Working with an E-commerce brand which had grown rapidly into new markets. The pace of growth and evolution of the brand had fallen into bad practices with their approach to analytics and measurement. For an E-com they should be data rich but were information poor which had a rippling impact on the wider business outcomes.
There were a few significant moments over a period of 12 months that led them to the questioning what they are getting out of their measurement approach.
- Google’s move away from Universal Analytics (GA3) to GA4 and the end of attribution
- An overspend on paid search meant for a period of 10 days there was no paid search running. The impact on overall sales was minimal
- Took place in one market then the same test was deployed in other markets, and it showed similar results
- Ran an out of home (OOH) campaign (in multiple markets) which was not deemed successful as it was not visible within attribution
As a brand they created their own barriers around measurement trusting Google Analytics and GA360 data driven attribution to be the single source of truth. This approach created challenges:
- Attribution was providing an inaccurate view
- Impacted the wider strategy
What was learnt they were data poor and information poor.
Google Analytics attribution provided an inaccurate view
Google Analytics (GA360) overall set-up and structure of GA3 was satisfactory. Each country had its own view supported by a main view, test view and a raw view. The country view made logical sense; it was driven to maximise Google Analytics capabilities around data driven attribution.
There were some big assumptions made in what GA360 attribution can provide. What was not taken into consideration.
- It will have a distorted view of the journey to sale as it only looks at the last 4 touch points
- October to December is the biggest quarter where on avg 70% of sales are generated
- Does not factor in brand awareness which will be at different levels per market
- The models reward the channels and tactics that drive traffic to site
- It won’t factor in external variables i.e., seasonality, pricing
The process of managing attribution within Google Analytics became very unproductive 70% of time was spent in getting data out of Google Analytics and updating reports. The custom attribution models used for each market / GA view the data was not available within the API to build into the dashboarding process meaning it had to be a manual process. The other challenges it created across mutli-markets:
- With how attribution models work they update daily this required the team to constantly update previous weeks / months reports
- The models were refreshed twice a year it required the team to update the past reports / dashboard / QBR’s
- There was no one source GA or dashboards that could be trusted (last click was disregarded)
- No documentation that clearly stated what attribution models were in play and the rationale behind the changes or previous changes
- Not all models were correctly updated across all the markets GA views
- Custom channel groupings were not updated regularly enough to add in new channel / tactics (source/medium/campaign)
- No documentation to support how channel groupings were defined
The wider strategy
The fixation on attribution had a domino impact on several areas:
Over investing into performance and declining brand interest
There was an over investment into performance putting a focus on efficiency which was driven from the attribution insights and a focus on ROAS. The core media mix was focused on driving short term results: Paid Search, Paid Social and Display (mainly banners). 70% of budget at one point was going into paid search and 95% of total budget going into performance marketing with no real investment into brand marketing. When looking at share of search data it indicates to losing interest in the brand led by the in-balance media mix where baseline sales would have weakened year on year.
Data driven attribution came into play in 2017 which led to a change in strategy and a significant increase in performance and paid search budgets which aligns to the declining interest of the brand. As Byron Sharp says paid search is like shelf space it’s an always on channel that is targeting active customers in the category (5%) not future customers who are not in the category (95%).
ROAS became king and the ultimate success metric for all within the business to measure performance. Individual channels were still performing well when measured against ROAS which is driven by the attribution within each platform (channel) as it benefited by nicking sales from other channels. As paid media budgets grew and became a bigger % of the mix the blended ROAS got much worse.
As the business was scaling it got harder to maintain a ROAS especially when paid is driving a high %.
Under Valuing Organic
Not playing the long game for SEO has been costly, it’s a missed opportunity in what is a very competitive category and search landscape to cut through. SEO plays a critical role in educating the audience in what Google call the messy middle. It’s building the mental availability with the audience that they trust the brand. If organically the brand is not available better known as physical availability within the search results to the customer, then another brand will be available. When the customer decides to make a purchase ideally, it’s driven by organic, this is where Paid Search can ensure the click is made which navigates the customer to the right landing page to convert.
The core foundations
With no real foundations to the Analytics and Measurement it provided an opportunity to start again aligning to the strategy going forward.
KPI’s to track
With the change in strategy it required a more robust set of KPI’s within the business where ROAS had dominated which has its flaws especially when it became the sole metric of success. There was an acceptance that LTV was not used correctly and when it was used to understand performance, marketing was siloed and in turn it became a cost centre.
There needed to be a revamp of the metrics to focus on.
Selecting the right KPI’s came down to:
- Does this metric +/- impact the business
- Does improving the metric help deliver against the business goals
- Does this metric help improve other metrics
The metrics were split into 3 views: Market, E-commerce, and Website. This split helps in ensuring the right KPI’s and the mix of KPI’s are focused on against the right areas.
Market – Being able to understand the dynamic of the market of tracking competitors as well tracking your own brand performance provides a different context to a lot of the internal metrics. Which is why share of search provides the external view with it being a strong indicator on performance and SOS = SOM.
E-commerce – Gaining better insights on E-commerce performance comes from having the right mix of KPI’s. The focus on E-commerce performance is LTV:CAC Ratio but all other KPI’s play a significant role.
What LTV:CAC Ratio provides it’s the connection between marketing and finances from understanding:
- CAC: the cost of acquiring a customer (Marketing)
- LTV: the revenue generated that a customer generates over the lifetime considering the overall business costs (Finance)
The impact of measuring LTV:CAC Ratio has wider implications beyond marketing such as pricing, cashflow and valuation.
The other key KPI is % New Customers, for any E-commerce business driving new customers is critical knowing what is the % at any given period. Which helps to understand loyalty and what the Retention Rate is as there needs to be a balance between new and returning.
Website – Understanding the website funnel metrics provides huge amounts of insights to improving performance with each metric complimenting the other. Add to Cart Rate provides insights into the visitors and their intent are they interested in the product it helps also to understand is the right audience visiting the website. Cart Abandonment Rate is where the biggest improvement can be made which has a direct impact on revenue, high abandonment rate shows there is friction in the checkout experience. Conversion Rate indicates how certain tactics, segments are converting better or worse than others.
Is Google 360 providing the value in return for the price being paid. The simple answer is No. GA3 360 was only being used for the custom attribution and not for any of the other features that Google 360 provides. The plan for the Google 360 was to give it 2 years to re-assess the value that’s being provided around the Google 360. It provides an opportunity for the fee paid for Google 360 can be re-distributed into wider data and tech ecosystem.
When comparing the features between GA4 free v GA4 360 the key features are likely to be:
- Big Query – with the GA4 data collection requirements it will make it challenging to provide the insights within platform with the limitations it has. Which is where Big Query provides huge amount of flexibility it can also be the driver of integrating with other data sources to provide those richer insights.
- Audiences – this is very much dependant on the wider media & data strategy and the use of the Google ecosystem v non Google ecosystem.
With a push towards a new strategy, it required a change in the analytics and measurement approach and embracing a range of analytical techniques. The advantage of these techniques they don’t face the same data challenges as attribution.
The focus on Incrementality and Econometrics to bridge the measurement gap.
Incrementality running control v test activity measures the true uplift which will provide insights in how to reduce media wastage, scale and re-locate budgets. To successfully run incrementality a lot of the insights will be coming from Econometrics.
Econometrics will quantify true impact of each channel on sales, attribute success to those channels and help redistribute budgets by understanding diminishing returns. It will also factor in non-advertising variables such as economy, weather, pricing, promotions, and seasonality. Econometrics bridges the link to Brand Equity in being able to understand the base and the impact of brand awareness.
Brand Equity will quantify the long term impact of advertising on brand equity focusing on understanding the direct and in-direct effects and how they impact the base. Understand which channels help drive brand and which brand health measures are triggered.
The immediate focus is on incrementality which can be deployed via the platforms, and it should be an on-going process. Econometrics requires more planning but there are also other techniques that can be used to help answer different business questions:
- Share of search to help predict market share
- Paid, Earned and Owned to understand the relationship between the three with the goal to save budget on paid media
- Jones understanding the relationship between SOV + SOM with the goal to set budgets on SOV estimation
The decision was that long term managing Econometrics in-house makes sense for where the business is today in the short to medium term utilise the media agency capabilities or find a specialist partner who can help on the journey from:
- Phase 1: fully managed service
- Phase 2: team building + consultancy
- Phase 3: support and knowledge transfer
With a lot of moving parts across the business and time constraints on certain deliverables there needed to be a formalised plan.
The key time constraints that had to be factored in:
- Financial year is July to June
- 360 properties the deadline is June 2024 so less than a year to get GA4 set-up and running
- If insights from Econometrics is needed for 24/25 planning, then the process needs sufficient time minimum 6 months for planning and data collection
When building out the plan there were couple of key areas of conversation that dictated the direction of the plan:
In total the plan had 9 phases over a period of 10 months split into 4 core areas: GA4, Marketing Data Warehouse, Incrementality and Econometrics.
- Phase 1: Planning the analytics and measurement requirements
- Phase 2: Mapping out the Google Analytics (GA4) requirements
- Phase 3: Google Analytics (GA4) deployment
- Phase 4: Mapping out the marketing data warehouse requirements
- Phase 5: Data Warehouse deployment
- Phase 6: Build out a plan for incrementality testing by market
- Phase 7: Econometrics planning
- Phase 8: Roll out plan for incrementality testing
- Phase 9: Roll out Econometrics plan. Start the process for 1st round of Econometrics
Google Analytics 4: Getting the right implementation of GA4 is the most critical in the plan with the GA3 360 deadline for June 2024. Factoring in there will be a code freeze from October to early January. Getting GA4 implemented in good time ahead of the peak season from October to December was a key business requirement.
Marketing Data Warehouse: With being data poor and information poor it required a very different approach in how to use data. There was no frame of reference which made it challenging. With the business requirements changing, an exhaustive data collection building a marketing data warehouse (using Big Query) was the only viable solution. The marketing data warehouse had to be multiple purpose to be able to cut and the slice the data in different ways for stakeholders. Couple of key deadlines being by end August before peak season starts to help build a frame of reference:
- Getting historic channel and campaign data from multiple data sources
- Ensuring KPI’s in the framework are easily accessible and can be segmented out
Incrementality: With incrementality testing likely to start in mid-January there is sufficient time to build a plan using past learnings and insights by channel/tactics/platform for each market. January being a post peak season is a good time to start embedding incrementality testing into all channels and build up throughout the year. For the medium sized markets incrementality will be their main form of measurement (not ready to do Econometrics) for some smaller markets incrementality may not be possible yet due to volumes.
Econometrics: There are many factors when Econometrics goes live dependant on the operational model that is decided for short to long term. The preferred plan was to start the process with a partner by mid Q3 to provide insights for planning next financial year. The other variable with it being a multi-market, the plan for Econometrics is to go live with the 2/3 of the larger markets which take 70% of the overall budget.
The likely scenario as it’s the first steps into Econometrics will be to go live with one market then with 2 or 3 markets at the same time. For the success the business needs to understand the process, insights, optimisations it goes beyond creating nice charts it becomes a core part of the strategy and a constant education for all.
There has been a significant shift from being data poor, information poor to data rich and information rich. It’s only the start, it requires alignment top down to better understand the data supporting the business requirements.
With Econometrics and other analytics techniques on the roadmap central to driving growth and constant improvement in performance. This will be supported by constantly evolving the marketing / data warehouse which will be the cornerstone for measurement and the wider ecosystem. In the medium term the next hurdle will be how to extract most out of the rich first party data available which is likely to be in form of a CDP. What needs to be considered is the kind of CPD that is required:
- An all in one solution CDP i.e., like a segment
- Building a modular CPD i.e., stitching customer data together
Both options are viable solutions as it depends on the wider ecosystem that is in play which comes back to role of the Google stack. Google may be the answer it may not be the answer.
The steps taken so far ensures measurement is central to driving the Martech strategy.