Annual Protocol Product Design of Data Management Platform

Annual Protocol Product Design of Data Management Platform

2021｜Type: Platform Product｜Tag: Design for Service Improvment｜Role: Product, UX Design

Background

In the 3C industry, compared with competing media, there is still a gap in the service of our key accounts, and communication still requires a lot of labor and time costs. But the advantage we have is that we have more complete data. The key accounts in the 3C industry are expected to have a more timely and comprehensive understanding of the trade-in data through our product.

It is expected that all four major features will be released in DMP

Stock analysis: Under certain conditions, observe the holdings of the total market and segmented markets.
Sales analysis: Under certain conditions, observe the inflow of the total market and segmented markets.
Competitive analysis: Under certain conditions, the observation group flows out to the market and the market flows into the observation group.
Flow in-out analysis: Under certain conditions, observe the amount and proportion of inflow and outflow in the observation group and control group.

Stock analysis and sales analysis share the same method, and the holdings and the inflow are also two core indicators of the 3C industry, which can be integrated into a market overview.

Annual Protocol Product Design

Refers to the way of cooperation between the advertiser and media, commonly found in groups, generally, they signed on an annual protocol.

The advertiser's budget can be locked in advance for the media, also the annual protocol can be used to exchange resources to create more valuable cooperation opportunities for the advertiser.

The threshold for annual protocol products is 50 million for agreement-based advertising and 20 million for auction-based advertising.

Value Proposition

Through the design of market overview, competitive analysis and flow in-out analysis, it helps 3C industry advertisers to timely understand the trends, the exchange and flow in-out between brands, etc., and help advertisers formulate pre-delivery strategies, optimize during delivery, provide necessary marketing support and post-delivery measurement to assist advertisers in marketing decisions.

On this basis, it is expected to win a key account in the 3C industry to sign an annual protocol and use a budget of 70 million for advertising. In the follow-up, we will continue to win other key accounts to sign the annual protocol with successful cases.

Design Challenges

Computation is huge. High performance is required because of the complex combination of multi-dimensional conditions.
High real-time requirements. The data is updated daily, and the response of the query are required in seconds.
Sensitive to data. Security issues such as data leakage are highly strict.

Performance Assessment

The more complex the conditions of the three modules, the longer the calculation duration, which is far from the demand of the key account for a timely response. In particular, the flow-in-out analysis, which involves a bi-group needs to be calculated in real-time. To this end, we test the query performance of Hermes by building this test. The evaluation results of a single request are as follows:

Query Performance Testing

In addition, in the concurrency test, the number of concurrent requests returned within 10 seconds cannot be higher than 3, and some task flows take more than half an hour.

Through testing, we found that no matter how many brands and their sub-brands, holding periods, and prices are selected, no matter what the dimension is, the calculation duration cannot be achieved in seconds. On this basis, if we need to meet key account requirements, we can only provide the simplest filter, that is, a radio selection of brands. However, the insights of these selections are relatively simple, and it is necessary to ensure the depth and scalability of the insight.

After communicating with the key account, it is expected that the lite version performance can satisfy 100 to 500 concurrent requests and return within 10 seconds. The full performance is similar to the audience insight task, achieved by creating tasks and returning results in 1-3 hours.

Objective

Frameworks are clear.
Responses are timely.
Data is safe.

Design Strategy

Frameworks are clear.

Frameworks are clear, which required business value and information architecture are clear.

Firstly, the business value should be clear:

Market overview: Within the specified observation window, take the time unit as the dimension, and observe the stock and sales of the total market and segmented markets under certain conditions.
Competitive analysis: Within the specified observation window, take the time unit as the dimension, under certain conditions, the amount of the observation group flowing into the market - which includes the brand of the observation group and other brands - and the market flowing into the observation group.
Flow in-out analysis: Within the specified observation window, the observation group and the control group were designated, and the amount and proportion of inflow and outflow were observed under certain conditions.

Which can be summarized as:

Market overview: Window × Dimension × Filter
Competitive analysis: Window × Dimension × Filter for the observation group
Flow in-out analysis: Window × Dimension × Filter for both the observation group and the control group

Secondly, the layout should follow information architecture, which could be divided into condition area and result area:

Condition area: setup for windows, dimension, and filter
Result area: visualize the data, and provide segmentation and download

Layout

Responses are timely.

Two versions are offered for these three features, one is the standard version, which meets the most timely requirements. The other is the advanced version, which needs to submit the flow as a task and wait for the calculating time. In this flow, the conditions include all the brands, and their sub-brands, series, and editions as well as holding period, price, and audience.

The standard version is provided for all the customers of the 3C industry.

The Standard Capabilities

The advance version is provided for all the key accounts that have signed this annual protocol.

The Advanced Capabilities

Data is safe.

In order to protect the security of customer data, the data needs to be opaque during front-end rendering. It mainly includes metrics and units.

On the metric side, In the standard version, the use of protected metrics such as flow in-out amount in the Flow in-out analysis is prohibited.
On the unit side, the unit is 10k, with one decimal place; the percentage or rate is displayed in %, with two decimal places. (Same as ad delivery platform)

Delivery

The kanban in the standard version.

The Standard Pages

The kanban in the advance version.

The Advanced Pages

The submit flow in the advance version.

Fulfill the User Flow Through a Dialog

Data loading exception.

Data Anomalies

What I’ve learned

Take the 3C industry as an example, as well as the automobile industry, etc. are all annual protocol products. The annual protocol products are oriented to match customer needs, and the content required in the agreement should be completed. Under such a premise, taking the Data Management Platform as an example, huge computing puts forward higher requirements on performance. It is necessary to think in advance to complete a feasible design solution.