Working with an E-commerce brand the focus was around measurement how they were using Attribution which led to re-shaping their analytics and measurement approach. What came out of the stakeholder interviews, audits, and conversations was a constant theme around how much value was Google Analytics 360 providing. In the world of GA3 360 the main feature used was data driven attribution which had a rippling impact on the wider business.
The bottleneck were the insights built around data driven attribution, last click was totally disregarded which made it challenging as it meant a lot of the valuable reports and insights were ignored as well as segmentation.
Coming out of the stakeholder interviews there were two distinct camps when trying to understand the value Google Analytics 360 was providing:
- Camp 1: the heavy users of GA + attribution: loved the granularity of data and detailed insights. Frustrated by the process in using Google 360 attribution
- Camp 2: the users who did not touch GA but received the insights coming from GA. Found the data mis-leading, inconsistent, and not easy to understand
The concerns, frustrations from the stakeholder interviews were valid and aligned when it came to auditing GA3:
Google Analytics 3
- Each market had its own view with a defined filter but there was no ‘main’ view that had all market data
- The raw and main view only had data for the last 18 months, one of the market views had data going back all the way to Jan 2017
- Not all market views had the same filters applied
- Channel groupings were not updated regularly to add in new source / medium
- Channel groupings were not consistent in all market views
- Attribution models were not consistent in all market views
- Bounce rate was 10% to 15%
- Over the course of a year there were periods when sales dropped to around 10% of the average while sessions remained steady
- Product Clicks were not tracked
- Product Add To Cart and Product Checkouts tracking was swapped around
- Product List Views were not tracked against all product lists
- In total had 190 event categories
Google Tag Manager
- In total there were 197 tags, 172 triggers and 89 variables
- Prior to GA4 deployment via GTM the last edit was made 2 years ago
- Vendor tags i.e., Criteo were not actively running campaigns for the last 18 months but still had an active tag
- Multiple E-commerce purchase triggers. i.e., Facebook and Google Adwords did not have the same trigger as Google Analytics built on datalayer event and /thank-you URL
The audit, the stakeholder interviews and use of Google Analytics 360 it was clear a total overhaul was required.
With Google Analytics 360 being a significant investment, it was not providing the value in return looking at overall business performance in its totality.
(share of search data)
The plan was to give it 2 years then to re-assess the value that is being provided by Google 360. Not diminishing the value of what the Google ecosystem provides which is substantial but can the investment for Google 360 be re-distributed into the wider data and tech ecosystem to support the business and marketing objectives.
The 2 year plan
With the business and financial year (July to June) which is the same for the Google 360 contract. The 2 year plan would run across 2 financial years.
Key considerations of the 2 year plan:
- Budget was approved earlier in the year to continue with Google 360 for the next 12 months.
- By end of H1 (Dec) budgets for the next financial year 24/25 are submitted but not final budgets. Unlikely to be in a position by end of Dec to decide if Google 360 is worth the investment
- With no frame of reference, a poor set-up, no solid data to understand the impact of GA4 and 3 to 6 months required for setup.
- It needed minimum one year of using GA4 + data collection to understand if Google 360 is worth continuing with the investment beyond year 2
- GA4 had a base implementation (not tracking revenue) it required to start again
- With the core months September to December, it meant time was invested in ensuring an optimum GA3 main all-market view set-up. GA3 still critical to understand performance
- GA3 will provide ‘comparable’ data for year 1 v GA4
- From start of Year 1 – July GA4 would have a minimal viable e-commerce set-up which will be comparable to GA3
- By start of October (3 months into the 2 year plan – if not sooner) there would be a full GA4 set-up, advanced tracking, and data collection. Which will help in providing an indicator on the kind of costs for the core months
- Code freeze from October to early January
- From January (6 months into the 2 year plan) there will be a good 12 to 15 months in using GA4 360 providing good use cases and insights if the investment should continue or not
- If another solution is required there needs to be a period of minimum 3 to 4 months where the new solution is in place and the data can be compared v GA4 data
2 year (25 month) plan:
- Phase 1 – Month 1: GTM clean up + Deploying a base GA4 implementation + Building an optimum GA3 main ‘all-market’ view + GA4 baseline e-commerce set-up
- Phase 2 – Month 1: Building out GA4 tracking and data collection requirements
- Phase 3 – Month 1 to 3: Implementing GA4 tracking and data collection requirements
- Phase 4 – Month 3: Mapping out GA4 dashboard requirements
- Phase 5 – Month 4 to 6: Designing GA4 dashboards
- Phase 6 – Month 7: GA4 training for all users
- Phase 7 – Month 7 to 21: Generating insights / segmentation, audience activation and optimisation
- Phase 8 – Month 14: Catch-up on 2 year plan, insights on year 1, use cases + what solution is best if Google 360 is not needed
- Phase 9 – Month 20 to 21: Decision if Google 360 is worth the investment
- Phase 10 – Month 21 to 22: Deployment of new solution (if needed)
- Phase 11 – Month 23 to 25: Comparing data between new solution and GA4 / Big Query
One of the reasons to re-assess Google 360, what was the value being provided with the significant investment made. The pricing model has changed from GA3 360 v GA4 360.
GA3 360 – was based on a fixed fee model. It was an upfront cost which is easier to plan against.
GA4 360 – based on a usage based model. Makes it challenging to plan what the cost would be over the year and there is always a likelihood the costs could get out of control.
The additional costs for GA4 360:
- More than 25m events per month will incur additional costs
- Any events that are not business critical and not being used will incur additional costs
- Within GA3 360, there was a total of 190 event categories and max 70% were not business critical
- Sub properties (replacing GA3 views) or rollup properties will incur additional costs
- Sending events i.e., into sub properties will incur additional costs
As the volumes increases the cost of GA4 360 flattens out but there are too many unknowns which make it hard to plan against. The 2 year plan provides the opportunity to track the costings closely in that time there will be peaks and dips in the data.
6 months – GA4 Set-Up
Setting up GA4 and the use of its key features need to be focused on what GA4 free provides not what GA4 360 provides. When planning the tracking of events + data collections there needs to be consideration to what is needed for insights via segmentation, activation via audiences and day to day insight and reporting.
The core set-up of GA3 was built around views where each market had its own view. In GA4 there’s no direct concept of views the closest concept of views in GA4 is sub properties but this is only a feature on 360. Sub properties will incur additional costs from the events that are collected. To move into an efficient set-up but provides more flexibility is using Big Query. With no data warehouse solution in place starting off with Big Query is a great first step which is part of Google Cloud, and the options are boundless.
What Big Query can provide is the ability to cut and slice the data as per requirements with more richness than what GA3 views provided. The end users can view their data within a data studio dashboard and will not have to log in to use GA4 and worry about the complexity of it all. It pushes those users to invest time to understanding their data and drive insights from it then fumbling their way to getting data out of the platform. As new dashboards are built, or new markets are launched the roll out should be relatively seamless.
One of the key things to note with using Big Query there will be additional costs when querying the data but that should be manageable. With GA4 360 it can export upto 20bn events daily where for GA4 free it can only export 1m events daily.
With a multi-market site which is growing the 1m daily limit is likely not going to be enough. With Google 360 on board for 2 years it’s not an immediate issue but need to look at what are the solutions available that can provide same kind of data:
- Reflective Data provide the same GA4 data infrastructure so moving from 360 is likely to be seamless
- Supermetrics can provide a high level of aggregated reports
- If more granularity is needed, then using Google Analytics API via Python
- Analytics Canvas provides multiple Google Analytics connectors to pipe the data
- Use the Big Query intraday table which has no limits
With Big Query the way forward, it makes the data collection process easier as there won’t be such a big need to explore beyond what GA4 free provides.
|Feature||GA4 Free||GA4 360|
The other key feature is audiences, with GA4 360 it provides 400 audiences where with GA4 free it’s 100 audiences. Dependant on the requirements 100 audiences should be more than sufficient, with 400 audiences I would imagine high % of them won’t be used + the granularity of those audiences won’t provide any clear results or insights.
Using Big Query also mitigates:
- Data sampling thresholds: GA4 360 provides 1 billion events v GA4 free 10 million events
- With all dashboards, insights and reporting done in Data Studio via Big Query the use of Explorations will not be required
- Data retention: GA4 360 provides 50 months v GA4 free 14 months
- With Big Query collecting data from day 1 the 14 month retention period is not critical
Next 12 to 15 months – Insights and Optimisations
Come January there should be a good 3 to 4 months of data available layered on top of the datalayer and event data to provide rich insights. During the period of set-up there will be a vast number of dashboards created to provide the relevant insights.
The immediate action will be training to get a better understanding of GA4, understand the context of the data within the lens of the business model and the set-up of the dashboards. Then it’s about generating insights, understanding campaign performance, and building out audiences to activate within the Google ecosystem.
Final 6 months left of the 2 year plan
With a solid 12 to 15 months of data collection and the daily use of GA4 there should be a view with use cases if Google 360 is worth the investment or not.
The areas of assessment:
- Big Query: how many events are exported daily? (GA4 360 providing 20bn daily events v GA4 free providing 1m daily events)
- The no of daily / monthly events:
- Across a 12 month period, longer if possible
- A focus on the core months from September to December
- The non-peak months but also any campaign periods
- The no of daily / monthly events:
- Audiences: how many audiences have been created? (GA4 360 provide 400 audiences v GA4 free provide 100 audiences)
- What was the split of audiences created for reporting/insights and activation across the Google ecosystem
- Improvement in performance for the audiences that were used for activation
The other area of assessment would be how have the teams have adopted the shift from heavily using GA3 360 to now using mainly dashboards which is pulling in the data from GA4. Is the time investment helping deliver value through generating insights, optimisations etc.
If the decision is not to invest into Google 360 then a decision would need to be made on what is the best solution to provide similar or like for like data set. To have this workaround solution deployed so there is a minimum of 3 months of comparable data.