Creating our proprietary algorithm that identifies trends

In a constantly changing digital world, spotting a trend at the right moment can be the difference between being a leader or falling behind. That’s why, at Keytrends, we’ve created our own time series analysis algorithm to detect when a search is becoming a trend.

This algorithm not only allows us to predict trends but also categorize the type of alert it will trigger, whether it’s a smooth and progressive trend (soft alert) or a sudden spike (hard alert). Let’s break down how we do it and why it’s such a powerful tool for your marketing strategy.

Why create our own algorithm?

We could have relied on existing solutions, but every business has its own needs, and in our case, the available models didn’t meet ours. We wanted precision and reliability, avoiding false positives at all costs.

To achieve this, we focused on an analysis based on annual time series. Through this approach, we managed to eliminate the noise and seasonality that often confuse other models, keeping only the trends that have a real long-term impact.

AD 4nXdMxvcRGX Osbb GocI4KTLQiyTOC 1pCj32TWLx8HaSNNtySGwklltwLxbAIeeHdBofEs3LUxT SsT4O7ARi6tFvRVBCs5w5Pkwf808DIxSPXuHVLkKGzbcC7GpW1aEJL417QVxU 8jGPFXlAnkOed cP

How does our trend detection algorithm work?

Components of a time series


To understand how we detect trends, it’s important to know the three fundamental components of a time series:

  • Trend: Marks the long-term direction of the data, whether increasing, decreasing, or stable.
  • Seasonality: Patterns that repeat regularly, like search fluctuations during specific times of the year.
  • Noise: Random variations that don’t have a predictable pattern and can’t be explained by other factors.

The key to identifying a real trend is to remove noise and seasonality, focusing on the pure trend component of the data.

AD 4nXfvB9XbquZrWPbbHAxUWq8

The Hodrick-Prescott filter: cleaning the time series

To obtain a cleaner time series, we apply the Hodrick-Prescott filter, which helps smooth the data, removes irrelevant fluctuations, and leaves us with a clear trend. This lets us know if a keyword is growing in a meaningful way or if it’s just experiencing isolated peaks.

Detection of local maximums and minimums

Once the series is cleaned, the next step is to identify the local maximums and minimums. These points help us pinpoint the moments when a trend could be starting. If we detect a local minimum, it means that the search is beginning to grow. If, on the other hand, a local maximum appears, the trend is probably decreasing. This information tells us where the trend stands.

Trend threshold and type of alert

After identifying the minimums and maximums, we establish a growth threshold. If the series slope exceeds that threshold, the algorithm detects the trend and determines whether it’s a soft alert (progressive growth) or a hard alert (sudden spike). This is key to deciding if you should create evergreen content or react quickly to a short-term trend.

Why is this system useful?

This system not only allows you to anticipate what’s coming but also gives you a clear competitive advantage. When we detect a soft alert, you can capitalize on it by creating evergreen content at the right time. If it’s a hard alert, you can react quickly to capture short-term traffic. The best part is that we do this without false positives, ensuring that every identified trend has a real impact.

Keytrends in action: Riding the wave of trends

After three weeks of growth, we can detect whether a soft trend will turn into relevant evergreen content. In the case of hard alerts, we can identify within just two weeks if a sudden search spike deserves your attention. With Keytrends, you’ll never fall behind and can always react on time.

AD 4nXdEadGYirGhiRU2w70jwOD7k5dOq2ZhzCLNRo QuJ2CiX4JsAbMHTFzXFiuoaJzBomfi2K6 U2E3abSPr9Nk4UqYN2VE14lnb66PXjB IAPgrQ8bma7KijnLIyS7VeLbV6XIvY8fNCLYmogkO6Y9

Take advantage of trends and adjust your strategy

Trend analysis not only helps you anticipate what will be relevant but also allows you to adjust your strategy. By knowing exactly when a search is gaining traction, you can stay ahead of your competitors and create content that connects with your audience at the right moment.

Soft vs. Hard Alerts: How to use them?

  • Soft Alerts: Ideal for creating evergreen content. For example, if you detect that a keyword is starting to gain relevance, you can create solid content that remains useful and relevant for months or years.
  • Hard Alerts: Let you react quickly to sudden search spikes. An example might be the unexpected launch of a tech product, where the content should be published as soon as possible to capture traffic during the surge.

How to adapt your content to trends?

  • Evergreen Content: From soft alerts. Plan long-lasting content, such as guides, tutorials, or detailed articles that can continue generating traffic over time.
  • Quick Reaction: In the case of hard alerts, adjust your marketing campaigns to capture immediate interest. This could include social media posts, newsletters, or even exclusive promotions.

Case Study: How Keytrends detected an emerging trend

A client in the technology sector used Keytrends to detect a soft alert related to the search for “green technology.” Although the search volume was moderate, the trend showed progressive growth. Thanks to this alert, the client created evergreen content on the topic, positioning themselves as a leader in their niche before the trend exploded.

Conclusion: A tailored system to detect opportunities

At Keytrends, we’ve designed a unique solution that responds to the specific needs of trend analysis. Our time series algorithm eliminates the noise and focuses on what truly matters, giving you a competitive advantage in a market flooded with information. With our soft and hard alerts, you’ll always be one step ahead, capturing both the short-term searches and the evergreen trends that define the future of content.