Growth Engineering Services

August 18, 2020
Michael Taylor

This post is the second in a series on Growth Engineering:

---

You may have heard the term Growth Engineering, but what do Growth Engineers actually do? This post provides examples of the types of services a Growth Engineer can perform for your business. Though the work Growth Engineers do is fairly non-standardized, there are a set of core tasks and solutions to those problems that come up again and again.

This post is a menu of services that you can expect when working with a Growth Engineer — though of course not every engineer has experience in everything, and many of these problems can be solved to some extent with no-code tools as well. Treat this as a checklist of the various areas where engineering can help your marketing team, and see what's relevant to your specific problems right now.

Data Engineering:

Getting your data in the right place, in the right format.

  • Vendor selection
  • Tag management
  • API connectors
  • Web scraping
  • ETL pipelines
  • Data cleaning
  • Data warehousing
Vendor selection

With 8,000+ marketing tools to choose from, it can be difficult to know what stack you should be implementing for growth. Based on experience with popular vendors, with some technical ability, it’s possible to save considerable time and money by finding exactly the right tool for the job.

Tag management

Every tool you use relies on accurate tracking implementation to work — set your tracking tags up wrong and it’s garbage in, garbage out. Knowing the implementation details of major platforms, ability to translate documentation and experience with writing tracking plans is key.

API Connectors

The majority of popular marketing platforms now offer APIs for marketers to programmatically access their data and make decisions, even automate the execution of campaigns. Best practice with dependencies and ability to understand documentation are both important.

Web scraping

Many websites don’t offer a public API yet have valuable information available publicly. Web scraping is the ability to pull data from websites programmatically and at a scale impossible for any team of people to replicate manually. If it exists online, there’s some way to scrape it.

ETL pipelines

Most modern marketing automation requires pulling data from one source and pushing it into another system, consistently and in the right format. Expertise in Extract, Translate and Load tasks helps you ensure that consistency and experience with popular APIs is essential.

Data cleaning

Wherever you get your marketing data from, it’s unlikely to be in the right format for the analysis or automation task you have planned for it. Data cleaning typically takes up to 50% of the time of a Data Science task, and understanding common methods is important for efficiency.

Data warehousing

You have unparalleled access to information as a marketer, particularly if you’ve mastered pulling data from APIs or web scraping. Once you have it, storing that data in a structured database that the right people or programs can access securely becomes important.

Data Science:

Analyzing your data to find actionable insights.

  • Exploratory data analysis
  • Image classification
  • Natural language processing
  • RFM analysis
  • Customer cluster analysis
  • Cohort analysis
  • Lead scoring
  • Churn risk tagging
  • Attribution modelling
  • Monte-carlo simulation
Exploratory data analysis

You have performance data and are looking for insights to help drive your strategy. The ability to analyze and visualize that data using techniques that scale way past the point where Excel crashes is a foundational skill for any Growth Engineer, and part of almost every project.

Image classification

Machine learning has advanced significantly in recent years, to the level where it’s able to identify objects within an image with good accuracy. This can be used for a variety of use cases, for example advanced ad creative analysis or competitor monitoring of website changes.

Natural language processing

The field of NLP has several tried and tested techniques for modeling text to extract insights. For example pulling out instances of specific keywords or NGrams, calculating the reading level or translating text into a foreign language using machine learning translation algorithms.

RFM analysis

It’s important to segment your customers by purchase behavior to better target your campaigns. The practice of scoring customers based on recency (how long since they bought), frequency (how often do they buy) and monetary value (how much they spent), often proves to be useful.

Customer cluster analysis

Using techniques like K-Means clustering, you can segment your customers based on behavior into data-driven personas. This is useful for analyzing performance, setting your strategy, and also for building retargeting or CRM campaign audiences. 

Cohort analysis

The customers you signed up yesterday have had a very different experience to those who signed up a week ago or last year. To hold environmental factors constant and see the true underlying growth trend in your performance data, it’s essential to master cohort analysis.

Lead scoring

Your marketing campaigns are bringing in leads, but are they all good enough quality? Scoring leads by variables that are predictive of deals that closed, so that you can optimize your marketing campaigns by qualified leads, is essential for any sales-driven organization.

Churn risk tagging

Every customer you lose needs to be replaced by expensive customer acquisition campaigns. By scoring existing customers by behavior that correlates with churn, you can predict which users are at risk of leaving, and run a CRM campaign to intervene before you lose them.

Attribution modelling

It’s important you don’t trust everything your inherently biased analytics vendors and ad platforms are telling you. Building your own attribution model using advanced tag management and econometric techniques is an important way to verify independently what drives sales.

Monte-Carlo simulation

Before you make major decisions, it’s important that you model out your assumptions and make sure the initiative isn’t doomed to failure before it starts. Monte-carlo is a technique for simulating all potential outcomes of a decision, giving you more confidence and accuracy.

Product Development:

Building tools to automate repetitive marketing tasks.

  • Custom reporting
  • Competitor monitoring
  • Campaign builders
  • Ad optimization
  • Anomaly detection
  • Recommendation engines
  • Conversion optimization
  • Technical SEO
  • Engineering as marketing
Custom reporting

Sometimes your analytics tool just doesn’t cut it, and you can’t see the data you need in the format you need to see it. By writing custom scripts to connect APIs and process data in ETL pipelines, it’s possible to build whatever executive dashboard or recurring analysis needed.

Competitor monitoring

With use of web scraping and various monitoring and audit tools’ APIs it’s possible to keep tabs on your competitor’s activities like never before. This can help inform your strategy, explain fluctuations in your performance and even be used to make decisions on pricing or optimization.

Campaign builders

Once marketers get to building hundreds or thousands of campaigns, it becomes impossible for a team of any size or ability to keep up. This is where automated creation of large-scale campaigns from a product feed data can provide a huge competitive advantage.

Ad optimization

Marketers spend hours every day optimizing their bids and budgets, turning off poor performing campaigns and launching new ad creative. Most of this work can be automated, with smart rules based on data analysis, and timely notifications to keep the marketer in the driving seat.

Anomaly detection

Most marketers aren’t able to keep tabs on everything happening across their business and campaigns all the time. Yet getting notified of every fluctuation would be equally distracting. The solution is an anomaly detection system that finds only the most significant changes to alert you.

Recommendation engines

It’s one thing to spot a pattern in the data, but quite another to be able to automatically make a decision based on that insight. Recommendation engines can be used to automate decisions on what products to show, what creative to use and what optimization to make.

Conversion optimization

This skill set is the bedrock of any growth engineer, and a large part of the work available within product-led companies. Optimizing landing pages, improving the new user experience and driving impact on other key pages is one of the most direct ways to improve performance.

Technical SEO

SEO is one of the few channels where, if done well, you can build an evergreen asset that generates new customers over the long term for free. Therefore many growth engineers work on teams that optimize SEO for the core product or freemium products built to drive traffic.

Engineering as marketing

As content marketing becomes saturated and paid ads get more expensive, some companies have found success in offering free widgets or tools as a way to attract potential customers. Done well, these projects can deliver evergreen assets that attract new customers indefinitely.

---

If you enjoyed this post, check out the rest of the series:

September 6, 2022

More to read