How can the marketing customer acquisition team be data-driven?

How can the marketing customer acquisition team be data-driven? This article starts with an analysis of the current market launch problems, analyzes the current market launch architecture, and then proposes the concept of a data-driven launch market, and explains the process and implementation of this type of market from many aspects. It is recommended for people who are interested in data-driven launch markets, and I hope it will be helpful to everyone.

The results of consulting and research on some Internet companies show that many Internet companies do not make good use of the resources at hand. There is still a strong path dependence in the matter of marketing, which may be related to the fact that most of the marketing personnel are transformed from traditional marketing personnel.

The underlying logic of this article is that Internet businesses can be divided into multiple independent, iterative small closed loops. Within the closed loop, data and algorithmic decision-making can be used to reduce the requirements for practitioners and make them as SOP-oriented as possible. So we hope that this article can expand everyone's thinking about market acquisition.

1. Market acquisition team drives growth architecture

1. Problems with current market launch

The marketing department will have work contents such as brand promotion and customer acquisition, among which brand promotion inevitably involves brand placement and customer acquisition placement in channels. Usually, the current marketing department’s placement workflow is similar. Suppliers are used to provide topics, and then large companies determine whether the topics are feasible. After that, suppliers will submit KOLs for placement.

Figure 1 New media placement process for marketing personnel

Among them, the content screening and selection in new media is done by KOLs or UP hosts. Most KOLs and UP hosts follow the hot spots, and some UP hosts will plan the direction and rhythm of the content themselves, and use third-party tools to understand the changes in market information. The disadvantages of this process are:

1. This process results in a long delivery cycle

The overall process chain, from brief to agent to MCN to number selection to KOL, takes a very long time for final feedback and iteration. Such an iteration cycle is difficult to cope with the rapidly changing tastes of users in content consumption.

2. Most KOLs and UP hosts do not have strong content control capabilities

KOLs who are sensitive to content usually have many fans and high advertising costs. The number of KOL fans also follows a power law distribution. Most KOLs have a small number of fans due to their limited sensitivity to content and limited communication power. There will be a situation where the CAC price of top KOLs is expensive, while the CAC price of long-tail KOLs is low.

Do the fans of top KOLs necessarily contain a large number of target users? Possibly, but the cost of acquiring customers will be very high. Only by helping KOLs and UP owners select content and directions can the ROI of advertising be improved. Using KOLs to spread advertising can only obtain an average CAC.

3. The traditional delivery process may not be able to hit the target audience

This form of generating a brief and then looking for agents to find KOLs is because the fans covered by KOLs are fixed, or the KOLs in the hands of agents are fixed. In fact, their channels are fixed, which means that the target audience is almost fixed as well.

2. Why drive growth through technology?

Based on the above background, a data-driven process becomes particularly important because there is almost no arbitrage space and it is difficult to widen the gap between the media and marketers in content selection.

1. The acumen of media professionals alone is unlikely to outperform high-end marketing talent

If the media chooses the topics themselves, it will become a competition between the content sensitivity and topic selection of high-end talents in the market. Then the high-end talents in the industry will definitely win, and such high-end talents usually do not seek jobs in companies.

2. Due to over-reliance on people, any personnel changes will affect the content effect

This is a problem that will arise in all advertising and strategy making, because all strategies and advertising are decisions made by people. Once people leave, the previous successes and failures are taken away, leaving no sediment for newcomers. Newcomers have to start from scratch again.

3. People cannot guarantee that they will make the right choices all the time

Relying on people to make choices is a subjective choice, which is difficult to attribute and find a basis. It is more based on job level and the prestige accumulated from previous successful cases. Subjectivity is prone to disagreement because it is difficult to reach a consensus within the team from some dimensions based on subjective judgment.

So how do we think Internet companies should acquire customers and deliver content? The biggest advantage of Internet companies is their ability to digitize algorithms and data. The former can process massive amounts of data, while the latter allows users’ and internal staff’s operations to keep data assets within the company.

These are what ordinary agencies, 4A companies, and MCN companies do not have. Although many current tools can provide a summary of the latest popular information to facilitate content producers and manufacturers to screen topics, they cannot provide data precipitation based on previous user online feedback and historical statistical trend analysis.

3. Data-driven market launch mechanism

Therefore, we want to provide a process that can be driven by data to decide on the content to be delivered. This process does not require a high threshold for the professional ability of the delivery personnel. It is more about driving customer acquisition through algorithmic decisions based on organizational structure and data.

Data-driven delivery process

1. New process driver description

Data scientists and R&D personnel will first complete the screening of KOLs in the distribution channels.

1. First, filter the KOLs through the users you have found and crawl the data. First, clarify which KOLs can cover these indicators and dimensions.

2. Then find these KOLs through agents and MCN companies, and find out whether the agencies have channels to these KOLs.

After selecting a KOL, since the interaction between KOL and users is the closest, content submission can be done in two ways:

2.1 Brief direction of the content given by KOL.

2.2 Content Decision Model The direction of the Brief is given based on the analysis trend of historical data.

The brief submitted by the KOL passes through the content screening model, which gives an estimated number of views based on the historical content browsing.

3. You can use the company's private domain traffic, especially the target group's fan base, to conduct grayscale or AB experiments to observe the user effect. Considering the difficulty of video modification and iteration, moderate grayscale experiments are necessary before large-scale deployment. Through the feedback of the fan base, you can determine which content to launch in the end, or what to modify. Then launch on a large scale.

4. In addition to the data on various content platforms, the data should be visualized. The purpose of visualization is to make the best effort to see the source of customer acquisition, and ultimately to reduce the cost of obtaining data.

2. Program algorithm in the new process

In the new delivery process described above, which parts can be carried out by programs and algorithms? We believe there are three links:

01「KOL screening model」

03「Filter content model」

05「 Delivery Result Model」

In terms of implementation difficulty, "01 KOL screening model" is the easiest, and the most difficult screening model is "03 content screening model"

「01 KOL screening model」, capture the information set by each fan on Douyin, and then determine whether each KOL contains the target users and whether the number of users is sufficient.

"03 Content Screening Model" requires manually abstracting the content into multiple labels. This part of manual abstraction is the core work of the marketing department. The algorithm part of "03 Content Screening Model" is mainly to give the relationship between labels and exposure, and to find statistical confidence.

By continuously breaking down new content into tags and using data such as the final page views of the content, the content screening model will become more and more accurate.

3. Core tasks of the marketing department in the new process

First of all, content prediction itself is a very difficult task.

If prediction is needed, the content needs to be broken down, from being unquantifiable to being comparable in multiple dimensions.

Because if any content cannot be broken down, there is no way to predict it. If the traditional method of selecting materials by marketers is used, it can only reach the average level of the industry. It requires more selection ability than the average level of the industry, and people who are very sensitive to content will eventually leave the organization. Therefore, all the endowment resources of the marketing department rely more on the research and development and data capabilities of the Internet.

What marketers are good at is breaking down content into simple elements, and then using technology and data capabilities to perform attribution analysis on all videos with good or poor delivery effects. This is used to train AI decision models. As time goes by and the amount of data changes, the model's judgment will become more and more accurate, and the results of the decision model assisted by these data can effectively improve the success rate of content.

2. Implementation of data-driven marketing processes

Next, we will talk about how to implement this data-driven market from five aspects: data acquisition, system software tools, workflow, organizational structure, and job responsibilities.

1. Data Acquisition

From the perspective of the market or the company, the core of the marketing department or the acquisition of delivery data is to see the channels and sources of each user, to associate the channel sources with the subsequent consumption of the customers, to know the users coming from a certain channel, and how their subsequent consumption transactions on the platform contribute to the platform.

Usually the marketing department will be exposed to the following data:

1.1 Exposure data of the delivery

The delivery data can be obtained from the delivery platform. Each delivery platform (Kuaishou, Douyin, Bilibili, Xiaohongshu, Zhihu, etc.) will record the exposure UV of the relevant materials.

1.2 Customer acquisition channel traffic data

It is usually divided by channel, and key business behaviors such as registration, activation and other indicators are added to construct a matrix data.

Usually, in the channel detection part, we complete the association between the channel source and the customer's subsequent consumption.

1.3 Core transaction business data

The SKUs, order amounts, and number of orders purchased by users on the platform are data of this type. They are usually associated with user channels based on their channel IDs, so the contribution value of the channels can be clearly seen.

1.4 Content material data delivery

This part of data is unique to the market, namely the topic of the article. By analyzing the first-level data 1.1 exposure data, 1.2 channel traffic data, and 1.3 core transaction data, we can see from multiple dimensions the user quality and subsequent consumption contribution brought by each content dissemination.

2. Tool empowerment

Whether it is for an organization or an individual, if the average cost of acquiring something is very high.

(Here we don't consider that some experts can write SQL or extract data using data templates.) If you want an organization to continuously obtain data, data-driven is impossible. For example, you put a household appliance in a very difficult place to find it. You have to clean it after each use. Believe me, it will be covered with dust in a short time.

Therefore, the core of tool empowerment is to reduce the implementation cost of a team. Software tools are like lubricating oil on a bearing. The better you develop them, the less resistance there is and the faster the business will run.

The purple part is for tools that drive business and reduce driving costs. The better these tools are, the more they support the business, the lower the implementation cost of each operation will be, and the faster the iteration will be. There are two main ideas in general. One is the downward gray arrow, which continuously abstracts the business into automated tools, and the other is the upward arrow, which continuously empowers organizations and businesses through automated tools.

Many companies we have consulted with think that it is normal for them to iterate a feature or project every 1-2 months. However, if we can use tools to empower the organizational structure, we can actually increase the frequency of iteration.

3. Workflow

For the marketing department, a part that spends resources, the most important thing is to plan the market traffic acquisition plan as a whole and plan the allocation of resources as a whole.

The overall delivery consists of two parts: budget estimation and traffic estimation. Since most companies have to pay for traffic acquisition, the traffic estimation can basically calculate how much you need to spend.

So budget and traffic are two sides of the same coin. Let's talk about the workflow from the perspective of budget. Generally speaking, the market starts with planning for the whole year and preparing ammunition for traffic at each stage. So the workflow is planning for the whole year, planning for each quarter, and then implementing it on a weekly basis, and constantly adjusting. I don't want to waste words here to prove the importance of traffic planning. I just say that plans will definitely change. Making plans is not for strict implementation, but for thinking and planning.

The market traffic budget estimates the budget and traffic based on the business goals of the enterprise. The budget constraints are ROI and profit margin, which can be used to limit the budget.

The second is to look at the core budget that supports the strategic goal: Is there a long-term strategic plan to achieve net profit or seize the market? What capabilities are required and how much is the expected investment?

Bottom-up approach:

Vertical quantity: natural increment (time series/smoothed index)

Horizontal quality: Based on the strategic calculation of operational improvement (potential for improvement), i.e., quantitative conversion. Incremental space for upstream customers (time series prediction); account improvement value for customers of different sizes (combined with strategic estimation, etc.)

The logic for attracting new customers is similar, and corresponding estimates can be made based on the current traffic conversion rate.

Top-down approach:

The company's total GMV target -> the annual contribution of historical customers -> how many new customers are needed

4. Organizational Structure

The overall organizational structure of the Strategy Department consists of the following parts:

Media channel placement:

Mainly responsible for managing the delivery of each channel, it can be maintained according to different channels. The advantage of doing so is that you have a better understanding of the users of each channel and it is easier to calculate work performance.

Internet delivery:

The main thing is to have a better understanding of some technical delivery methods, such as oCPC delivery, etc., to be able to communicate with relevant developers, and to know some rules of the Apple Store and Android App Store.

Market middle platform:

Products and R&D are more about providing tool systems for the delivery process, continuously reducing the cost of implementation.

DS (Data scientist), BI, and data engineering:

BI work content: Cooperate with various channel parties and CMO to convert available data into T+1 dashboards.

DS job content: Mainly to break down data to give predictions including KPIs and strategic interventions for various channels and business personnel. Assist in guiding the experiments and strategies of various business parties.

Data Engineering: Systematically collect and store the data needed by market colleagues.

Finance and Risk Control:

Job Description: Mainly responsible for resource planning and capital expenditure strategy. Participate in BI and DS to provide budget strategy based on traffic and conversion rate, calculate the discount strength of each activity, control the cost of traffic acquisition as a whole, and cooperate with operations or sales to provide product pricing.

Event operation and category operation:

Job Description: The overall work includes designing and planning operational activities, including preferential strategies, applying for corresponding budgets, configuring corresponding activity tool strategies, and connecting with product R&D for customized development.

Activity operation: Generally, operation activities are planned by a unified team. These people tend to be operational strategies. Then they are launched in various channels. Of course, each channel can organize and plan operation activities according to its own characteristics.

Category operation: Determine whether to configure independent category operation to manage the merchandise for the event based on workload and revenue. When your platform has a very rich SKU, you can assign a separate person to plan categories for attracting new users.

Content and brand:

Job Description: Correspond to content producers and KOLs, and review the briefs submitted by KOLs through cooperation with DS. And evaluate and measure the effect of content dissemination through DS and BI.

In the middle office, every time a campaign is launched and new customers are attracted, DS, finance, risk control, event operation, category operation, content and brand are departments that work closely together. The core logic will use data to evaluate prices, risks of campaigns, and the value of content. Evaluate the categories of campaigns.

5. Work responsibilities

According to our understanding of the marketing department, the core indicators are as follows:

• Number of new app downloads (new device ID or new device ID + new account)
• Number of registered users acquired through marketing channels
• Number of first-time new customers, GMV of new customers (traffic outside the APP)

(1) Number of new app downloads

The number of new customer downloads in the APP should be the responsibility of the marketing department. Overall, ASO and the content of the app store should be the responsibility of the market. From the funnel model, we can see that users will only buy when they have awareness. Therefore, if the Awareness (brand awareness) of new customers is the responsibility of the marketing department, then the download of new customer APPs should be in the market, and the overall brand promotion of ASO will also tend to favor new customers. Then ASO will also belong to the marketing department.

Responsibilities of marketers and operators (operators after activation)

Additional note: The reinstallation of old customers should be borne by the person in charge of APP operation. Since the old customers have already established contact channels, the recall of old customers belongs to the retention operation, and the operation team should be responsible for this indicator.

(2) Number of registered users acquired through marketing channels

The main consideration is who can do the work and has the authority to influence the users in this link. Since the marketing department is responsible for user acquisition, getting users to download or register an account belongs to the marketing link.

Additional explanation: Prevent over-promising, such as buying low-quality traffic, with a large number of registered users but a very small number of first-purchase users. The marketing department can have a clear view of the overall channel, from exposure UV, registration number, APP downloads, and first-purchase GMV. However, considering that the user registration to first purchase is affected by product experience, SKU, etc., it is not necessarily a conversion problem caused by traffic. Therefore, the number of registered users in the marketing department is a more reliable indicator.

(3) New customer first purchase, new customer GMV (traffic outside the APP)

The marketing department can let users experience consumption and purchase outside the client terminal first, and let users experience their first purchase outside the terminal so that more users can achieve deeper market awareness and brand recognition. Since the outside terminal is where the marketing department works, the first purchase of new customers outside the terminal and GMV should be the responsibility of the marketing department.

In general, the brand marketing department is responsible for more of the upper funnel, such as market awareness, new customer app downloads, user registration, which is equivalent to customer acquisition. Of course, it can also plan fission activities and referrals. The external end is responsible for purchasing revenue.

However, purchases involve operational roles and capabilities such as subsidy calculation, product selection, and event planning. From an organizational perspective, the marketing department can recruit employees with relevant capabilities, or the operations department can provide relevant and capable personnel specifically for market connection. If the base is under the marketing department, these methods are all acceptable.

Author: Arun's Growth Research Institute

Source: WeChat public account "Arun's Growth Research Institute (ID: arungrowth365)"