How a CMO Can Build an Effective Marketing Analytics System

Businesses that collect data, analyze it, and make informed decisions based on that data, grow and develop faster than those who don’t. This article explains how to build an analytics system for your business and why martech tools and analysts are essential.

What is marketing analytics? It’s a tool by which a company evaluates what generates profit. However, analytics itself doesn’t directly bring income to the business. With analytics, the Chief Marketing Officer (CMO) can only get various reports in order to draw conclusions and make decisions.

Let’s take an example from our practice at OWOX. The usual task for a CMO is to execute an online advertising sales plan. How can analytics be useful in this case? Analysts might provide Google Analytics reports, ad cost data only from Google Ads, and data for online transactions on the website that doesn’t match the data from the CRM. As a result, calculating ROAS will be tough.

Let’s say a company has been using analytics for a long time, has collected data in cloud storage, and has configured reports that are created automatically. Does this solve all the company’s marketing problems? Not really. After all, marketing directors are interested not just in data but in data that will help them understand what opportunities are available in order to execute the plan for the next month or quarter.

In order to accurately perform the plan and achieve all their goals, marketers need analytics to identify:

• growth zones (how to exceed the plan)
• risk zones (what interferes with implementing the plan)

"Before thinking about two zones, I'd emphasize hiring the right people to help build that system.

CMO's are not the ones who should be building the system. CMOs should do more of a strategic thinking, choosing and controlling the direction of the marketing, etc. You know, the "higher-level thingies".

When it comes to analytics, there are too many things that are just too technical for CMOs. Thus the right employees should fill in this gap. CMOs should be consuming the processed and (hopefully?) visualized data and then make decisions."

Julius Fedorovicius, founder of analyticsmania.com.

We’ve already admitted that automated reports alone aren’t enough to solve marketing problems. Data needs to be interpreted to get answers, and the Make Everything OK button hasn’t been invented yet. (We’re working on it!) Thus, the marketing director needs to do these things to keep productivity high:

1. Set up an analytics system that shows risks and growth zones
2. Hire analysts who process data, add context, and provide conclusions

The fact that CMOs are increasing their technology budgets suggests that marketing analytics services are not just an expensive toy anymore. They’re real working tools that help the marketing department execute the plan and get bonuses.

"For smaller companies, the process can be simplified and it could be good enough to start with hiring someone who is good with Google's ecosystem (GA, GTM, GDS, sheets) but ideally the team should consist of at least two people.

1. The technical person who's good at data collection (let's call him/her analytics engineer).
2. Data Analyst who can then clean the data, play around with it and analyze (from what I've seen, usually people analyzing data are not very skilled with the technical part of data collection, that's why analytics engineer will help).

The people you hire could help you choose/build the right analytics stack."

Julius Fedorovicius, founder of analyticsmania.com.

A business can successfully use either out-of-the-box products or its own solutions. To avoid unnecessary money and time expenditures, a company must clearly understand what it needs.

When it comes to developing your own solutions, you should pay attention to the development and support team as well as documentation.

If your specialist who writes data collection scripts and is responsible for configuring the entire system quits, you need to be sure the employee who replaces them will be able to support the projects.

If you want to purchase a ready-made solution, first evaluate the amount of data you need to process. If you have 10 sales a day from online advertising, you don’t need to invest in powerful and expensive services. For this amount of work, it will be enough to use Google Analytics or Excel.

While your turnover and needs are growing, there’s an increasing need for additional metrics, automation, and higher data processing speeds. For example, in the SaaS business (IT solutions with subscription fees), you can begin to receive real value from your own data warehouse or the use of machine learning algorithms if your annual turnover exceeds $200,000.

However, note that even a ready-made service that can collect, process, calculate, predict, and visualize data needs to be adapted for a specific business. Many businesses have unique logic and key planning parameters. Therefore, the structure of the imported data and data visualizations will differ. In addition, from our experience, customers don’t like to learn how to work in new interfaces but want their data presented in a common way.

To summarize, our recipe we draw on at OWOX in creating marketing analytics systems is this:

• Businesses should have full access to their data
• Data must be delivered in an interface that’s familiar to the decision-maker

There are many features and differences among businesses even within one niche. For example, in retail, the business of electronics is conceptually different from the business of selling clothes or household goods. Let’s say there’s a different frequency of purchases and a different emphasis on working with new and current customers.

As another example, many companies are emphasizing the development of mobile applications and client analytics. Such companies have completely different marketing models, methods for plan performance, and metrics compared to companies based around web applications. That’s why it’s difficult to combine and process data.

Therefore, it’s important for a business to have direct access to its marketing and product data and not to rely on the capabilities of a particular service and its visualization system.

Companies are looking for analysts for the needs and pain that have developed in the company at the moment. And if there are a lot of pains (and there always are), then the search for an analyst can take months or even years. And at the end of the day, there’s a high probability that the specialist hired will not be able to handle all the tasks.

It’s important to separate and systematize analysts’ areas of responsibility. Nowadays, many companies are looking for analysts who will set up metrics, merge data, build hypotheses based on this data, provide recommendations, make sure the conversion rate immediately grows, and ensure that advertising campaigns begin to pay off. They’re an analyst; they can handle it!

In real life, it doesn’t work like that. Your analysts will either not have enough time or not have enough skills. For example:

• A specialist can be an excellent technician (setting up data collection, writing requests, and preparing data for analysis) but lack knowledge of how to meaningfully assess the obtained data.
• A company may be large, and everything may be organized so that only one analyst can deliver data. All day long, that analyst will essentially be a voice interface to the data.

What you should do is determine which areas of responsibility analytics can strengthen by measuring data and points of growth. For example, in the retail business, this is marketing, product, and the customer experience.

If specialists are overloaded with preparing data and reports, it’s worth allocating this technical analytics function to a separate unit or transferring it to a partner.

It’s more profitable for a company when analytics is a working function and isn’t limited by the capabilities of one specialist. Otherwise, the company will hit an iron ceiling.

For example, let’s say an analyst suddenly decides to abandon certain third-party solutions and recreate them himself after studying and becoming interested in the R language. The thinking is that the business will stop overpaying for external solutions. But in fact, the analyst is developing their own interest instead of being engaged in their direct tasks.

The company, in return, will receive an unstable solution without support if the analyst quits. The benefits are questionable. Any business shouldn’t be limited to what the analyst can do.

Another thing is when a company forms a full-fledged internal analytics and data department that supports the growth objectives of different departments. In this case, the company can be sure of data quality and the resources that evaluate the development of marketing, product, and other divisions.

The Chief Data Officer (CDO) or Head of Analytics determines which data-based solutions should be developed and supported first.

For example, let’s consider a real case from a client: A bank automated the selection of a client manager based on information collected about the interests and actions of customers. Also, the bank started to consider the conversion of this manager to close transactions with similar customers’ profiles.

In developing a solution for the company, it’s important for analysts to get answers to the following questions:

• What are the strategic goals of the analytics customer?
• What decisions will be made and what responsibilities will be assigned to achieve the strategic goals?
• What questions do customers need to answer to make a decision?
• What graphs, reports, and recommendations can help answer questions?
• Which metrics should be used in the solution?

It’s important to work out the tasks for analysts so as not to miss important details and, as a result, to give the customer a working and useful solution.

Let’s summarize. In order for data to work for the business, your analysts should:

1. know what customer goals and solutions really depend on the data
2. consider how much profit these decisions affect
3. regularly monitor the quality of data and report promptly on deviations from the norm

And of course, it’s in the CDO’s interest to increase the proportion of employees who use data by themselves rather than needing the constant assistance of a team of analysts.

"I think that analytics in marketing should be considered as supplements for an already experienced/strong athlete. They can make him/her faster/stronger/etc. If you are already doing pretty well, then analytics can help you improve that.

Speaking of more concrete examples (regarding the two zones mentioned before), here are some.

Growth zones:

• Finding 20% of products that bring 80% of revenue. Check their traffic sources, demographics, etc. and try to identify which segments are working better than others. This might help you identify opportunities where to increase budgets (for example to make adjustments to your paid advertising).

Risk zones:

• Tracking of errors (and other negative actions, checkout errors, signup errors, etc.). These are the primary struggles of customers/clients. By identifying them and fixing you can reduce friction thus increase the conversion volume."

Julius Fedorovicius, founder of analyticsmania.com.

It can take six months to close a vacancy for a strong analyst. That’s a long time. A partner is a great solution to get started right away and not wait so long to get value. There are two cases when working with a partner really helps to develop analytics in a company.

1. The partner has significantly stronger expertise than the company currently has. For example, say a business already wants to work with data or at least start collecting it, and there are no analysts in the company or they’re focused on other tasks. In this case, the company should find a partner who already has experience working with similar cases in the niche.
2. The partner’s team is hired for a certain function. For example, it could be implementing analytics, providing Google Tag Manager support, or reporting on marketing. This approach works when companies need a large and qualified team to perform specific tasks and in-house employees are more valuable when focused on data analysis and interpretation.

Important! With any approach, it’s worth building relationships with a partner who’s transparent in their decision-making. In other words, your partner must understand your company’s strategic goals and objectives and your analytics requirements in order to work proactively.

What are your strategic goals, time frame, and budget? For example, at the annual Go Analytics! conference, people regularly speak about how to use paid analytics products worth hundreds of thousands of dollars. But there are also speakers who say that analytics can be built for $130 a month. Both cases are true and real.

Big companies that actively use Google Marketing Platform products and combine advertising data with sales data use Google Analytics 360. For other companies with smaller volumes, it’s enough to use features of the free Google Analytics.

Simply put, if your annual revenue is between $200,000 and $1 million, it’s time to collect data in your database but it’s too early to invest efforts in working with machine learning algorithms. You won’t have enough information about your customers yet.

Focus on analytics functionality and profitability for your business. First of all, analytics should help answer the questions Where are my risks? and Where are my growth zones? when implementing your plan. If it doesn’t answer these questions, then it’s just taking away your time and budget.

"Analytics is something that connects the past, present, and the future. We analyze the data created and collected in the past. In the present, we try to predict what would happen in the future. And finally, we try to influence the future by reallocating our efforts in a new way."

Mikko Piippo, digital analytics & optimization consultant, co-founder, and partner of Hopkins.

• Start building analytics from defining the development level of the business.
• It isn’t necessary to wait for a large amount of data or look for a super analyst for six months. Start by contacting partners who have experience building marketing analytics models in your area.
• Choose the appropriate model: a ready product modified for your business or a system developed from scratch just for you.
• Consider hiring an analyst who’s well versed in the context of your business even if they don’t have the most advanced skills. They’ll be able to catch on the infrastructure from the very beginning and further scale it and improve it.