This is a two-part post with some ideas about the stories that data tells (author) and using data to tell a story (star).
Data as Star
When I was running strategy for iZS, Zurich Scudder’s VC attempt during the first dot-com run, I read an HBR article (the physical, ridiculously priced magazine) talking about companies producing two things; a product and a data stream. Today’s environment has sensitized those of us who are paying attention to the fact that companies know a s*%$ ton about us. So with all this data flying around why are we still married to the old-school, direct data collection. Or limited by the lack thereof. We need to get smarter to solve for two problems:
- What we are asking our measures to measure has changed but the measure has not.
- Measures that don’t exist.
In order to use data as the star of our story, it must exist and it has to be believable (source, methodology, red-face test). Other than that, data in support of a pre-constructed narrative too often a creative endeavor. We use L&D spend benchmarks as a way of showing how our company is leading the pack in our industry. we use the same metric but compare it to the highest performers in the world (where the company is a bit more middle of the pack) when we submit our annual budget. While data authored narratives can also be fraught with biases, assumptions, and intent at least it starts with and is anchored by the data.
Two data scientists delivering a talk to a group of startups in Delaware remind the audience that data is like a bikini. Very intriguing but revealing nothing. Loved those guys.
I have a job, kinda.
I worked with a company that did a large volume of formation documents. These are the documents that need to be completed and filed with states and the Federal government when you start a company. They reached out to me to look for additional revenue streams, partnerships, or any other ideas that would help strengthen their already leading position.
Of the many ideas that my team and the internal team developed the one I loved was a data play. At the time, ADP has just begun releasing their employment data index. A set of publicly announced numbers that suggested the current state of employment. You may ask, “don’t we already have a number from the Bureau of Labor Statistics (BLS) for that?” We do. It sucks. If you go underneath the BLS numbers you realize the number’s number is up. Don’t take my word just google it. One of the many voices you will come across is Dr. Robert Shapiro’s from EconVue,
Dr. Shapira is no slouch having served as Under Secretary of Commerce for Economic Affairs for four years. Here is how he describes the reason why people at the outset of the pandemic were deemed not “available” for work and therefore excluded from the unemployment calculation.
“In fact, millions of Americans were not “available” for work in April because they were caring for children whose schools were closed—and millions of people didn’t look for new jobs because the avalanche of layoffs made a job search pointless. They fit a textbook definition of individuals whom the BLS excludes from the ranks of the labor force, and so do not count as unemployed.”
So tell me again why we don’t need a data point that simply counts the number of paychecks it issues for its clients every two weeks. Seems closer to the source. Still not perfect, but important.
So our data play for the formation king was just as simple as ADP’s. We already knew how many new businesses were being formed. We knew what percentage of the overall formations it represented (more than representative). Hell, we even knew what type of business was being formed (barbershop, restaurant, welding.) So why not create an entrepreneurial index that showed how active the startup spirit was and in which sectors they were seeking their fortunes. This information could be useful for commercial real estate, business banking, tax incentives, and more. Again, imperfect but important. They chose other options but this post is also me pitching the idea to them again (call me).
How will the rapidly growing gig economy make the current measures even less calibrated to reality? We have to look at new and better measures for important things that must be measured properly. If we are to use the unemployment data as a proxy for how well we are doing, as workers and as an economy, then it better be as close to right as possible.
The economy goes as pickup sales go
This proxy is real and has been around for a long time. DataTrek is a believer that, “pickup truck sales can be a helpful indicator since the trucks are used in a vast array of businesses, and are typically a discretionary purchase.” More here. To paraphrase Forum Corporation co-founder, Richard Whitely, quoting some ancient wisdom from someone…
“We cannot see the wind. We know it exists by watching the leaves.” – Unknown
“True dat.” – Me
So the pickup proxy may be a bit of a stretch but it has a history (relatively accurate directionally) and broad nodding acceptance. While these correlations may seem silly finding one that works can provide insight, and even advantage. Say for example you are the first to see the relationships between two data sets, Home Depot revenue and Housing Sales. Whenever housing sales go up revenue goes down implying that people are buying rather than building. Or revenues rises when home sales go down implying more fixing up less moving. Either way (or both) there is an insight into consumer behavior and as long as there is a time lag between dataset results there is time to exploit it.
So in the case of the health of the economy what other leaves could we be watching with our all-access pass to the data-sphere. So how about a healthy neighborhood index? It could consist of:
- Credit card receipts for all the businesses in the ‘hood (local business revenue)
- Utility late payments (resident disposable cash)
- Major crime statistics (gun violence, home invasions, assaults)
- Building permits issued (new development)
- Employment by local businesses (growth, stability)
- Aggregate savings growth
- Property maintenance complaints filed
- Utility events in the last time period (power outage, water quality, etc)
- Solar installations (environmentally consciousness)
None of these ask residents how healthy is your neighborhood. NPS is a great proxy but redemption of referral codes is better. Leaves don’t lie. People can say that being green or putting money away for retirement is very important to them but counting solar panels, recycling bins and 401k balances are the truth.
L&D some days feels paralyzed by not having the direct answer to the question, “did the learning make a difference?” Too many levels removed. Too many variables in motion. So we say we don’t have the data. No data, no credit. It is a binary directional question. Precision is not required just a solid proxy that meets our requirements, especially the red-face one. Look for meaningful correlations. Not the “umbrella sales go up every time it rains,” kind.
Longer sales calls could be an indication of better customer engagement skills. Lower T&E expenses could indicate stronger adoption of Zoom. Increased sticky note purchases could be the basis for a corporate innovation index by including cross-department meetings (pulled from the attendee lists in our work calendars) and Ted.com views from the work internet (IT has this just like they have your entire browser history.)
If you want or need data to be the star of your show, it is out there. Sometimes you just need to not look directly at it. Some people say that selling is not convincing but rather helping a customer do the right thing. Data can do both. It can convince and help. How it is used is up to those of us who use it.
“Use your powers for good, you will.” – Yoda?
Part 2 – What’s My Backstory? [coming soon]