Scaling Analytics

Does it scale?

That’s the number 1 question for analytics leadership in 2014.

Three Trends Against Our Favour

Devices are proliferating and their categorization is blurring.

The neat division between desktop and cellphone has blurred into a continuum of browser based experiences and native experiences. This results in an acceleration of new interactions and instrumentation challenges.

There are more people entering digital than are effectively trained in digital.

Traditional areas of the economy are dying and people are following some of that money into digital. Normally this would be in our favor, but far more people are entering digital that have been, or can be, trained up in a reasonable period of time. It has been September for a long time.

The increase in popularity and adoption of true agile methods to software management.

There are more continuous improvement shops. On the dark side, instrumentation does break more frequently. On the bright side, instrumentation can be updated more frequently.

Three Trends In Our Favour

Digital spending is increasing.

A lot more money is flowing into digital. Not all of it is going towards headcount. A lot of it going towards software and systems.

The aggregate number of digital analytics professional and data scientists is increasing.

The absolute number of people who work with digital data is increasing.

More analytical infrastructure is open sourced.

More technology is produced by more people, in a more reliable and scalable way.

Scaling

Scaling Testing

It is impossible to fire thousands or millions of media components by hand. There’s a huge opportunity to pro-actively test instrumentation, on an ongoing basis, and iteratively improve tagging. In other words, use agile methods and some open source technology to tackle the challenge head on.

Scaling Education

It is impossible to train a user base of hundreds by spending time with each one, one on one, repeating the interpretation of visit. Different people learn differently. Some learn by doing. Some learn by reading. Some learn by watching and listening. To scale, we’re going to have go beyond the written word.

Scaling Improvement

It is impossible to argue that analytics is delivering better if there is no evidence that analytics is fundamentally delivering better. Executing a single A/B test every other quarter doesn’t scale. Executing dashboards by hand at some cadence doesn’t scale. The way that improvement itself is scaled, and the tempo of change increases to a new normal, commands adoption of technologies that scale.

Conclusion

There is more technology and people coming at analytics leaders than there is technology and people to thrive. Manual processes, including ‘pizza and spreadsheet’ and ‘hackethons’ may have sort of worked a decade ago. It’s clear that they’re not up to this decades challenges.

The trends in our favour far outweigh the headwaters against us. It’s possible to make huge impacts, if the system is architected to scale.