Information flow methodology helps you reduce costs by 40%!

Let’s first talk about a key principle that has always influenced delivery, the North Star Metric .

When a product attracts users from the front end to the final transaction, it will inevitably go through multiple stages, corresponding to N data indicators. The idea of ​​the North Star indicator is to clearly understand which North Star indicator we should focus on most. Other data indicators are to serve the purpose of completing this North Star indicator. This approach is equivalent to denoising the data in data analysis , preventing us from being affected by other data indicators, so that we can focus more on data analysis and strategic decision-making.

For example, we can try to conduct a simple analysis of WeChat according to the North Star Indicator. Imagine if WeChat’s North Star indicator is monetization, then when we open WeChat today and enter the product’s opening screen animation, I think it will not be a person looking at the moon, but mooncake-related ads. Then, when we enter the WeChat product interface, it will be filled with various related ads. But obviously, we are not harassed by ads when using WeChat now, and there are only a small number of ad spaces . So based on WeChat’s current stage, we guess that WeChat’s North Star indicator may be: “ the number of messages sent .”

For example, in my current business, because the front-end lead conversion costs and the back-end transaction costs have become stable, while ensuring that costs are controllable, our most important task is to ensure the number of leads.

2. Backend guides frontend delivery strategy

With the data indicators in the first step, we have a clear goal. Next, we just need to develop a strategy around this goal. Because we are in the automotive industry, I will use a car to illustrate my marketing strategy. Comparing all relevant factors to the parts of a car, the launch strategy is the working principle of the car. Let me tell you in detail about my delivery strategy:

• Destination: Data Indicators
• Wheel: Agent
• Gasoline : Material Output
• Throttle: Screening Materials - Plan Fission
• Steering wheel: plan classification
• Engine: The back end guides the front end

Destination: Data Indicators

I compare the entire delivery strategy to a car. Our destination is the single data indicator we obtained through the first step of analysis.

Wheel: Agent

The agents we work with are the four wheels. Because they can only see the front-end data, and the advertising strategy should be guided by the back-end, so we develop a set of feasible and cyclical executable strategy plans and synchronize them to the agents.

A good plan must be feasible so as to reduce a lot of communication costs with agents. It must be recyclable so that our partners can enter this circulation system and handle more things spontaneously, which can greatly reduce communication costs and free up time for us to handle more things.

The following is the entire delivery strategy that I synchronized with the agency:

Gasoline: 5w3h method, continuous material output

The regular weekly material output is gasoline as the raw material supply for this vehicle. This step is related to whether our vehicle can continue to operate and provide a continuous output of raw materials for the subsequent closed loop.

According to the 5w3h method, you can get a lot of scenes by cross-arranging and combining various dimensions, and produce corresponding materials based on these scenes, with more than 3 texts for each scene. By renovating the competitor's materials and combining them with your own business, you can get another batch of materials. Through these two methods, you can get a large amount of scene materials to be tested every week.

The following are the scenarios and copywriting output using the 5w3h methodology:

Throttle: Screening Materials - Plan Fission

After the previous step, we have obtained a large amount of materials. In this step, we need to screen these materials and conduct in-depth research on the materials with large quantities. The most common method of targeted in-depth research is to control variables, generally starting from two dimensions: materials and orientation.

1. First, keep the material unchanged and take targeting as a single variable. After many targeting tests, I found car interest tags, car crowd DMP, and car headline number DMP.

These three directions will each have a run-out rate of more than 80%, that is to say, the run-out materials screened out in the previous step can be split into three run-out plans after this step, and the directions are (car interest tags, car crowd DMP, car headline number DMP). In this way, the purpose of rapid increase is achieved, and each plan fission can bring a peak in the amount of leads.

2. Keep the orientation unchanged, take the material as a single variable, and conduct in-depth research on the running material. You can adjust and test dimensions such as color tone, pictures, text, and patterns. Finally, place advertisements to plan fission

The following is how I modified the orientation to fission three volume plans from one volume plan (I won’t show you the specific data)

Steering wheel: Data feedback-plan classification

At this point, we already have a lot of running plans. Now what we need to do is to combine the transaction data of our product backend and label the running plans selected in the previous step. This step is equivalent to the steering wheel of a car to control the direction of the entire delivery.

Category A plan: transaction cost is less than X

Plan B: Closing cost XY

Type C plan: Closing cost is greater than Y

The purpose of this step is that although some plans have a lot of volume, the transactions brought by the back-end are not good. Therefore, if these seemingly "good" plans are increased in volume, it is actually a waste and will lead to higher transaction costs. Therefore, we must combine our own transaction data on the back-end to connect the front-end, so that I will know which plans to increase in volume and which plans to control the budget, rather than controlling the plans through the cost of leads converted by the front-end form.

Engine: Back-end guides front-end marketing thinking

The last step is to label and classify the plans and then implement targeted strategies for volume expansion. The following are the operational ideas for various plans:

A-type plan volume expansion strategy: Do not make any operations to avoid damaging the traffic model, and completely release the budget. It should be noted that A-type plans should be kept an eye on to avoid situations where the audit fails without knowing it.

Type B plan: The budget is released, but the plan situation should be monitored in real time. Type B plan is the focus, and all the expansion strategies are placed on this type of plan.

Type C plan: To control the budget, observe the cost after about 2 weeks. If the cost remains high, abandon the plan.

【Summarize】

• Destination: Data Indicators
• Wheel: Agent
• Gasoline: Material Output
• Throttle: Filter Materials
• Steering wheel: plan classification
• Start: The rear end guides the front end

Gasoline-Material Output

1. Regularly output scenarios: 3-5 scenarios per account per week, with more than 3 copywritings for each scenario

2. Renovation and output of competitive product materials

3. Class A plan material translation and copying

Throttle-Filter Materials

Carry out targeted in-depth research on the selected running volume materials: use the control variable method, starting from two dimensions, materials and orientation.

1. Keep the material unchanged and modify the targeting . The three targetings are: car interest tags, car crowd DMP, and car headline DMP, which can be split into three plans respectively.

2. Maintain the orientation and dig deeper into the material

Steering wheel-plan classification

Combined with the back-end transaction data, the selected plans and fission plans are marked (ABC)

Category A plan: Closing cost is less than X

Plan B: Closing cost XY

Type C plan: Closing cost is greater than Y

Engine - rear end guides front end

After labeling and classifying the plans, carry out targeted expansion strategies

A-type plan volume expansion strategy: do not make any operation to avoid damaging the traffic model, and completely release the budget. It should be noted that: A-type plans should be monitored to avoid situations where the audit fails without knowing it

Type B plan: The budget is released, but the plan situation should be monitored in real time. Type B plan is the focus, and all the expansion strategies are placed on this type of plan.

Type C plan: To control the budget, observe the cost after about 2 weeks. If the cost remains high, abandon the plan.

Finally, the core strategy framework diagram of this article is attached: How the "car" works

Finally, advertising is a dynamic cycle. There is no one-size-fits-all strategy. Advertisements must be adjusted in real time based on data results, re-labeled every 7-10 days of data accumulation, and the back-end data is used to guide front-end delivery, allowing the account to enter a dynamic cycle.

Source: Wang Zhipeng