Thanks for reading my blog. I'm Tin who is passionate about digital marketing. I usually write about the new trends in digital technology, tips & tricks for optimizing digital marketing campaigns. Stay tuned with my blog!

Latest News

    No posts were found.

Share this

How to run a successful online marketing team

Many people have asked me about how to run an online marketing team successfully, especially for a big e-commerce company like Shopee. So in this post, I would like to share what functions and elements an online marketing team needs to operate successfully and healthily.

Looking at the mindmap below, there are 4 core elements to structure online marketing channels: Ads structure, Data structure, Analysis, and Reporting.

Ads structure

Ads structure is like a foundation of a channel that defines the main objective of a channel (user acquisition, order driving, brand awareness, etc.) and also decides the structure of online marketing data by defining the name convention for our ads structure from campaign level to a creative level. If the structure of the ad is defined incorrectly in the first place, then your data generated later on will be useless or lead to wrong optimizations.

The ad structure should be defined at the business level, not an online marketing perspective only which means what is your business aiming right now? (user growth, order generation or revenue) then ads structure will be set up accordingly. For Shopee, we divide our campaigns into 2 kinds: prospecting and converting campaigns.

The prospecting campaigns’ main target is acquiring new customers at a low cost. Its main KPIs are CPI (cost per install), CPR (cost per register), CAC (customer acquisition cost) which let you know how much you have to pay for a new customer and the main metric you should an eye on is CAC as you can pay for very-cheap-CPI traffic sources but most of them are trash, then end up a very high CAC. These metrics will be tougher when your market penetration is higher, usually after a 2-year acquisition then you need to look for new acquisition sources or optimize your customer retention rate.

We divide the prospecting campaign into 2 kinds: Always-on and Promotion based:

For Always-on campaigns, we mainly use dynamic campaigns that pull product info from a product feed and learn user behaviors for targeting. With this method, we can scale up easily as millions of products can be shown without help from the design team and targeting is more accurate.

The 2nd kind is promotion based campaigns which we use initiatives to drive most users and usually, these initiatives usually solve common problems of most users. The promotion campaigns can be based on a seasonality trend like Xmas holiday for instance when customers look for Xmax products to celebrate the holiday, so the products must be specific and selected for these kinds of events and target audiences are targeted as well. However, bearing in mind that audience overlapping can happen if you run multiple promotions based at the same time so make sure your promotions are different and target audiences are also different to minimize the overlap as much as possible.

In addition to expanding the market share, order driving is also the 2nd target for us as customers can explore more products on Shopee as they become loyal to our platform. Then conversion campaigns will be involved in this stage. The conversion campaigns could be driven by many channels like retargeting campaigns, affiliate, direct sales CRM and so on. They are also divided into 2 kinds: always-on and promotion based and heavily spent on the big campaign days.

The ratio between prospecting and conversion really depends on the objective of each business. For well-strategized and long-term focused organizations like Shopee, we can split around 80/20, 80 for prospecting and 20 for conversion. However, for local companies which are very focusing on revenue like Adayroi (the one I used to work for), is spending around 80 – 90% of their marketing budget for conversion as their target is ROI.

The 3rd component of the well-designed ad structure is Naming conversion. I’ve seen a lot of companies don’t have any naming rules for their ads channels or they do have but it is very basic and keeps changing every time a new digital lead comes in. Eventually, they end up with a very messy and unstable database. Therefore, the naming convention needs to be designed by a centralized BI team that makes sure the name convention is being used correctly and data is correct and reliable. The well-designed naming convention will help your team

Data structure

The data structure is the foundation of any optimization activities of a digital marketing team. If your company has a well-designed data structure, the digital team can easily manage and optimize their campaigns at scale. The data can be pivoted at any level that your team wants to deep dive.

For Shopee, we consolidate data from 3 main sources: media platforms (like Google, Facebook), tracking platforms most from 2 sources: AppsFlyer for app tracking and Google Analytics mainly for Web Tracking and Backend data which contains the net data like net orders, GMV, Item sold, etc.

In order to build a well-structured data structure which can long last for years without any major modifications, you should answer these questions:

  • What kinds of traffic sources do you right now? For example, the remarketing report cannot be mixed up with the install report. I saw a lot of companies out there are mixing up these reports together but it is wrong as their targets and purposes are different.
  • What metrics do you want to keep track of?
  • What are your main metrics for each traffic source? CPI, CPO, CAC or what?
  • How deep do you want to optimize? Platform level? Campaign level? Ad set level? Creative level? The deeper you want to optimize, the more complex your name convention is.

After defining a suitable data format, you have to have strong BI guy to build up a data model and make it run automatically every day. Last but not least, your data needs to be visualized in a dashboard that you can easily access and make quick decisions for the channels you are managing.

Reporting

After designing a data structure, the next thing to do is build reports that can help you understand your channels’ performance and enable you to make quick decisions to optimize them on a daily basis.

There are 2 kinds of reports I often use, channel report and sale report.

For the channel report, there are 3 kinds of reports we are using: app install report, retargeting report and web report. As mentioned above, we have to understand the objective of each channel to build a suitable report. Why can’t the app install report and retargeting report be merged? as their objective is so different, app install channels are aimed for user acquisition and the main metrics are CPI, CPR, and CAC while retargeting channels are aimed for sales generation and the metrics should be looked at are CPO and CAC. Some companies calculate the CPI by summing the app install cost and retargeting cost then divided by the sum of installs generated from both kinds of campaigns. But this way is not correct as installs generated from retargeting are very small (the are called reinstalled apps which an app had been uninstalled from the user’s device, then reinstalled after 90 days so this install will be attributed as a new install).

For the sales report, I often look at the sales numbers of the specific categories I’m trying to push traffic to be aligned with the business strategy from the top management as a whole, whether the sales go up after push more effort or just go flat. Some questions you should answer by reading this report:

  • What products are people buying most right now and why? Then just pick 20% of products generating 80% of revenue for running ads.
  • People buy because of promotions or they buy simply because they have demand. What are the main drivers causing people to buy? Free shipping promotions? Voucher codes?
  • Are there any resellers of particular products? If yes, how to leverage them or eliminate them.
  • What products do people usually buy for the first time on your platform for trial? Why don’t create a promotion scheme to encourage them to try and make their first experience on your market intuitive so they will come again.
  • What are your target customers for each category? And why do they buy? For example, for the electronic category, the main target customers are males aged from 18 to 34, living in big cities. They buy because they can easily compare prices amongst online stories and use promotion codes to have a much better price.

After having answers for all the above questions, you can create a special campaign which can full fill most of your customers’ needs and combine with online marketing efforts so you can win the game.

I’ve observed many companies, they failed simply because they only can do a good job for sales or marketing. They usually outsource the not-good job to agency, then the flow between sales and marketing is broken and gets sucked up by the bad execution of whom they outsource to. In my opinion, I will never outsource the job I don’t know to a third-party as in the long run, we will gain zero experience by doing this. I rather than hire an expert and build an in-house team to build from scratch brick by brick, then in one year, the team is definitely better than thousands of agencies out there.

Data analysis

This the most important part of an online marketing team. At Shopee, we have a lot of questions that need to be answered every week. Why did registers increase? Why did organic installs increase? Was it due to fraud? How did airing TVC impact on business metrics? Why are buyers from a specific region increasing? Who are the heavy buyers and why do they buy a lot? Are they resellers? We have to understand the movement of business and customer behavior as well as channel performance to have the most impactful initiative which can lift the business up in the most effective and efficient way. To do this, the business analysis guys need to jump into the field. We usually apply the problem-solving method of McKinsey into every aspect of our job. We breakdown the problem into small parts to understand the root cause of the problem and debate the most feasible and effective solutions for the most significant part of the problem. Then we test it to see whether our hypothesis is true if the uplift is high, we roll out the solution to the mass scale of the business.