As an excellent operator, how can you perform real-time analysis?

With the increasing number of operational methods such as flash sales, popular articles, and rapid product iterations, the previous hourly traffic monitoring can no longer meet the needs of operators. We need to record every minute, every visitor, and every operation on the website platform to meet our needs for real-time monitoring and quick decision-making.

This minute-level data monitoring and analysis is what we are going to introduce today as “real-time analysis”.

1. Three major indicators that need to be paid attention to in real-time analysis

There are many indicators that need to be paid attention to in data-driven operations, such as PV, UV, conversion rate , retention rate , etc. Ignoring result-oriented indicators such as retention and conversion, in minute-level real-time monitoring, operations mainly focus on three types of data indicators of the website platform:

Number of visiting users Visiting source Visiting behavior

User visits, visit sources and visit behaviors are of great significance to the operation of the website platform:

The minute-level visits (module 1 in the figure below) can help us understand the traffic trend and detect traffic anomalies in a timely manner; the monitoring of the access source (module 2 in the figure below) helps us understand the real-time access source and weight, and prepare for channel optimization; the real-time monitoring of the access behavior (module 3 in the figure below) helps us understand the user's access preferences, and facilitates the subsequent optimization of website content.

In existing SaaS (Software as a Service) products, the above-mentioned real-time indicator modules are unified into one background page. This design makes it easier for operators to understand the real-time situation at a glance and make plans.

2. Real-time analysis explained through three cases

Identify problems from unusual traffic spikes

Operators generally pay more attention to the PV, UV and trends of the website platform, which are also the basic indicators for website traffic analysis. Traffic analysis based on daily or hourly granularity is relatively rough and will cover up many details of traffic fluctuations at time nodes. What would we find if we observe traffic at a granularity of minutes?

The number of user visits to a certain content community surged from 16:30 to 16:35 on July 16, about four times the usual number (as shown in the circle in the figure above). The community operators immediately discovered this anomaly and found out with the help of [Access Source] that the top access source of this node was WeChat (mp.weixinbridge.com), but they did not know the specific reason at that time. In an article shared in the circle of friends later, it was discovered that a big operations expert recommended the community platform in a WeChat group sharing, which contributed to the surge in community visits from 16:30 to 16:35. The community’s PR decisively seized this opportunity and invited the operations expert to the community to share knowledge, which had a very good communication effect.

This is a typical case of discovering cooperative channels through surging traffic, which is worth considering for all companies. On the contrary, if the traffic plummets or even drops to zero, then you should immediately check whether the website/APP is normal so that you can fix the problem in time.

Precision delivery: channel optimization and anti-fraud

As an operator, if a product has placed advertisements on major channels, you can use [visit source] to monitor the advertising effect of the channel at all times, and then determine the number and quality of visiting users brought by the channel.

An Internet company recently conducted a series of channel delivery tests. Through [visit sources], they found that the volume brought by two of the channels was very small and the price was high, so they immediately withdrew the campaign after a short launch. At the same time, real-time analysis can also be used to prevent fraud. A sudden increase in traffic from a single channel in a short period of time is likely a sign of fake orders or traffic fraud. One morning, the website's traffic experienced two abnormal peaks in succession, and during that period, most of the traffic came from one channel. The operations staff were very alert to this and found out that the agent was cheating and using robots to increase the volume. Afterwards, the company decisively abandoned the agency channel. The above two actions saved the company a lot of losses.

Real-time monitoring makes product operation more efficient

Nowadays, the iteration speed of Internet products is getting faster and faster. Product operations need to track newly launched products or functions and evaluate product effects or market feedback.

In the field of Internet finance, there are cases of group fraud (fraudulent application, i.e. purchasing financial products). Take a certain Internet finance company as an example. Due to risk control reasons, the intensity of external publicity will be controlled, and the number of visiting users every day is basically stable. One day, the Internet finance company launched a new financial product. The company's operations staff found that the number of visiting users increased sharply through [Real-time Trend of Visiting Users], and then found that the number of visits to a certain page in the product was particularly high through [Active Web Pages]. After investigation, it was determined that this was a loophole in the product, which would cause the company to lose a large amount of funds. They decisively took repair measures and put it back online. If we still use traditional traffic monitoring methods, it may take two or three days to discover the loophole, by which time the loss of funds may reach millions.

3. Data-driven refined operations

From the initial "idea" of a product or operating method to its final launch, operators need to use data to measure its performance and market feedback. At the same time, we discover problems from the data, propose hypotheses, and continuously upgrade and iterate; thus forming a virtuous cycle of "idea — product — data" and driving business and customer growth.

During the operation process, the more timely the data feedback, the faster we iterate and the higher the efficiency of the operation. 1.01 to the power of 365 is approximately equal to 38; in other words, real-time analysis can achieve continuous and rapid small iterations, which accumulate to become a huge improvement in operations and the enterprise.