Latest news with #forecasting
Fast Company
12 hours ago
- Business
- Fast Company
Stop flying blind: Build a data-driven revenue model with bottom-up forecasting
One of the biggest challenges early-stage startup founders face is predicting and managing revenue growth. In most organizations, this looks like top-down forecasting and starts with determining the target revenue goal for that period. This goal is often based on ambition, investor expectations, or competitive pressure. But this method is flawed; you can't set revenue goals based on a finger held to the wind. What should be an exercise rooted in science lacks real-world input, and thus prevents leaders from obtaining the performance-based insights necessary to make data-driven decisions. The solution? A reality-based assumption model, known as a 'bottom-up' go-to-market (GTM) forecast. Here's what to know: WHAT IS A BOTTOM-UP FORECAST? In short, it's a more reliable way to determine how various functions and tactics are expected to contribute to revenue. This methodology allows businesses to draw a straight line to their revenue goals based on assumptions or known variables. This is very similar to a 'revenue bridge,' a tool designed to outline how portfolio companies intend to achieve their ultimate revenue goals. HOW A BOTTOM-UP REVENUE MODEL WORKS This approach to forecasting first involves segmenting the business by its core functions, which include sales, marketing, and customer success. Then, each business unit should outline the activities it performs that contribute to revenue acquisition or expansion. This data can be pulled from historical campaigns and updated for current market conditions. However, if this historical data isn't available, a little creativity is necessary. There are a plethora of benchmark data sources available (full disclosure: We provide custom benchmarks based on industry verticals and business models to our portfolio companies). Ultimately, with a bit of legwork, organizations can develop a set of assumptions to manage, with the goal of getting more and more confident over time. Of course, budget is a major factor in this exercise, which is a common constraint. However, there is no more effective way to justify budgets than with supporting bottom-up assumptions. This differs greatly from the top-down approach, which works backward from a target and effectively tells everyone to 'run' at it. That's flying blind. WHY IS BOTTOM-UP FORECASTING IMPORTANT? Top-down revenue models tend to be more pie-in-the-sky than rooted in reality. When a company forecasts this way, it's more likely to overestimate its growth and underestimate the costs and challenges of scaling. It can also miss critical allowances for changes in the process. Markets and audiences don't remain static, organizations need to account for change that comes with penetration, competition, and market headwinds or tailwinds. On the flip side, bottom-up forces business leaders to take a good, hard look at their teams' actual performance or assumptions of performance. Even if the topline isn't sexy initially, leaders will know where they're heading if their employees continue to perform at past levels. This can potentially define what the chasm of 'go-get' revenue looks like versus the expectation. Optimistic but grounded in data is far better than optimistic and hopeful. Additionally, a bottom-up approach is not only a financial plan—it's also an operational plan. This method covers the goals, costs, and resources needed to achieve the company plan. As a result, this framework becomes a living, breathing tool to be updated and discussed on a regular basis. THE MAP TO PERFORMANCE-BASED BUDGETING A bottom-up pro forma isn't just essential for forecasting; it's also the foundation for budgeting decisions. Before you know what funds you have to allocate, you need to know where growth is expected and how each department and resource contributes to it. For example, how many sales reps do you need to hire to meet your revenue targets? If you don't have an understanding of what those resources should yield in terms of performance, your budget—and their compensation—will be a total shot in the dark. But once you have a pro forma, you can use it to inform budgets and have confidence in your decision-making as results begin to filter in. Based on the early-stage companies we evaluate and invest in, underperforming sales reps stay in a role three to six months longer on average than they should. Human nature wants to see our investments produce, but consider the opportunity cost of making key decisions too late. Time is the one variable that remains constant. ADJUSTING YOUR MODEL AFTER A FUNDING ROUND After a funding round, the pressure to scale quickly is intense. Without a reliable GTM model, businesses risk over-promising, burning out, and missing key opportunities. That's why it's critical to revisit and update your bottom-up pro forma based on current realities. Ask yourself: How many hires do we need, and how fast? Does our 'quota-in-field' align with our revenue goals? Which marketing channels are most effective, and how will increased spend impact leads? What revenue can we expect from existing customers versus new sales? This discipline helps to build internal clarity and inspire confidence with investors. KPIS AND BENCHMARKS KPIs and benchmarks are foundational to a bottom-up pro forma. While they vary by company, industry, and goals, key benchmarks include sales rep efficiency, customer acquisition cost (CAC), and revenue per employee. By consistently tracking these metrics, you can fine-tune your go-to-market model and drive better performance. Just as important are cultural KPIs. These metrics define how your team thinks and operates. The best CEOs use a select few KPIs as both their north star and operational compass, aligning strategy and execution across the business. What numbers should become the shared language of your company? THE BOTTOM LINE IN BOTTOM-UP Early-stage startups looking to scale should build reliable revenue acquisition through both top-down and bottom-up forecasting. This will help create a science of growth for your organization. By understanding your growth levers and required resources, you can avoid the trap of overpromising and underdelivering. The goal of a well-constructed forecast model isn't perfect accuracy, but to build confidence in a framework that evolves with real results. That framework becomes more accurate and predictive over time. Before you know it, you have that coveted map, and while others are flying into mountains, you're cruising safely toward performance.
Khaleej Times
3 days ago
- Business
- Khaleej Times
End of free trade era? Tariff war could stall growth for 3 years, CFO warns
Free trade, as the world has known it, is over at least for the next three years, according to a global shipping company head. Ali Abouda, Group Chief Financial Officer (CFO) at Gulf Navigation Holding PJSC, also observed that the world is now in an "active trade war" and finance professionals must resort to frequent forecasting and scenario planning to navigate the uncertain times. "Budgeting has probably become a redundant exercise," said Abouda. "We are going through a very, very unpredictable time in terms of business environment. If there's no stability where you operate, then everything becomes quite difficult and challenging. Budgeting should be replaced by scenario forecast and I think a very frequent forecast should be the name of the game." Ali was in conversation with Khaleej Times Chief Content Officer Ted Kemp about the impact of US tariffs on finance professionals at the sixth edition of the New Age Finance and Accounting (NAFA) summit organized by Khaleej Times. The event saw CFOs, finance leaders, policymakers, and fintech innovators engage in dialogue on various topics including taxes, ESG, and reskilling the workforce. How does Section 301 impact trade? One of the topics of discussion during the conversation was Section 301. Kemp questioned how the section of the Trade Act of 1974, which targets Chinese ships, would impact trade. Abouda explained that 90 per cent of global trade was accounted for by shipping - a field dominated by China. "The US took a decision to keep the brain in the US and move the muscles elsewhere. Section 301 targets Chinese-built, owned, or operated vessels calling at US ports, imposing penalties of $500,000 to $1 million per call in addition to the tariffs. "The resolution is expected to be effective October 4, and if this goes through, I think there will be a structural change in the whole supply chain. China, in one way or another, control about 60 percent of the shipping, either through building or through financing or operating or resources," he said. Unclear directives from US Abouda added that while the Trump administration was agile in decision-making, it lacked clarity. "If the US wants to build more ships, that's all good, except that it isn't something that's going to happen quickly," he said. "I don't know what's the thought process behind it." He gave the example of how US President Donald Trump encouraged more American companies to drill oil, with his 'drill, baby, drill' slogan. "That's practically not possible, because the US doesn't have the capacity today to start increasing production by 3 or 4 million barrels immediately," he said. "That needs a lot of regulations to be in place and a lot of investments, which the US is not ready for. If I was one of the leaders of those companies, I would think twice. If you go with the mandates of an administration that has three years left, maybe the next administration will have a totally different view," Abouda added. He said that Trump's constant shift in tariff policies was also damaging to the industry. "The most recent example is, he imposed a 50 percent tariff on Europe and two days later, it was pushed to July 9," he said. "So, he's creating an environment which is very difficult to do business in."
Bloomberg
3 days ago
- Business
- Bloomberg
How AI Could Change the Future of Weather Forecasts
Technology Explainer Weather forecasting helps industries avert billions in losses from extreme events, and AI could make predictions faster, cheaper and more accurate. Weather forecasting has gone through incremental but tremendous progress in past decades. By one metric, today's five-day forecast is now as accurate as a three-day forecast was in 2000. Entire ecosystems rely on weather forecasting, and any improvements — particularly as climate change heightens weather volatility — can help not just individuals to better manage risks, but also entire industries to avert billions in economic losses. In the US alone, an estimated one-third of the economy, or about $3 trillion, is sensitive to the weather and climate.
Geeky Gadgets
4 days ago
- Business
- Geeky Gadgets
Forecast Anything with Transformers with Chronos or PatchTST
What if you could predict the future—not just in abstract terms, but with actionable precision? From forecasting energy demand to anticipating retail trends, the ability to make accurate predictions has become a cornerstone of modern decision-making. Enter transformer-based models, a new advancement originally designed for natural language processing but now transforming time-series forecasting. Among these, Chronos and PatchTST have emerged as standout tools, offering unparalleled accuracy and adaptability for even the most complex datasets. Whether you're grappling with noisy healthcare data or modeling long-term climate trends, these models promise to redefine what's possible in predictive analytics. In this exploration, Trelis Research explains how transformers like Chronos and PatchTST are reshaping the forecasting landscape. We'll delve into their unique architectures, such as self-attention mechanisms and data segmentation into 'patches,' that allow them to capture intricate patterns and long-range dependencies with ease. Along the way, you'll discover their real-world applications across industries like finance, energy, and healthcare, and learn why their scalability and precision make them indispensable tools for tackling today's forecasting challenges. By the end, you might just see forecasting not as a daunting task, but as an opportunity to unlock new possibilities. Transformer Models for Forecasting What Makes Transformer-Based Models Ideal for Forecasting? Originally developed for natural language processing, transformers have demonstrated remarkable versatility in time-series forecasting. Unlike traditional statistical methods or recurrent neural networks, transformers process entire sequences simultaneously, allowing them to capture long-range dependencies in data. This unique capability allows them to handle complex datasets with greater speed and accuracy. From financial metrics to environmental data, transformers excel at identifying patterns and trends, making them a preferred choice for modern forecasting tasks. Their adaptability is another key strength. Transformers can be fine-tuned to suit various datasets and forecasting objectives, making sure optimal performance across industries. This flexibility, combined with their ability to process high-dimensional data efficiently, positions transformers as a fantastic force in predictive analytics. Chronos: A Flexible and Scalable Forecasting Model Chronos is a transformer-based model specifically designed to simplify forecasting across multiple domains. Its architecture uses self-attention mechanisms to detect intricate patterns and trends in time-series data. This makes Chronos particularly effective in scenarios where understanding complex temporal relationships is critical, such as stock market analysis, supply chain optimization, or energy demand forecasting. One of Chronos's standout features is its scalability. By incorporating advanced feature engineering and efficient training processes, Chronos maintains high performance even when working with large and complex datasets. This scalability ensures that the model remains reliable and accurate, regardless of the size or complexity of the forecasting task. Its ability to adapt to various industries and applications makes it a versatile tool for organizations aiming to enhance their predictive capabilities. Time-Series Forecasting with Chronos and PatchTST: A Complete Guide Watch this video on YouTube. Below are more guides on transformers from our extensive range of articles. PatchTST: A Targeted Approach to Time-Series Data PatchTST adopts a specialized approach to time-series forecasting by dividing data into smaller segments, or 'patches.' This segmentation enables the model to focus on localized patterns within the data before synthesizing broader insights. This method is particularly advantageous when dealing with irregular or noisy datasets, such as those encountered in healthcare or environmental monitoring. The modular design of PatchTST allows for extensive customization, allowing users to tailor the model to specific forecasting tasks. For example, in healthcare, PatchTST can be fine-tuned to monitor patient data and predict health outcomes, even when the data is highly variable. This targeted approach ensures that the model delivers precise and actionable insights, making it a valuable tool for industries that rely on accurate and timely predictions. Real-World Applications of Transformer-Based Forecasting The adaptability and precision of Chronos and PatchTST make them highly valuable across a variety of industries. Key applications include: Energy Management: Predicting electricity demand to optimize grid operations, reduce costs, and improve sustainability. Predicting electricity demand to optimize grid operations, reduce costs, and improve sustainability. Retail: Forecasting sales trends to enhance inventory planning, minimize waste, and improve customer satisfaction. Forecasting sales trends to enhance inventory planning, minimize waste, and improve customer satisfaction. Finance: Analyzing market trends to guide investment strategies, manage risks, and identify opportunities. Analyzing market trends to guide investment strategies, manage risks, and identify opportunities. Healthcare: Monitoring patient data to predict health outcomes, streamline care delivery, and improve resource allocation. Monitoring patient data to predict health outcomes, streamline care delivery, and improve resource allocation. Climate Science: Modeling weather patterns to enhance disaster preparedness, optimize resource management, and support environmental research. These applications highlight the versatility of transformer-based models, demonstrating their ability to address diverse forecasting challenges with precision and efficiency. Why Choose Transformer-Based Models? Transformer-based models offer several distinct advantages over traditional forecasting methods, including: Scalability: Capable of processing large datasets with high dimensionality, making them suitable for complex forecasting tasks. Capable of processing large datasets with high dimensionality, making them suitable for complex forecasting tasks. Accuracy: Superior performance due to their ability to capture long-term dependencies and intricate patterns in data. Superior performance due to their ability to capture long-term dependencies and intricate patterns in data. Flexibility: Adaptable to a wide range of industries and forecasting objectives, making sure relevance across diverse applications. Adaptable to a wide range of industries and forecasting objectives, making sure relevance across diverse applications. Efficiency: Faster training and inference times compared to recurrent models, allowing quicker deployment and results. These advantages make transformers an ideal choice for organizations seeking to enhance their forecasting capabilities and make data-driven decisions with confidence. Industry Adoption and Future Potential Industries worldwide are increasingly adopting transformer-based models like Chronos and PatchTST to address complex forecasting challenges. Examples of their application include: Utility Companies: Using these models to predict energy consumption patterns, optimize grid efficiency, and reduce operational costs. Using these models to predict energy consumption patterns, optimize grid efficiency, and reduce operational costs. Retailers: Using forecasting tools to streamline supply chains, reduce inventory costs, and improve customer satisfaction. Using forecasting tools to streamline supply chains, reduce inventory costs, and improve customer satisfaction. Healthcare Providers: Enhancing patient monitoring and predictive analytics to improve care delivery and resource management. Enhancing patient monitoring and predictive analytics to improve care delivery and resource management. Financial Institutions: Employing these models for market analysis, risk management, and investment strategy development. As transformer-based technologies continue to evolve, their applications are expected to expand further, driving innovation and improving decision-making across sectors. By addressing increasingly complex forecasting needs, these models are poised to play a pivotal role in shaping the future of predictive analytics. Transforming Forecasting with Chronos and PatchTST Chronos and PatchTST exemplify the potential of transformer-based forecasting models to transform predictive analytics. By combining advanced architectures with practical applications, these models empower organizations to forecast with precision, efficiency, and confidence. Whether you're managing resources, optimizing operations, or planning for the future, transformer-based solutions provide a reliable foundation for informed decision-making. Their ability to adapt to diverse industries and challenges ensures that they remain at the forefront of forecasting innovation, allowing you to navigate complex prediction tasks with ease. Media Credit: Trelis Research Filed Under: AI, Guides Latest Geeky Gadgets Deals Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, Geeky Gadgets may earn an affiliate commission. Learn about our Disclosure Policy.
TechCrunch
23-05-2025
- Science
- TechCrunch
Microsoft says its Aurora AI can accurately predict air quality, typhoons, and more
One of Microsoft's latest AI models can accurately predict air quality, hurricanes, typhoons, and other weather-related phenomena, the company claims. In a paper published in the journal Nature and an accompanying blog post this week, Microsoft detailed Aurora, which the tech giant says can forecast atmospheric events with greater precision and speed than traditional meteorological approaches. Aurora, which has been trained on more than a million hours of data from satellites, radar and weather stations, simulations, and forecasts, can be fine-tuned with additional data to make predictions for particular weather events. AI weather models are nothing new. Google DeepMind has released a handful over the past several years, including WeatherNext, which the lab claims beats some of the world's best forecasting systems. Microsoft is positioning Aurora as one of the field's top performers — and a potential boon for labs studying weather science. In experiments, Aurora predicted Typhoon Doksuri's landfall in the Philippines four days in advance of the actual event, beating some expert predictions, Microsoft says. The model also bested the National Hurricane Center in forecasting five-day tropical cyclone tracks for the 2022-2023 season, and successfully predicted the 2022 Iraq sandstorm. Image Credits:Microsoft While Aurora required substantial computing infrastructure to train, Microsoft says the model is highly efficient to run. It generates forecasts in seconds compared to the hours traditional systems take using supercomputer hardware. Microsoft, which has made the source code and model weights publicly available, says that it's incorporating Aurora's AI modeling into its MSN Weather app via a specialized version of the model that produces hourly forecasts, including for clouds.
