputs to ensure they feel accurate and culturally aligned, not just statistically correct.
5. Integrate & Aggregate Metrics
Once the model produces sentiment outputs, those insights need a visual home. Dashboards consolidate metrics like net sentiment, trendlines, share of voice, and audience segmentation, often enhanced with real-time alerts. These visuals turn complex data into something decision-makers can interpret at a glance a living interface that keeps emotional awareness front and center.
6. Generate Insight & Drivers
This is where the story behind the numbers begins to emerge. By applying statistical techniques like regression or cluster analysis, agencies can pinpoint the drivers behind sentiment shifts, what’s fueling delight, and what’s sparking frustration. Maybe “customer service” consistently drives positive sentiment, while “pricing” triggers downturns. These insights reveal emotional cause-and-effect patterns that traditional analytics often miss.
7. Act & Intervene
Data becomes powerful only when it informs action. In this phase, agencies translate insights into targeted interventions: refining brand messages, adjusting tone of voice, optimizing service touchpoints, or testing creative alternatives. The aim is not just to react to sentiment but to shape it, closing the feedback loop between perception and performance.
8. Test & Monitor
With interventions in place, outcomes are measured and compared against the baseline. Did sentiment improve? Did the audience trust rebound? Post-mortem audits reveal what worked and where the model may have drifted. These learnings recalibrate the system, ensuring that each campaign makes the next iteration smarter.
9. Evolve the Program
Finally, sentiment intelligence becomes part of the client’s broader brand governance. Agencies help institutionalize the practice through quarterly reviews, training sessions, and integration with other KPIs from brand equity to customer experience. Over time, sentiment tracking expands across business units, products, and regions, creating a culture where emotional data informs every strategic decision.
From Insight to Intelligence
The power of SENTI-GATE lies in its cyclical nature. It’s not a one-off report but a living system continuously sensing, learning, and evolving. By embedding emotional intelligence into brand governance, agencies turn sentiment from a reactive metric into a proactive advantage.
In a region where reputation moves markets and perception defines value, the ability to listen deeply and act decisively is what sets great agencies apart. SENTI-GATE™ isn’t just a framework, it’s the bridge between data and empathy, helping brands in Dubai and across the GCC navigate the emotional economy with confidence.
Multilingual and Cultural Layers in GCC Sentiment
Multilingual Complexity and Dialect Nuance
In the GCC, language isn’t just a medium it’s a living mosaic. Conversations move fluidly between Modern Standard Arabic, local dialects like Emirati or Khaleeji Arabic, and English, often within the same sentence. This kind of code-switching makes sentiment detection far more intricate than in monolingual markets.
To interpret these emotions accurately, sentiment models must understand:
- Local idioms, sarcasm, and mixed Arabic-English expressions
- Non-standard transliterations (Arabic written in Latin script)
- Regional slang, emojis, and evolving meme expressions
Failing to recognize these nuances can lead to serious misreads classifying humor as hostility or irony as praise. Many global sentiment tools still struggle with Gulf Arabic’s subtleties, which is why localized linguistic calibration is indispensable.
Cultural Context, Symbolism & the “Surface vs. Deep” Tone
Words aren’t the only carriers of sentiment symbols, imagery, and metaphors often hold cultural weight that generic models simply miss.
A campaign evoking falconry or pearl diving may stir national pride among Gulf audiences, yet register as neutral or irrelevant in Western-trained models. That’s why human-in-the-loop review real people validating AI insights is vital. Analysts familiar with local culture can catch those layers of meaning that algorithms gloss over, ensuring sentiment isn’t just technically correct but culturally true.
Signal vs. Noise: Disambiguation Matters
In regional analysis, brand names frequently appear in broader contexts that aren’t directly about the client. For instance, “Dubai” surfaces in thousands of conversations about tourism, economy, or events not all of them linked to a specific brand.
Effective sentiment architecture includes disambiguation filters to separate client-specific chatter from general noise. Similarly, sentiment surrounding an industry (say, “real estate” or “banking”) can unintentionally skew a brand’s perception. Distinguishing macro sentiment trends from brand-specific sentiment is crucial for reliable insights.
Temporal Spikes & Event Bias
The Gulf region’s event-driven media environment creates frequent sentiment spikes from global summits and sports events to Expo milestones and policy shifts. Without contextual normalization, these fluctuations might be misread as crises.
Dashboards should therefore include event layers and anomaly detection, allowing analysts to attribute sentiment changes to specific external triggers rather than internal brand issues.
Model Drift & Lexicon Aging
Language evolves faster than any algorithm. New slang, memes, and ironic formats appear overnight. Left unchecked, even the best sentiment models degrade over time.
Agencies should build a continuous feedback loop retraining models quarterly or post-campaign, refreshing lexicons, and validating outputs against human annotations. This ensures that your sentiment system remains accurate, current, and culturally responsive.
Balancing Quantitative and Qualitative Insight
Sentiment scores and net percentages provide direction but numbers alone can’t tell the whole story. The richest insights come from reading representative posts, clustering conversations, and understanding the narrative arcs behind the metrics.
Qualitative sampling humanizes the data. It captures tone, humor, and context those intangible cues that define public emotion far better than a graph ever could.
Dubai Case Study: Sentiment in Action
Client: A boutique luxury hospitality brand launching a flagship hotel in Downtown Dubai
Objective: Optimize brand positioning during the pre-launch and opening phases
1. Baseline Sentiment Sweep
The agency begins by analyzing sentiment around Dubai’s luxury hotel market, tracking conversations across Instagram, TripAdvisor, Arabic-language forums, and travel blogs from the past year. This establishes a benchmark for tone, competitive themes, and emotional triggers.
2. Creative Testing
Two pre-launch campaign concepts are tested:
- Soft-luxury emphasizing heritage and craftsmanship
- Futurism highlighting skyline innovation and modern design
Early teaser posts reveal that English-language audiences respond more positively to the futurism theme, while Arabic audiences show warmer sentiment toward heritage-driven messages. The campaign is fine-tuned for bilingual emotional balance.
3. Sensitivity Monitoring
In the first week post-launch, a few guests posted complaints about check-in delays. Within 90 minutes, the sentiment dashboard flags a negative spike. The client’s operations team responds swiftly with transparent messaging and service recovery offers. The issue dissipates before reaching the press proof that speed and empathy can defuse potential crises.
4. Evolution Monitoring
Over six months, sentiment by attribute (service, food, location, design) is tracked. A consistent pocket of mild dissatisfaction emerges around WiFi speed. After the client upgrades connectivity, net sentiment in that category jumps from –8 to +3 within two months, a tangible ROI on insight-driven action.
5. Competitive Sentiment Benchmarking
The agency also tracks peer luxury hotels. By year’s end, the client’s net sentiment surpasses competitors by +7 points. This metric becomes a central talking point in investor pitches positioning the brand not just as beautiful, but beloved.
Positioning Your Agency as a Thought Leader in Sentiment-Driven Branding
To stand out in Dubai’s crowded branding landscape, don’t treat sentiment analysis as an add-on to make it your intellectual property. Here’s how:
1. Create a Proprietary Sentiment Playbook
Develop in-house frameworks (like your SENTI-GATE model), along with GCC-specific lexicons and dialect modules. These tools become part of your proprietary asset base, reinforcing your credibility as a regional expert.
2. Launch a GCC Sentiment Benchmark Index
Publish a recurring Brand Sentiment Index across key sectors hospitality, retail, fintech, real estate. Share it quarterly or biannually. Such benchmarking attracts media coverage and positions your agency as an authority on emotional brand performance.
3. Offer Sentiment Clinics or Audits
Provide clients with a concise sentiment audit and a snapshot of their current emotional standing, complete with recommendations. It’s a low-barrier entry service that often leads to deeper strategic engagements.
4. Run Workshops & Training
Educate your clients’ marketing, PR, and CX teams. Teach them how to interpret sentiment outputs, integrate them into decision gates, and sustain continuous feedback loops. Empowered clients value long-term partnerships.
5. Embed Insights into Retainers
For ongoing clients, integrate sentiment dashboards, alert systems, and quarterly reviews into retainer agreements. This shifts your role from creative vendor to strategic reputation partner.
6. Lead with Publishing & Thought Leadership
Regularly share white papers, webinars, or case-based articles on sentiment trends in the GCC. Cite anonymized client work or market examples to demonstrate expertise. Over time, your agency becomes synonymous with sentiment-led brand strategy in the region.
When clients see that your agency doesn’t just execute campaigns but interprets emotion and engineers perception, you elevate from design partner to reputation architect.
Conclusion
In a market overflowing with campaigns and claims, the real differentiator isn’t what brands say it’s how deeply they resonate. Sentiment analysis is the scalpel that reveals that resonance.
For Dubai-based branding agencies, mastering this discipline means evolving from storytellers to perception engineers. Using frameworks like SENTI-GATE™, you can transform scattered sentiment data into strategic intelligence closing the loop between what people feel and how brands respond.
As competition intensifies across sectors, the winners won’t be the loudest, they’ll be the ones who listen best. Agencies that can read emotion as clearly as they read metrics will define the next era of User & Market Branding Perception in the GCC.
In this new paradigm, emotion becomes data, and data becomes empathy, a continuous dialogue between brand and audience. Sentiment intelligence doesn’t just track perception; it shapes trust, refines storytelling, and future-proofs reputation. By mastering the language of emotion, your agency ensures brands don’t merely occupy market share they own mindshare.
Ultimately, the future of brand leadership in Dubai belongs to those who see insight as feeling quantified brands that measure not only visibility, but belief.
FAQ
1. What is sentiment analysis in branding?
Sentiment analysis is the process of using AI and data tools to evaluate customer opinions, emotions, and attitudes toward a brand. By analyzing reviews, social media comments, and feedback, businesses can understand how people truly feel about their products and services.
2. How does sentiment analysis help improve brand perception?
Sentiment analysis helps brands identify strengths, weaknesses, and customer concerns in real time. This insight allows businesses to address issues quickly, refine messaging, and improve customer experience—ultimately shaping a more positive brand image.
3. What types of data are used for sentiment analysis?
Sentiment analysis uses data from multiple sources such as online reviews, social media posts, surveys, emails, chat interactions, and customer support conversations. These inputs provide a comprehensive view of public opinion about a brand.
4. How can businesses use sentiment analysis for better decision-making?
By tracking customer sentiment trends, businesses can adjust marketing strategies, improve products, enhance service quality, and respond proactively to negative feedback. It turns customer opinions into actionable business insights.
5. What tools can companies use for sentiment analysis?
There are many tools available, including social listening platforms, AI-powered analytics software, and CRM systems that track customer feedback. These tools help brands monitor conversations and measure perception accurately over time.
