If business performance metrics are the cheerleaders of a marketplace’s success levels, then software performance metrics maintain the stadium in which they perform. The reality is that software performance will always have an outsized impact on user experience, revenue, and overall business success, whether you understand the technical intricacies or not.
The challenge is often that marketplace startups are aware that there will be issues that they know they don’t know (known unknowns), but more challenging are the issues they don’t know they don’t know (unknown unknowns).
The answer is to use the right metrics to monitor how systems are performing. Then you will be able to identify, track and manage the knowns, the known unknowns and the unknown unknowns. In this post, we will cover seven key performance metrics that marketplace operators can use as a departure point.
Why software performance is crucial for online marketplaces
Accurate information on software performance metrics helps to pinpoint areas for performance improvement. This is the information that marketplace owners can use to learn about those unknown issues.
In the early stages, a marketplace startup won’t have a clear forecast of traffic spikes (known unknowns), so the capability to monitor performance in real-time will allow developers to make application or server changes to manage scalability.
One of the biggest design issues that marketplaces face is optimising the user experience. With appropriate performance metrics, user issues like fast load times and search can be tracked, and changes to applications or memory usage can be re-designed.
Marketplace software performance priorities vary
The software performance priorities of a marketplace will depend on the type of marketplace and the individual business model. For instance, a services marketplace, like TaskRabbit, will require a different software infrastructure to an interactive platform such as an auction marketplace like eBay, or a streaming platform like Netflix, which both need to support high volumes of concurrent requests.
In terms of business model, some typical differences include:
- B2C marketplaces prioritise fast load times and smooth user experiences to drive impulse purchases.
- B2B marketplaces may focus on robust search, filtering, and comparison.
- Niche marketplaces tailor performance to specific product categories
- General marketplaces need to balance performance across a wider range of products.
- Subscription-based marketplaces prioritise stability and reliability over speed.
- Transaction-based marketplaces need to optimise for peak traffic and checkout speed.
Marketplace software performance metrics
There are a few performance metrics that marketplace owners can use to evaluate and assess the efficiency, responsiveness, and overall health of their platform. These provide insights into how well an application is performing, from both a user and technical perspective.
1. Speed (Throughput and Response Time)
Speed is one of the key business requirements for a good web experience and to drive conversion.
Response time is one component of speed that measures how long the platform’s servers take to respond to requests from users. It is important to monitor response times as this directly impacts users who want quick access to a service or application. This is done by calculating the average response time – total response times divided by total number of requests over a specified period of time. A low average response time implies better performance in terms of responding to requests or inputs.
Throughput is used to refer to the number of requests an application can handle within a given period of time. It is often measured as requests per second or transactions per second. If throughput is consistently observed to be reaching a maximum limit, it can be an indicator that the server or application infrastructure needs to be scaled up. This may lead to deploying additional servers or rewriting the application to be less resource-intensive.
For an e-commerce marketplace, response time is critical as each additional second of load time can reduce conversion rates by more than four percent.
2. Capacity
Capacity indicates how the marketplace handles increased future load demands with additional hardware resources. As user growth increases, monitoring resource utilisation (like CPU, memory, and network bandwidth) is essential.
Database capacity can be scaled horizontally or vertically to support growing user demand, but the choice of infrastructure architecture will have cost and performance implications. This usually boils down to choosing between relational (SQL) and dynamic (NoSQL) databases.
Poorly written software can result in an excessive capacity load. From a business perspective, insufficient capacity leads to high response times and constrained throughput, which have a negative impact on user experience.
3. Efficiency
Efficiency is the amount of compute resources required to complete a transaction. Efficient resource usage is crucial to ensure optimised response time and infrastructure cost. An application is considered efficient if it handles business transactions with reasonable resource consumption, and with acceptable response time and throughput.
The importance of efficiently written software has a direct bearing on capacity consumption and thus cost. If software is not written well it can lead to poor user response times and excessive capacity consumption.
4. Scalability
Scalability measures the ability of a marketplace to handle increased workloads without performance degradation. It assesses how well and quickly a system can scale up, which is essential for managing fluctuating demand over time. Performance indicators include response time, throughput, resource utilisation and availability.
If a system is not scalable, speed and resilience will be impacted. Poor scalability leads to degraded response time and throughput.
Technical factors that affect scalability include hardware resources, software architecture, and database design, which can be improved in different ways such as load balancing, caching, queueing systems, microservices architecture and database sharding or replication.
5. Stability
Stability refers to an application’s ability to maintain consistent and predictable behaviour over a long period of time. A stable application avoids crashes, errors, and performance degradation, ensuring a seamless user experience.
A stable application ensures the cost of infrastructure will not need to be unexpectedly upgraded. It also means the customer experience remains good. A low and stable response time over a long period of time is a good indicator of stability.
Marketplace owners should be aware of whether developers are following best practices to ensure software stability: semantic versioning, test early/test often, regression testing, and detailed up-to-date documentation.
6. Resilience
Resilience is the ability of an application to recover from errors, disruptions, or failures while still providing the best possible service. A resilient application easily handles failures, either by automatically recovering or by minimising the impact on users. It includes mechanisms like failover, redundancy, and disaster recovery plans to keep the application operational under adverse conditions. It therefore plays a key role in a marketplace to maintain user satisfaction, security, and business continuity.
One way to maintain resilience is to add a secondary system, this is adding redundancy. The downside of course is the added cost. Software can also be made more resilient by using code libraries such as Resilience4j.
7. Availability
Availability is the percentage of time a marketplace is operational and accessible to users. It is crucial for marketplaces because they rely on continuous availability for a good user experience.
It is usually measured using indicators such as percentage uptime, number of completed transactions, and error rates. For high availability, uptime needs to be over 95%. For example, a 99% uptime means the system is down for about seven minutes in a month.
Every minute a marketplace is down directly flows to the bottom line in terms of lost sales. For example, in 2013, Amazon had store downtime for 40 minutes, which cost an estimated US$5 million.
Identifying the right marketplace software performance metrics
Tracking performance without knowing what constitutes good or bad is pointless. For example, good response times and throughput won’t necessarily ensure higher conversion rates. A product search feature should thus prioritise accuracy over speed.
Identifying the right metrics to optimise your marketplace software performance is not a simple affair. That’s why marketplace startups should start looking at software performance metrics during the early stages of development.
Using a prototype or Minimum Viable Platform (MVP) during the discovery phase allows you to test whether the tech stack and feature set are the right ones in terms of scalability, flexibility, speed and reliability. At this stage, you can start to uncover those unknowns using the performance metrics.
It can be tempting to choose an off-the-shelf solution for this purpose, but history has shown that going this route can lead to serious performance bottlenecks. It is very common that marketplaces launch with a turnkey marketplace builder that seems to work initially, but as the site scales a multitude of technical problems usually reveal themselves.
When fast-growing neobank, SoShop, hit 50,000 transactions per day their platform started buckling under the pressure. The solution was a custom tech stack that could scale as their business grew. Affordable Art Fair, a marketplace for contemporary art, also hit a dead end when their Magento template became difficult to adapt, while slow site speed alienated users. A custom platform helped them increase turnover by 50%.
A product roadmap for your marketplace can help to map performance priorities. By incorporating performance metrics into the roadmap, you can use data on the metrics to plan and prioritise the performance indicators for each item on the roadmap. The roadmap can also be used to involve different departments to ensure they get what they want from the product. Performance-related features or functionalities can be used to align with product-market fit, such as functionality or customer acquisition.
Using third-party providers and services to improve performance
These days online marketplace startups don’t need to build each component of their technology stack from scratch. Instead, they can rely on numerous cloud software solutions from providers like Microsoft Azure or Amazon Web Services to avoid having to fund upfront capital investment and get the benefits of the dynamic scalability that the cloud enables.
Using a cloud service, however, does not obviate the need for performance monitoring. In fact, monitoring is necessary to ensure that performance bottlenecks don’t occur and that any increase in resource usage stays within the budget.
Integration with services like payment gateways needs to be monitored so that problems like bottlenecks or gateway downtime are quickly spotted and fixed.
An experienced software development agency can also be a powerful resource. They will have the skills and insights to help identify relevant software performance metrics, implement the right tech stack for those metrics, and then monitor them.
Testing your marketplace’s software performance
Part of the MVP stage is to test the marketplace’s performance. Without instrumentation, marketplace owners will not have the data to understand how to track software performance. Application performance management (APM) is software that monitors and manages the performance of an application. APM tools can help identify problems like latency and system failures, and provide the information to monitor the other six measures.
Different APM tools may need to be used as they tend to specialise in particular areas. Sentry is an exception-monitoring tool that can alert developers to the impact of errors and where they originate from. Another tool, Amazon CloudWatch, collects data from logs, metrics, and system events, and presents it as a dashboard so that developers and application administrators have greater visibility into the health and performance of an application.
During the test phase, marketplaces should use their APM tools to carry out load and stress testing, which allows developers to see a system’s behaviour under normal and peak load conditions, and when the system is liable to break down when it is pushed beyond its normal operating capacity. Other tests, such as endurance testing, are used to identify potential performance degradation under a sustained load over an extended period, or spike tests which simulate a system’s response to sudden traffic surges.
Technical health as part of marketplace software performance
A marketplace’s technical health looks at the performance, reliability, and efficiency of its underlying technology. Examples of good technical health are fast load times, minimal errors, and high uptime. Understanding technical health requires data which comes from the APM systems implemented.
Sentry is a good APM system for monitoring errors. CloudWatch can provide the real-time information needed to monitor load times and system availability.
Conclusion
Software performance is not just a technical aspect of an online marketplace, it also drives business success. By prioritising performance, marketplaces can create exceptional user experiences, increase sales, and build a sustainable competitive advantage.