The most important question regarding marketing analytics today is not how important analytics is to long-term marketing success. The critical question is how much impact is analytics driving to one’s business. And better yet, how can we, as marketing leaders, make analytics teams more efficient and effective.
Gartner research consistently confirms that analytics will be a long-lasting priority for marketers. Look no further than our 2018 CMO Spend Survey where marketing and customer analytics ranked as the most vital capability to supporting the delivery of marketing strategy over the next 18 months.
Unfortunately, here are the remarks that we, Gartner analysts, often receive from our clients (and these are despite advanced marketing analytics teams having upwards of twenty or more people):
- ‘Our marketing measurement isn’t trustworthy’
- ‘We don’t have the insights to personalise our customers’ experiences’
- ‘Analytics requests take too long to be delivered’
What do all of these have in common? They are symptoms. Symptoms of a common root cause— too few sophisticated analytics skills.
Let’s unpack each of them.
Why might marketing measurement be suspicious?
- Likely because an organisation is using predictive analytics methods that are beyond their skill level (e.g. MMM or algorithmic MTA that often benefit from a statistics expert). Or…
- Because they aren’t using standard statistical methods that bring more confidence in measurement (e.g. reducing sampling bias or measuring statistical significance…ones that a statistician can easily implement).
Why might personalisation be unfulfilled?
- Likely because their data is siloed and they can’t easily aggregate it (i.e. they need data engineers). Or…
- Because they don’t have skills to make sense of their data or use predictive models like next-best-action (i.e. they would benefit from statisticians).
Why might analytics requests take too long?
- Likely because analysts spend time re-aggregating and re-cleansing data (over and over again) to inform analytics insights (i.e. something data engineers can automate). Or…
- Because analysts lack knowledge of what interactions or profile attributes are predictive of conversion (i.e. something a statistician can model or use feature engineering to help determine).
So the simple question is how do you convince leaders to let you hire more sophisticated skills (or instead cover the cost of training existing teammates to become more sophisticated)?
- Focus on the root cause of your stakeholders’ frustrations (e.g. the constant bottlenecks in insight delivery that result from inefficient data processes that a data engineer can automate; or the inaccurate campaign measurements that a statistician can bring rigor to). Estimate the unnecessary cost or missed revenue from them.
- Convince your stakeholders that you can solve their frustrations through the addition (or training) of data engineers and statisticians (who can de-silo, automate, make sense of, and model your data).
- Collaborate in creating a business case that conveys the incremental value (increased sales, profits, conversions) that these roles could immediately contribute to.
- Convince your stakeholders that you will deploy these new resources against their objectives if they vouch for your business case as being necessary.
If the above comes across as simple, that’s because it is. The good news is it’s the strategy I used at my previous employer, a Fortune 50 retailer, and it worked multiple times. The trick behind your business case becoming approved is that it’s not just you, the analytics leader, requesting for more headcount. It’s your business stakeholders vouching that this additional headcount will address their underlying issue— the root cause of their frustrations.