What is a last mile problem?
In transportation, there’s a challenge of scale vs specificity. Imagine a highly systematized network of trains that can get you anywhere in a 100 mile radius. But each station is 5 miles from any other station.
This network would be incredible for moving large numbers of people the 90 miles it takes them to commute into a central city. But what about the handful of miles from the station to the office?
Some say the answer is bike shares, some say more localized train networks, some say the inconvenience of those miles is enough to drive commuters back into their cars.
Whatever your feelings, it’s not the sort of problem that has a clear solution - it’s a problem that has haunted transportation engineers for decades.
Data product can be viewed the same way: while getting data into a state where it’s even usable by the majority of stakeholders is a massive feat, that challenge of getting folks exactly where they’re going remains.
The Last Mile Problem Isn’t New - but it is changing
If you’ve been in data for a while, the last mile problem may sound familiar to you. As early as 2017, folks at Bain were coining the “last mile of analytics” as the major problem pressing the data organizations they work with. As Chris Brahm of Bain put it,
“Seventy percent of enterprises view advanced analytics as a critical strategic priority, but only 10% actually believe they’re achieving anywhere near the full potential value of those analytics. What’s the source of that value realization gap? Well, we believe one of the biggest sources that we see clients struggling with is the last mile. That is the gap between great analytic output and actual changed behavior that creates value in the enterprise — whether it’s a frontline worker, a manager, or even a machine.”
In the age of analytics, this was the breakdown between seeing and doing - the difference between having a dashboard and knowing what to do with those insights. This was a massive adoption challenge - how could companies get people to implement critical thinking as a process?
But the world is changing - with the onset of AI and more advanced SaaS apps, we’re automating away the dashboard and moving straight to action. DSPs are auto-optimizing, CRMs are auto-lead scoring, and CDPs are auto-recommending the next step in the customer journey. We’re overwhelmingly putting the burden of acting on insights into the hands of algorithms, or distilling those insights straight into next steps recommendations.
Where workflows are king, the last mile problem is about customization, not consideration
In short, we live in the age of activation workflows now - everything is a domino in a greater scheme of triggers and automations. Data is flowing through pipelines into platforms that segment and decision, that automatically insert customers into campaigns.
Engineering that machine is a massive and complex responsibility. AI agents require quality data that is structured to be interpreted. Automations run on data that is meaningfully connected.
So the challenge today is less about getting people to adopt critical thinking cultures. Instead, the last mile is about ensuring that data can be easily processed into a format, location and level of detail that is usable for any AI agent, person, or automation platform.
Solving the modern last mile challenge
Today, data teams need tools to perform fast-paced, ad hoc data work. To do this, they need to:
- Use AI purposefully: There's a time and a place for AI, and the teams we've seen succeed with it have overwhelmingly used it upstream to perform tasks that would be exceedingly difficult with SQL. This means finding an AI tool that allows you to prompt directly against your data. We like to use prompts that pass json-formatted events to prep our data for Salesforce.
- Bypass data cleaning: Low quality data never leads to good outcomes. But finding the best datasets and ensuring that they are clean, deduped, formatted correctly and comprehensible to a destination platform can take hours. To get around this, we like to use liquid template to bypass any regex work required for case sensitivity and format agreement.
- Keep data unified: We do all of our last mile transform work in an environment that syncs directly back to our warehouse to avoid creating any fractures in our source of truth.
Ready to get started with the last mile transform? Check out our tools for transforming data in the warehouse.