As a real-time trading company, eToro must provide users with reliable market rates as quickly as possible, necessitating close monitoring of the quality of the connection from both the client and server side. eToro had been using open-source tools to monitor the metrics from their Price Streams service that sends price quotes to their users. However, the company quickly realized that it needed to expand the number of metrics being monitored and faced resource challenges adapting their traditional monitoring tools to meet the new demands. With stringent regulations in Cyprus and the UK, eToro treats any trading error or problem as critical.

The company, an affiliate network with over 200,000 members, was struggling to monitor business and technical incidents that were impacting their bottom line. The dynamic nature of their marketplace and the extensive metrics they had to track made it difficult to monitor changes in real-time. Factors such as changes in search engine algorithms and third-party trends, as well as changes in affiliate accounts, could significantly impact their business. The tools they were using required them to set thresholds manually, which allowed time for incidents to escalate.

The digital printing company sells commercial digital presses for digital printing of items such as pictures, labels, large format prints, etc. Its customers are print houses around the world. The business model is that the customers purchase the printing press up front and also pay per page printed. The company provides the ink and spare parts and most of the support. Whenever a press is down, the company loses both in revenue (due to fewer prints) and in support costs. Customer satisfaction also suffers as a result. Typically, whenever there was a problem, the customers’ first instinct would be to start replacing spare parts that the company provides, and only afterwards they might call support. The initial support call costs the company several hundred dollars, and still may not resolve the problem. If an issue persisted or recurred, an expert would be sent (at the cost to the company of a few thousands of dollars per call). In this process, the company lost revenue from presses that were malfunctioning, paid a lot in support and parts, and also eroded its customer satisfaction.

Redis Labs, a company in a high-growth phase, was acquiring many enterprise customers in the Fortune 500 and Global 1000. It needed to scale its customer service while maximizing efficiency and minimizing time and resources. As Redis Labs scaled, it became responsible for managing tens of thousands of databases and could no longer manually monitor their usage patterns individually. The company wanted their monitoring to operate on a more granular level, picking up incidents that might otherwise go unnoticed. With the growing volume of databases also came a wider variety of usage patterns, which couldn’t be properly tracked with the fixed alerting that had proved sufficient in the company’s early days.

Payoneer, a global cross-border payments platform, was facing several challenges. They needed to replace their traditional monitoring system with one that could scale with their rapidly growing business. They also wanted to provide mission-critical monitoring-as-a-service to internal groups. Another challenge was to eliminate wasted engineering effort by preventing false positives on operational metrics. Lastly, they wanted to prevent revenue loss by accurately forecasting demand for funds. Given that the company’s operations span so many countries and involve a massive number of partners and their APIs, Payoneer continuously monitors nearly 200,000 metrics to ensure it meets SLAs and general reliability and performance targets.

When David Ashirov joined Freshly, a prepared meal delivery service, the company lacked systems to measure and evaluate data. The business was largely reliant on human intuition to gauge its performance. This approach was sufficient for a startup, but as the company grew, it became clear that human intuition could not scale. Ashirov's primary challenge was to build a data fabric, a system that would connect data across the company, allowing for easy querying of every bit of data without unnecessary complications. The goal was to create a single source of information for any business question, fostering trust in the data among the company's employees.

Tinkoff, Russia's leading fintech innovator, was facing challenges in managing the exponential surge in data due to the rise of innovative fintech and digital banking solutions. The success of their fintech model was heavily dependent on the quality of the customer experience they provided. However, monitoring, managing, and reconciling the vast amount of data was compromising their internal productivity and resources. They needed a technology that would not only guarantee the highest level of customer satisfaction but also ensure operational efficiency across their platform.

Uprise, an ad-tech company, uses a 'continuous delivery' approach for its software development, pushing around 20 new software releases into production each day. Each new release can affect the platform’s performance, making it crucial to monitor results in a timely fashion to determine if the new release should be kept in production or rolled back. The ad tech environment itself has many moving parts, each of which is a potential point of failure. These can include server issues, changes at the ad affiliates, introduction of ad blocking software, or even fraud. Whenever a problem occurs, isolating the source can require complex, time-consuming analysis. Identifying issues in the first place is also tricky, since network traffic behaves seasonally. With the traffic naturally reaching various peaks and valleys throughout the day, noticing a 20% loss or gain at any given point is next to impossible.

GetYourGuide, a global booking platform for travelers, was facing challenges in spotting issues in their business data in real-time. They were taking too long to identify problems, which led to revenue losses in their cloud services and marketing budgets, and negatively impacted user experience. The company needed an automated solution to control cloud costs, track product usage for changing revenue, and monitor marketing activity and ad spend. The global pandemic further complicated matters, as the travel industry was severely impacted, and GetYourGuide had to adjust its operations and spending accordingly.

With the rollout of 3G and 4G technologies, telecommunication service offerings have grown. Cell phone usage has skyrocketed. Voice, video, and data have converged to offer rich new services that customers rely upon. High-definition video consumption and other services are consuming more network bandwidth, making it more important than ever to accurately manage and maintain network performance. In this case study, a leading provider of telecommunications services needed to ensure end-customer satisfaction and quickly mitigate any network performance issues, where any incident could easily cost them millions of dollars. They had to be able to monitor service assurance, as well as analyze data at detailed levels to track the underlying quality of network performance and to avoid unexpected outages. Managing multiple communication applications and platforms required advanced network monitoring and orchestration to ensure optimal network performance. The company didn’t have clear visibility into how network resources were used. Their available network management tools were static and only addressed specific needs, unable to provide reliable, transparent data for insights in real-time.

Companies are generating more data than ever before, and traditional dashboards are unable to keep up with the volume and complexity of the vital business data collected. This is particularly true for companies using a Snowflake warehouse. The businesses served by Anodot have millions of customers across the globe and must manage millions of daily business metrics involving product usage, application performance, APIs, log-ins, and payment gateways, among others. Traditional manual business monitoring solutions cause significant delays of at least 24 hours or longer in detecting and resolving critical incidents, which threaten to impact customer satisfaction, brand equity, and the company’s bottom line. Transactional and customer experience data is too volatile for static monitoring. Since business data is complex and dynamic, AI/ML-based autonomous solutions are critical for achieving business outcomes and avoiding blind spots. Static monitoring approaches based on dashboards, and manual thresholds aren’t sensitive, robust, or agile enough to withstand this challenge. AI-based early detection of revenue issues and business system failures is nonnegotiable.

King, a leading mobile gaming company, was facing challenges in managing business incidents in real time. The company's most popular franchise, Candy Crush, along with 17 other games, were in production. The incident management team, responsible for investigating incidents and assessing the losses incurred, needed a tool to detect and address these incidents promptly. The goal was to minimize the impact on revenue by spotting incidents as soon as possible. The team was also tasked with monitoring 18 Key Performance Indicators (KPIs) for each game, which amounted to a significant number of metrics. The challenge was not only to monitor these metrics but also to differentiate them based on the platform, build, and country. The existing tools were not sufficient for this task, as they were not adept at detecting subtle anomalies in business KPIs.

PUMA, a global eCommerce giant, was facing difficulties in monitoring all revenue aspects of their 45 eCommerce websites. They lacked a tool to distinguish what was normal or abnormal across their platforms. For instance, an issue with gift card purchases in Switzerland went unnoticed, which could have resulted in significant financial loss if discovered later. PUMA's Senior DevOps Manager, Michael Gaskin, was interested in Anodot based on the experience he had with another Anodot customer. He understood the challenges PUMA was facing and sought a solution to monitor their websites more effectively.

Vimeo, a leading professional video platform, was facing the challenge of identifying critical signals in their decade's worth of data that could be used to improve operations, monetize services, and advance the business. The company's existing rule-based monitoring system was not able to understand each KPI's context or dig deeper into its permutations to find hard-to-detect anomalies. The company's growth was a big factor in the decision to adopt Anodot to quickly identify anomalies and trends in the data. The existing monitoring and alerting tools were based on hard-coded thresholds that couldn't cope with Vimeo's hyper growth. The threshold approach wasn't scalable because the numbers changed all the time.

Browsi, a startup providing large-scale publishers with AI tools to gauge the visibility and impact of their online ad inventory, was struggling with handling the enormous amounts of data it collects daily. The company needed an autonomous monitoring system to assess Key Performance Indicators (KPIs) such as page views, ad impressions, and other metrics for online movement and ad interaction. Browsi required real-time capabilities to respond to business or technical issues as they arise and to alert their customers of potential problems. Before integrating Anodot, Browsi had a limited view of its data systems and was alerted of technical or business problems only a day, and in some cases only several days, after they occurred. These delays and faults translated into a loss of revenue that could easily reach thousands of dollars.

Before using Anodot, Wix's data analysts manually measured and analyzed vast amounts of data. This included activity related to customers' actions on Wix, such as success and failure rates while opening the Wix website editors, checkouts at e-commerce sites hosted by Wix, logins by premium customers, and other important events. The analysts spent a great deal of time scrutinizing reports and graphs to try to detect issues, but important issues were sometimes identified hours to days after they had occurred. Wix needed a real-time alert system that would indicate issues without manual threshold settings in the key metrics.

As Minute Media’s business scaled, it became increasingly difficult to keep tabs on incidents that impacted user experience, revenue, and costs. The company needed a solution that could help identify underlying issues in the platform to prevent penalties with Google and other supply-side platforms, understand issues with the integrity of data aggregated from a wide variety of sources, improve the ad profit margins, especially during consistently changing patterns such as the pandemic, and prevent revenue loss by quickly notifying of fraudulent bot clicks on video ads. Prior to Minute Media adopting Anodot, data analysts extracted data and worked with it manually to try to spot anomalies or trends. This manual process was untenable, especially as the company grew, leading Minute Media to begin looking for an automated solution to identify anomalies in the business data.

Penske Media Corporation (PMC) was facing significant delays in discovering important incidents in their active, online business. The company was using Google Analytics’ alerting function to track business incidents but found it inadequate due to the millions of users across dozens of household-name and professional publications. The initial use case for PMC was to start using Anodot to track its Google Analytics activity, for example, to identify anomalous behavior in impressions or click-through rates for advertising units.

Telecommunications companies are facing the challenge of managing and monitoring an increasing number of products, campaigns, retail channels, prepaid and roaming services, billing, customer experience and support, and order and fraud management operations. The complexity and dynamic nature of business data make it difficult for static monitoring approaches to effectively track these metrics. Despite redundancies in data centers and telecom networks, outages and incidents still occur, impacting network, business, and customer experience management operations. The cost of these incidents can be significant, with a 2016 survey indicating that the average cost of a data center outage rose 7% from 2013 to 2016. For a telco operator with annual revenues of $1B, annual incident costs can range between $11.6M-$41.1M, depending on the types of systems used for monitoring.

The company, a mobile gaming giant, relies heavily on its in-house developed cross-promotional system for revenue. Any bug or change in the system could lead to more than 15% loss in in-app purchases. The company used to monitor impressions, clicks, and conversions of their cross promotions on a weekly basis using Tableau dashboards. However, this manual process was slow and inefficient, often leading to delayed insights on glitches. For instance, a new promotion caused major crashes across several platforms, but the issue was not logged until the next day. It took almost four days for the company to realize the problem, and it was only discovered when checking another unrelated system.

Xandr’s marketplace operates at a scale and complexity that are hard to fathom. The company serves multiple billions of ads every single day, handles 45 million transactions per second, and processes more than 175 terabytes of data. Xandr’s platforms make a lot of complex business decisions to reach the right customers for the marketers. When glitches occur and blank ads are served, all parties lose money. This has to be detected and resolved quickly before losses mount. The extensive nature of Xandr’s partnerships meant that issues could take a week or more to detect and resolve. Xandr’s infrastructure includes thousands of servers and hundreds of applications across its global data centers. The company used a variety of disparate tools to monitor the performance of the infrastructure itself as well as the delivery of the ads.

Rubicon Project, one of the largest ad exchanges in the world, processes trillions of transactions each month in real-time auctions. The company receives more than 13 trillion bid requests per month, handled in its seven global data centers, housing more than 55,000 CPUs. However, the Tech Ops team could not monitor more complex aspects of business and trends, especially not in real time. For instance, Rubicon needed real-time insight if a large institutional buyer deviated from its normal transaction trend by any percentage in one of the global data centers at any hour of the day or night. Such deviations could have a devastating effect on the exchange if there was a delay in addressing it with the customer. Along the bid stream, there were many potential areas for communication or technical breakdown, which would prevent the bid from going into the auction, and negatively affect overall bid health.

Magnite, the world’s largest independent sell-side advertising platform, processes trillions of transactions each month in real-time auctions that each occur within 40 milliseconds. The company's internal teams and existing tools could not scale to handle the growing volume and velocity of data. They needed real-time insight into incidents that were being detected too late, such as anomalies in normal transaction volume from a large buyer. Their manual alerting system with static thresholds also created costly alert noise and false positives. Magnite works with many demand-side platforms (DSPs) across its global data centers in different time zones. Along the bid stream, there are many potential areas for communication or technical breakdown, which would prevent the bid from going into the auction, and negatively affect overall bid health.

NetSeer, a leading adtech company, was facing challenges with its business and operational KPI tracking. The company was using several tools such as Graphite and alerting systems, but they were not accurately alerted on key business problems. The standard static thresholds were causing either too many false positives, or not enough alerts. For instance, the company tracks the number of ad calls to their front end and back end throughout the day and night. Daytime requests are typically 20 times more than nighttime requests, and with a static threshold, even a significant drop in daytime requests would not trigger any notification. Additionally, performance issues would crop up from time to time when new services were implemented and the NetSeer team had no way to identify them quickly.