Refined delivery operation of huge amount of Qianchuan delivery

Part.01 Data to look at before launching

01 Live broadcast room data

The first thing to look at is the transaction trend chart. The most important goal of our live streaming sales is to close deals. The ability to close deals is the basis for all other work. Without deals, everything else is empty.

Whether the transaction trend chart is stable is the key factor in whether the pitcher will make continuous delivery. If the transaction trend of a live broadcast room shows great ups and downs, then the guy asserts that half of the Qianchuan expenses are wasted. It can be said that the transaction trend determines the marketing budget.

Pitchers all like live broadcasts with stable transactions, which can rise smoothly but never fall continuously.

Next, let’s look at the popularity trend chart. Popularity is the data that determines a live broadcast room’s ability to retain people. Every time Qianchuan pushes a wave of people in, but the live broadcast room’s ability to retain people is poor and its popularity fluctuates obviously, this means that he is not a good anchor.

When people come in the anchor room, some will naturally leave. A good anchor will use all possible tactics based on real-time popularity data to stabilize the number of people in the room. Consumers have a herd mentality, and if there are not many people in the room, they will basically make subjective judgments.

If there are not many people, at least do a good job of converting old fans, so that old fans can see the host's personality and efforts, and continue to see the popularity increase over a period of time. When a threshold is reached, retaining new fans will become less difficult.

Looking at the interaction trend chart again, when the interaction starts to increase rapidly at a certain point in time, it is either a welfare tea or a hot-selling product. This is the best time to attract interaction, so you can grasp the rhythm of the live broadcast room based on the interaction trend chart. More interactions = greater weight of the live broadcast room, which is a major factor in the system giving away free traffic.

Then look at the user portrait of the transaction, because product purchase is a user behavior, and the demand for each product determines its respective user portrait. What kind of people are buying the products in the live broadcast room? Are there more men or women? Do people in the south or the north like it? How to identify who likes our products more, young people or middle-aged people? The pitcher is to dig out the people with the highest conversion rate from all Douyin users and send them into the live broadcast room.

Finally, looking at the traffic structure, we need to see how the traffic sources of each live broadcast are distributed, whether there are more old fans or new fans, and more free fans or paid fans. The pitcher needs to ensure that the traffic structure of the live broadcast room is healthy, so that it can be a sustainable live broadcast room.

If it is a mature live broadcast room, the proportion of paid traffic during the delivery process should not be too high. The reasonable range is 5%-15%, and it is best not to exceed 30%. If the paid traffic is too high, it means that the product click-through rate and payment rate are relatively low, and it may not be able to leverage the increase in ROI.

02 Product data

When we were advertising, we often encountered the problem of not being able to advertise. No matter how much money we added or how many people we attracted, we could not increase the volume. The basis of Qianchuan’s sales is display. Although there are many goals to choose from when building a plan, in the end, the system will automatically convert it into “the value of every glance”. On your side, you can only get 0.2 yuan in profit for one recommendation, but on other companies you can get 0.3 yuan. So of course I will sell this recommendation to other companies.

So when looking at product data, you are constantly looking for the next hot product. Product A has a good click-through rate, but the order rate is a bit low, so you need to find ways to improve the conversion rate. The sales volume of product B is not bad, but it is achieved by selling it at a low price, so the selling price needs to be increased to find a balance between price and sales volume.

In addition, you also need to create your own hot-selling product data model. For example, for merchants selling department stores, the product click-through rate is more than 20%, the conversion rate is more than 10%, and the customer unit price range is between 10-50 yuan. This is the basic hot-selling model of the industry.

03 Marketing Data

If the live broadcast room data and product data are good, it is time for Qianchuan to attack, which is what we often call using funds to accelerate the closed loop, and the funds here are Qianchuan's budget.

Now is the time to do the accounting. Add up the product cost, operating cost, placement cost, and return amount, and compare them with the transaction amount to see whether you are making a profit or a loss.

For example, for clothing, if the ROI of general-priced goods reaches 7-8, you may be able to make money. With 15% investment, 40% product cost, 10% operating cost, and 25% return amount, you will have a gross profit of 10%.

For department store items with low unit prices, the ROI is basically only around 2-3. The cost is low and the return rate is low, so even with such a low ROI, you can make money.

After completing these three data checks, I understood the relevant data of the live broadcast, recorded it, analyzed the suitable users, selected the hot products, and spent money to increase traffic. I think the results should not be too bad.

Part.02 Qianchuan Project A/B Test

When choosing plan goals, the maturity of the live broadcast room should also be considered. In the early stage, clicks, increasing fans and placing orders will be carried out in parallel. In the later stage, for a mature live broadcast room, you generally only need to place orders and complete transactions.

Entering the live broadcast room: popularity is the main factor, production deviation

Clicks on products in the live broadcast room: Mainly popular, in production

Back to the AB plan, in fact, if you don’t look at any information about products, live broadcast rooms, etc., you can build thousands of plans, because there are so many dimensions to choose from. Just based on the age, gender, and region of the population, you can arrange dozens of plans.

But not all plans are necessary, otherwise there would be no time to build a plan and test it.

Finalize the certain factors and classify and test the uncertain factors. This is the correct operation of the AB plan.

For example, if we are focusing on light luxury style in women's clothing industry, then the gender is female and the age is 25-45 years old, which is very clear. As for the region, we can establish four dimensions of 1st-4th tier cities. If we do the test with all other settings the same, we can get the result.

Remember, AB testing must be a single-factor test. If there are two variables, the results will have no reference value.

After all, the AB plan is designed to help us capture the best people. Today I tested the regions and found that the data in first-tier cities was the best. Then tomorrow, I want to know whether the girls who like my clothes also like food, home furnishings, beauty or travel. At this time, I have to create multiple plans to test and obtain data.

I will even test the lower-tier regions. Perhaps the conversion rate among girls with certain interests will be particularly good.

If you have a large budget, you can make more plans. Generally, each live broadcast will be measured in several aspects for comparison. Once the data comes out, you can plan tomorrow's test content.

When it comes to plan names, every pitcher has his own style. Generally, when building a plan, he or she should follow his or her personal habits. Factors such as age, goal, gender, region, interest, bid, etc. can be included in the name. The key is to efficiently get the desired results when doing data analysis.

Behavioral categories and interest groups are generally more accurate, but the expert group has a stickiness problem, and it may be more difficult to grab the volume.

Bid from low to high, but remember not to modify the bid all the time. You can use the method of copying the plan to increase the bid, and slowly test out a bid that can increase the volume.

Run multiple plans in parallel, compare consumption and ROI, and select the best and eliminate the worst, especially the high-quality plans. Quickly copy the plans and bid 1% higher to grab the volume.

To sum up: plan type + crowd targeting + bidding method + creative type to conduct data testing and horse racing, find high-investment and low-cost plans, find accurate value traffic, and improve ROI

The purpose of data review is actually to optimize the plan. According to the consumption and ROI of the series plan, we can determine which plan in the series is good and what is good about it. If every plan in the series is very poor, we also need to find the reason, whether it is because the bid is too low and the product cannot be put out or the target audience is too narrow, which affects the consumption rate.

The specific operation is as follows: first, all plans are stratified into high-quality plans, medium plans and deviation plans.

Establish a daily summary table of important data to see whether the data is in a stable period or a slowly rising period. If any abnormalities occur, hold a meeting in time to summarize and find out the root cause and where the problem occurred in the flow of people, goods and venues.

Qianchuan is not omnipotent. Live streaming with goods must be an organic combination of "people, goods, venue and flow". There are many tricks involved. I have only mentioned a few of them. We welcome more friends to discuss together.