Taking APP as an example, let’s share a data idea of ​​advertising placement

The three core indicators of product operation are: new customer acquisition , activity, and retention . I believe every one of you is familiar with it.

The focus of operational strategies may be different in different product stages, but various new customer acquisition activities are certainly familiar, such as H5 activities, advertising , soft-text dissemination, channel exchange , etc.

Each round of campaign activities needs to measure the ROI ratio, and each activity needs to be reviewed and discussed after its completion. Each round of iterative campaign may verify some assumptions.

Our goal is to maximize the output effect by optimizing through various channels with reasonable input costs.

However, in actual operational delivery, we often encounter various delivery methods:

This channel said that we cannot achieve direct jump download;

The channel said, we are a brand, high exposure is our advantage, conversion is none of our business;

Sometimes, there are 3 different locations x 3 materials x 3 ways to jump on one channel. After jumping to the app market , I don’t know whether the app is downloaded or opened. How to measure the effect? ​​I feel a little dizzy...

But in fact, what we care about are the following directions: reasonable investment, optimized path, and maximized output. In short, what we care about is: input-output ratio .

(Attract new customers through different channels)

How to find the maximum input-output ratio is a headache for every partner responsible for attracting new business.

Judging from the process in the above figure, there are several links that we can optimize: whether the product is attractive, whether the channel is of high quality, whether the delivery method is efficient, and whether the effect meets expectations.

We will not discuss product positioning here for the time being, we will only discuss the optimization of the effectiveness of various channel delivery.

I have 100 ways to place your shots. Do you dare to try?

Here, although we take APP to attract new users as an example and discuss a multi-channel effect data analysis system, it is actually applicable to every product, even if it is not an APP product, or your new user acquisition is not centered on increasing APP users, you can reuse this data analysis framework.

Set expectations

First of all, for any campaign, we must define its target output. Only under a clear effect measurement system can the comparison of channel data be meaningful and the ROI of multiple channels be comparable.

For APP new user promotion activities, common expected results may be:

• Increase awareness of the app (for the purpose of spreading it widely)
• Become an APP user (download and activate the APP)
• Complete specific actions in the app (such as purchase, identity verification, or receiving gift packs)

Let’s take the common new user acquisition - “APP download and activation” as an example , and first build a data funnel to see what data we need to observe in the effect link.

If we want to monitor the effect data of ③ in the above content, we will continue to add the next level of funnel after activation to count processes such as adding to the shopping cart, confirming the order, and paying.

Based on such a funnel process, we can build monitoring indicators . Sometimes the monitored data may not achieve the ideal statistical indicators, and we may need to supplement other monitoring indicators. Sometimes such precise data granularity is not necessary.

The data indicators we finally monitored are as follows (Table 1):

As for the performance data, it can be achieved through MTA's new report - installation source analysis.

Why is the “ Network Percentage Used for Downloads ” added here?

The reason is that from download completion to installation, these two events are native events of the APP application market, and developers cannot actively collect them. However, we still care about the conversion steps, mainly because we are worried that if the user is in a poor network environment, the download behavior may be terminated. Therefore, auxiliary field data such as "percentage of network used" is used here.

The auxiliary field data here is mainly prepared to optimize the conversion effect. After determining the goal, our second step is to record the data of the delivery process.

Delivery process record

The main purpose of recording this part of the data is, firstly, to facilitate analysis and comparison during review, and secondly, to provide more direction during optimization and iteration.

According to Eric Ries's "The Lean Startup ": Every experimental iteration is actually to obtain a verified cognition.

When applied to the promotion and delivery system, the design of each delivery strategy is to verify the effectiveness of a delivery strategy and to improve the output effect.

This link is a multi-branch, and we need to make some version records to better locate the problem. The process diagram is as follows:

The recording process table 2 is:

(Table 2 Advertisement delivery strategy record)

*Depending on the delivery method, the recording method may be slightly different

**Specific strategy information can be recorded in the detailed table

Then map the effect data in Table 1 to Table 2 to get a complete record of the campaign. Then you can use the data to verify the effectiveness of the specific delivery strategy.

To verify the delivery strategy of the target, this needs to be planned before the delivery begins, so that effective conclusions can be obtained during data analysis.

Analysis and optimization results

For the large amount of detailed data collected, we can then do some data analysis on the results.

First, we may want to verify the impact of the delivery material on banner-1 of channel 1. Then, if we take channel 1 out of Table 2, we can get the following table:

(Table 3. Impact of delivery materials on banner-1 of channel 1)

If we want to compare the difference between the Banner and Feeds streams on Channel 1, we can get:

We can also conduct data analysis on multiple dimensions such as the effects of targeted delivery to different groups of people, different delivery times, and even channel effect quality comparison, to verify our strategies through data and optimize the input-output ratio.

Therefore, although there are thousands of ways to deliver the content, the effect path is always similar.

If you have a variety of delivery strategies, don't be anxious or panic. Sort out the paths, control the strategy versions, use A/B TEST to speed up the verification and iteration, and finally get an effective delivery strategy and maximized ROI effect.

The above is a sharing of the data indicator system for delivery effects. Similar analysis ideas can be reused in delivery plans for different needs, including some online and offline operational activities. In fact, this framework can also be reused.