Marketing Analytics has evolved; it is no longer just ‘web analytics.’ When more marketing budget is spent, you need additional insights that Google Analytics or similar tools can’t offer. That doesn’t mean they have no more reason for existence, but they just become part of a bigger marketing analytics stack instead of relying solely on them.
This is a normal transition. When resourcing is scaling through additional budget or headcount, you’re extending the capabilities of a marketing organization. An early-stage startup doesn’t need incrementality testing as everything they do will be incremental. But once you reach a certain threshold in the budget (rough guess $5M yearly), every dollar spent should be optimized to yield higher results.
Transitioning through the maturity stages in Marketing Analytics
I wrote extensively in several blog posts two years ago about setting up our marketing data lake & warehouse (see the last post). The series dived deeper into the tactical execution but, in some areas, lacked context on where it adds value. With learnings since then, it seemed timely to provide an update on where we are today and how we evaluate the future.
In short, RVshare was a bit behind for its age in its marketing analytics setup. Five years ago, we didn’t have an extensive Google Analytics set up, let alone a marketing data lake & warehouse. Yet, once we got the basics up and running, we kept pushing to catch up and give ourselves a lead in the years following. This wasn’t an overnight journey (hence why the last post was over two years ago), but it represents the evolution I see in startups.
The different stages:
- Startup: In the early stages of every company, you’re focused on reaching product market fit. So driving users to your product, validating your hypothesis, and learning from them is the most important goal. After this initial stage, you’re often focused on increasing adoption to ensure you can survive the following company’s stages. The importance of growing channels, conversion rate, and overall marketing mix increases (often scaling from 1 to multiple sources that work).
- Scale Up: You’re a few years into your journey, and especially for higher retention products/services, you need to focus on not just acquiring more users but making sure that they’re economically the right ones. Such metrics like CLTV are more critical as upsell/cross-sell become a play. Improving cohort efficiency is a leading metric for many companies at this stage.
Note: CLTV/CAC ratios are becoming much more relevant in 2023’s economy and likely will pop up as soon as you start allocating spending on marketing channels in the startup stage to increase adoption.
- Mature: Once you’re in a mature company stage, you often have more than the basics set up. Enhancing your customer data and running experimentation (brand lift, geo lift, incrementality testing) becomes a significant objective.
Not every piece of logic applies the same way. But past the product-market-fit stage, you can scale.
From Storage to Insights
The initial phases were about extracting (E) & loading (L) data into our marketing analytics warehouse (Google BigQuery). We initially built extensions to overlap data sources to get additional insights (at scale), like matching organic and paid search keyword-level data. The majority of these use cases were meant to support the following:
- Data Storage; archiving data for marketing platforms requires scale and (cheap) storage.
- Scaling Analysis; past 50 thousand rows it becomes increasingly hard to analyze data sets in an Excel/Google sheet.
- Data Enhancement; combining larger data sets with internal or external data.
- Rinse & Repeat; replicating a process to extract, transform, and combine data.
Once you’re past the startup phase in the evolution of marketing analytics, you have an increased need to transform (T) data. It needs to be cleaner, enhanced, and contextual so it can be leveraged for custom and deep analysis. Besides that, it also needs to power the ability to be segmented effectively so you can start to work towards segmentation for testing (cohorts, messaging, lift, etc.).
Moving toward Customer Centricity
In the initial phases of building out marketing analytics as a function (& capability), you’re focused on setting up the initial vendor(s) for tracking/measurement: Amplitude, Google Analytics 4, Mixpanel, Piwik, Fathom, Snowplow, take your pick. But somehow, including myself, we’ve confused this as the holy grail of marketing analytics (or worse, customer analytics). But ask these questions:
- Can you segment your customers and activate those users via email by just using the vendor? No.
- Can you combine the data for refunds/cancellations with your acquisition costs? Nope.
- Can it tell you, by itself, what the uplift is of a specific campaign or how it impacts CLTV? Nada.
Simply because the answer to all of them is that it requires multiple vendors to be connected and for a human to analyze the data (or at least come to a conclusion and validate it’s correct), while AI will start to play a more prominent role in some stages, it won’t overtake the decision-making process and judgment that comes with that, that humans excel in (and a big reason why managers/executives exist, setting a vision to head a particular direction).
Our Marketing Analytics Stack Today: Focus on Customer Analytics
- Beyond implementing platforms such as Google Ads, Pinterest, and Facebook Ads, focus on vendors that provide direct insight into customer behavior, such as your ESP, call center, and text messages. These are all customer data-focused attributes and vendors you can connect to via a user identifier (ID, email, phone, address). Most have ETL connectors in Stitch Data, Fivetran, Supermetrics, etc.
- Predictions: The team at Wilde.ai helped us predict future purchase behavior to work through customer segments.
- Enhancement: This can mean anything, depending on your own business. In our case, we wanted to gain more demographic insights into the household setup for our users, so obtaining more attributes that we could segment on was useful.
The Evolution of Marketing Analytics
Every tech company’s favorite metric in 2023 is the CAC/LTV ratio. The higher the outcome, the better a company is potentially suited to grow sustainably. It’s easy to calculate and control this metric in the early stages of marketing analytics because your stack and total marketing spend aren’t high enough. However, when your strategy matures, it becomes more complicated, and you have many levers to pull (just in Marketing) to influence this metric. So the capabilities within marketing analytics need to adapt to this new reality. Some angles on marketing analytics that come into play:
- Marketing Mix: measuring all the different setups (creative, ad strategy, frequency cap, etc.) that come into play by measuring your
- Attribution Modeling: what channels make an impact on the purchase journey of the user? This can be done both qualitatively and quantitatively.
- Media Mix Modeling: how is a different mix of media channels impacting the purchase journey, and is this making the mix more efficient (leading to a better CAC/LTV ratio).
- Customer Lifetime Value (CLTV): how are different user segments or attributes impacting the CLTV of your cohorts?
- Incrementality Testing: If you invest more money in channel/geo/cohort X, will it perform better than segment Y? Can additional investments lead to incrementality, or will it just increase CAC?
- Forecasting: There will be an increased need to understand what is to come based on your existing data. It is either for financial planning or to leverage the right channels at the right time.
We’re maturing through some of those stages ourselves; some are further ahead. Maybe we won’t ever get to some of them based on our scale, but having additional insight into some (attribution, CLTV) has given us a lot of insight and has led to significant returns (well worth the investments). How are you evolving?