Website Data analysis: Conversion rate of four modules six levels

Intermediary transaction http://www.aliyun.com/zixun/aggregation/6858.html ">seo diagnose Taobao guest cloud host technology Hall

First of all to say a more interesting thing, my little girlfriend has always felt that my body is too greasy, so I entered the glorious weight loss period. Finally, yesterday, 66.7% of all the people I met said that I was skinny. I am very happy to go home weighing, the scale told me, I am still a fat person.

Thought for a long time. I think it was the 66.7% people who cheated me: those 2 were too bad. (yes, that day I met 3 people altogether)

The story tells us: too little data base affects data analysis

A lot of people talk about the conversion rate, but they are all about a certain value of the study, think carefully, solve the problem may be more than one way.

How many layers of filtering did the user finally form the final effective customer?

What variables determine that the customer voluntarily divides itself into a specific classification?

The following is a simple generalization of the conversion rate at different levels, four modules, six levels

Increasing the conversion rate of any one level will lead to a rise in final sales, so when a road goes to the end, you can look at other roads. Each layer is optimized, there will be unexpected results

Traffic is kingly, but the most we should do is to improve the quality of their own shops, so that people into the shop more to stay in the store to become effective users.

This reminds me of some of the evils of thin people (allow me to spit in here), they eat every day in a mess of all kinds of energy substances, and finally even a skinny!! Every day in front of you said: Ah I eat all kinds of food is not fat ah what ... I really think they have nothing to be proud of!! Their conversion rate so low waste a lot of national food!!

Let's talk about the conversion rate.

Most of the time, if you can find something that users are interested in earlier than others, you are the one who makes the most money.

So you should know what will be the best baby to sell in a week?

Determine whether a baby can be a condition of the explosion:

Have sent friends once very confused, so many data software, why give the final conclusion is not the same?

That's because their data models are based on different dimensions, so we need to understand what they're basing their judgment on.

The steps to choosing a potential baby can be summed up as:

Step one: Look for more potential treasures in each category

Step Two: Compare the ultimate potential value of different kinds of potential baby, select one or several potential baby for related optimization

The two charts are the flow chart and sales chart of the top five-ranked baby in the underwear category.

Then we'll find that a lot of things are interesting.

(Artificially ignoring the new baby, because we're looking for the potential of the next few weeks, baby.) )

The first baby stay time is 557.28 seconds, skipping rate is only 12.85%, 62 pieces a week. But the chain growth rate is zero! However, careful study found that he was May 5 just updated products, can not calculate the chain growth rate. This baby is worth watching.

The fourth baby, all the data are very good performance, just last week on the new, need to upgrade the flow of the baby, you can have a very substantial sales.

Let's take a look at these two baby trends.

The fourth paragraph baby data base is small, the analysis result is easy to have deviation, so pull up this kind of baby to observe. This baby in February has been on the shelves, sales can, but there are 2 bad reviews, so after the shelves have a certain impact.

The first baby seems to show a slight downward trend, but compared to sales and traffic, conversion rate is getting better. So there will be a rebound in future performance. Instead of always falling

It can be inferred that the first baby is in the underwear category of the potential baby.

The next step is to repeat the above and compare all the categories in the store.

Summary：

1 Data base too small will affect data analysis, for new product selection potential baby, do not rely on all data software.

2 accurate analysis under different classes will affect the final decision.

3 first focus on upgrading their own store conversion rate, and then consider attracting more traffic. The skinny people who don't eat fat are definitely not good enough.

41 results will have a lot of decision variable impact, to balance all the data, weigh the impact of each data.

5 Data visualization is very important and can help us make clearer decisions.

Original: http://bbs.paidai.com/topic/98112