Marketers often find that when their business undergoes fast or significant growth, their marketing analytics setup fails to keep up. After all, as marketing spending increases, so does the cost of a mistake: a 1% inefficiency on a $20m budget with 2x target return on ad spend (ROAS) can cost you $400k incremental revenue.
At some point, relying on slow-loading, error-prone spreadsheets or a single marketing measurement provider may no longer be adequate. The cost of errors is too high.
This article explores the many benefits of investing in a data warehouse to address your most significant marketing challenges, saving you time and unlocking deeper, richer insights that ultimately lead to increased revenue, more customers, and greater frequency, all at a lower cost.
We have found that clients can improve their marketing efficiency by 15-30% as a result of implementing a comprehensive marketing datamart.
This guide follows three steps:
- Start With Why. Understanding what a marketing datamart is capable of and the impact it will bring
- Estimate the Costs. Planning the implementation requirements and understanding the costs
- Build the Business Case. Building a complete business case for investing in a marketing datamart
Step One: Start With Why
A marketing datamart is a focused section of the data warehouse designed to speed up the time it takes analysts to answer marketing-related analytics questions.
Perhaps you want to measure the efficiency of your ad spend on Google last year? Or you might want to get an idea of what’s likely to happen if you cut your Meta budget by 30%?
In essence, it is a centralized single source of truth for clean, accurate marketing data.
Managed correctly, a marketing datamart can form the foundation for all your future marketing analytics and optimization goals - completely removing the need for analysts to start from scratch each time a new question arises.
Before embarking on any marketing analytics project, it’s crucial to consider the desired outcome and what you expect as a result. Ultimately, the goal is to generate more revenue from a larger customer base, with increased frequency at a lower cost.
Marketing datamarts contribute to achieving this by:
- Optimizing return on ad spend by improving marketing measurement capabilities
- Lowering the cost per action by improving ad-platform connections
- Saving team time through automation and by productionizing reporting
We have found that investments in a marketing datamart often result in marketing efficiency improvements - measured by cost per action - ranging from 15-30%.
Nielsen estimates that 25-50% of marketing spend is wasted due to companies not truly knowing marketing return on investment (ROI). Forrester estimates that 15-20% marketing performance improvements are possible by improving measurement systems alone.
Whilst these numbers sound high, a marketing datamart can unlock a range of levers that each contribute significantly to improved marketing performance. These include:
- Sending server-side conversions unlocks improvements to ad optimization, leading to a 10-30% reduction in CPA (depending on industry and geography)
- Having a concrete understanding of user lifetime value and the early signals that help you predict this allows you to move to value-based optimization leading to a 10-15% increase in conversion value for the same spend
- Building a Media Mix Model can fill measurement gaps left by hard-to-measure channels, such as iOS where 75% of users opt out of tracking
- Running incrementality tests to identify where ad spend does not lead to conversions. P&G, Uber, eBay, and Airbnb all saved hundreds of millions of dollars this way
- Building a Message Mix Model allows a more granular understanding of creative, which Nielson estimate accounts for 47% of ad performance
- Improving email relevance through personalization and reporting. One of our clients boosted new customer activation by 13% as a result.
- Fixing errors in spreadsheet reporting led one of our clients to adjust the CPA for search down by 26% leading to more efficient budget allocation decisions
All of these use cases are centered around maximizing the efficiency of marketing through improved utilization of data. Of course, you could tackle each lever individually, but this approach can lead to wasteful duplicative work among engineers, data scientists, and analysts.
Furthermore, purchasing SaaS tools to solve individual issues can lead to data silos and integration challenges. One of the key benefits of a datamart is you are taking full ownership of all your current and historical marketing data - in a single place.
Step Two: Estimating the Costs
Start with the outputs
Start collecting and prioritizing a list of the business questions you want detailed answers for. The goal is to create a list of future dashboards and reports you would like to see. Some of the questions are basic:
- What is our return on ad spend, by source, and by campaign?
- Does paid and organic traffic interact with our site or product differently?
- What is our most efficient channel and campaign?
- Is email marketing an effective marketing channel?
- What is the baseline performance of each channel?
They can also be more complex:
- How incremental is each marketing channel?
- What is the optimal channel mix at the current spending level?
- When increasing the budget by a certain percentage, what is the likely marginal gain?
- What channels see the largest diminishing returns to spending?
- Do my TV ads have lagged effects on when or whether my users buy from me?
With the desired outputs in mind, you can now start to think about the tables and models you need to create these outputs. For reporting, consider the requirements of each dashboard and what data columns you need for the final tables to create your dashboard. For automation, explore the data that is likely to be required in the API call.
The key value of a marketing datamart comes from identifying the building blocks that contribute most to your outputs. For example, it is likely that all marketing datamarts will contain one or more tables on customer profiles, campaign history, conversion data, and email engagement.
Investing time upfront to ensure the accuracy and reliability of these tables provides benefits for future analytics work, as they are often combined and analyzed for insights.
While many tables will be reusable, it is important to note that sophisticated measurement techniques, like incrementality testing and media mix modeling, require specific data requirements and cross-disciplinary expertise in data science, data engineering, and marketing.
Understand what source data you are going to need
So the output tables are now roughly designed. Now you must plan how to get the data into the data warehouse. The data in question is likely to include the following:
- Spend and budget data from digital ad platforms such as Facebook, Google, LinkedIn, Snapchat, TikTok, etc.
- Your behavioral events data captured by Segment, SnowPlow, or mParticle
- Additional customer and usage data from your production database
- CRM / customer success data from systems like Salesforce, HubSpot, and Zendesk
- Email and direct marketing engagement from marketing automation platforms such as HubSpot, Mailchimp, Braze, Iterable, and Klaviyo
- E-commerce and payments data from platforms like Shopify, Stripe, and PayPal
- Data from any other marketing tooling such as multi-touch attribution providers or mobile measurement providers
The challenges here are twofold:
- Managing the loading, cleaning, and preparation of data
- Handling some of the less technically advanced ad platforms and getting spend data into your warehouse via CSV loading
This often demands in-depth technical and marketing knowledge - which requires close alignment with your platform or data engineering team as well as your marketing personnel.
Estimate the costs
Data engineering resources are the largest single cost driver of any marketing datamart. Whether you decide to keep your operation in-house or use an agency, try to put together some time and cost estimates to understand the likely overall buildout cost.
You should also factor in the time pressures on other required stakeholders, such as data science personnel, and your own marketing team. This is crucial, as their close involvement is necessary for guiding the data engineers and addressing any resulting questions.
Step Three: Build the Business Case
To decide whether investing in a marketing datamart is worthwhile, you need to calculate the performance improvement "hurdle rate".
This can be accomplished by dividing the estimated project cost by your marketing budget, then multiplying the result by your expected lifetime value to customer acquisition cost (LTV:CAC) ratio, and finally adjusting it based on your payback period for the project.
For example: if your annual marketing budget is $20 million, your expected LTV:CAC is 2x, the marketing datamart will cost $250,000, and you are looking for payback in a year, then your performance improvement hurdle rate would be 0.6%.
By considering this alongside Step One: Start With Why, you can determine the viability of your mooted investment in a marketing datamart. In the example above, if you believe you will get more than 0.6% performance improvement, then the marketing datamart would be net positive for your business.
Implementing a marketing datamart can lead to improvements in marketing efficiency, ranging from 15-30%, and can help you avoid wasting 25-50% of your marketing spend. With more effective use of data, you can significantly improve your marketing performance.
Following these steps above should help you build a centralized single source of truth for clean, accurate marketing data that can optimize return on ad spend, lower cost per action, and save team time through automation and reporting.