This post was co-published with ELGL - one of the most innovative organizations for local government professionals.

As much as we hate to admit it, we humans are a fairly predictable bunch. Our habits, routines, and preferences generally remain the same until some external circumstance changes or life event such as moving, getting married, changing jobs, etc. occurs. (If you doubt me, I would highly recommend Charles Dugg’s book “The Power of Habit” to learn why).

This is why in job interviews, interviewers usually ask a few “Tell me about a time when…” questions. They are trying to learn about past behaviors of the interviewee in order to gauge potential for future success. Unfortunately, these questions sometimes fail to accurately predict a new employee’s performance for a couple of reasons:

Asking only 2-3 of these types of questions hit on relatively few data points when compared with the whole of an interviewee’s prior work experience,
The interviewee may not be answering in a completely truthful manner or it may be biased to shine a more positive light on the individual, and
The circumstances which allowed the interviewee to be successful in the past might not be present in the new role/company.

While it may be hard to predict the future behavior of one new employee based on a few interviews, the circumstances change dramatically when we look at customer data. Not only do we usually have years of data to pull from, but assuming it was collected properly, we can also be assured that it’s accurate and unbiased. Additionally, while a small portion of our customer base is always in flux, the way in which the majority of our customers use our services today will be similar to how they will use them next week.

We’ve all heard stories about how big companies have used data to advertise, cater, and sell to their customers. One of my favorite examples comes from a company that I’m a customer of – Netflix. Although I’ve never spoken with, met, or e-mailed a Netflix employee, the company knows me pretty well through our direct connection as I access their content. They know not only what I like to watch, but even when, how, and which movies I finish in one sitting and which ones I turn off halfway through.

Through their data, Netflix also knew that they had a large group of customers that

Watched “The Social Network” from beginning to end,
Liked movies with Kevin Spacey, and
Liked the British version of “House of Cards.”

When talk of an American version of “House of Cards” starring Kevin Spacey and directed by David Fincher (who also directed “The Social Network”) started, Netflix found themselves in a Venn diagram with the new show and a bunch of $$$ in the middle and jumped on the opportunity. Based on the subsequent success of House of Cards and others, Netflix’s business model has expanded to include creating much more new original content, most of which they’ve used data to figure out what their customers wanted even before their customers knew they wanted it. Although there may never be a sure bet in business, data sure is helping create safer ones.

While Netflix and local government operate in two very different spheres, it doesn’t mean that the use of data is relegated to tech companies. When local governments use some of the same methods to predict and provide better services to our customers, we do more than simply provide a new source of entertainment; we have the potential to change lives.

In parks & recreation, it’s common to try to market a new program by reaching out to past customers who have signed up for a similar program. For example, let’s pretend that we’re offering a new one-day class on a Tuesday night costing $40 where participants will have the chance to construct and take home their own terrarium. This class would require advanced registration in order to make sure that enough supplies were on hand. Normally, most agencies would send out a mass email to past participants of nature-based programs and hope for the best. However, most park & recreation departments have much more data available to them that is going unused.

What if instead, they examined their past participant data to find out which customers in the past have registered in advance for one-day nature based classes in the price range of $30-$50 that took place on a weeknight? It would certainly narrow down the marketing list to a smaller number, but this customization may allow the department to send out 75 postcards instead of 1,000 emails and will most likely result in a much higher registration rate. Even better yet, what if they looked at their customer data in advance of creating the program and built a series of classes around the most popular registration method, topic, day of the week, price, etc.?

Other examples include:

Using customer transaction data to optimize in-person customer service hours and/or online chat availability,
Using timeclock data to predict which employees are on track to cross certain thresholds such as overtime or annual limits where pensions and health care would have to be provided (not simply to withhold benefits, but to ensure that they are offered deliberatively and budgeted for),
Analyzing customer demographics to optimize marketing efforts for reaching out to new customers and quickly measuring the return on marketing investments,
Analyzing which hashtags and subjects lead to the most retweets by the most influential followers,
Analyzing daily visits to a swimming pool versus the daily weather to get a better prediction of how many people will come through the door on any given day based on temperature, humidity, precipitation, etc.,
Analyzing accident/incident patterns to steer the audiences and topics of additional staff safety training, and
Using beacons or other smart devices to monitor traffic in a park or public space to optimize security, trash removal, park maintenance, most popular features, etc.

The best part of all of this is that these aren’t hypothetical/several years down the road type of scenarios; many of them are already in place at my own agency. Unfortunately I see many of my peers missing out on these opportunities because of some misunderstandings or incorrect assumptions. If I had to give some quick advice for local government agencies starting out or trying to improve their data efforts, it would be the following:

1) Industry benchmarks may not align with your current agency priorities. In order for industry-wide benchmarks to exist, they have to be generic enough to apply to an entire industry, usually over a long period of time. Benchmarks can serve as a good starting point and inspiration for improvement, and on the whole, they may be helpful as an annual spot-check in how your agency is doing, but they may not drive the specific change that your community is working towards. What you measure matters, and in order to drive continuous improvement, you need access to continuous data focused on the areas of desired change, not a once-a-year comparison report.

2) Data transparency does not equal data analysis. Let me start by saying that I’m all for government transparency efforts. My own agency includes live results of our performance measures on our website for our community. However, putting it out there doesn’t mean that

anyone is looking at it (or that it’s made to be easily found and accessed),
that people looking at it can understand what it is,
that it’s what’s helpful to our community, and
that it shows what’s both good and bad about agency performance, i.e. sharing data in an effort to actually be good, not just look good.

Simply putting the data out there without also analyzing and reviewing it to drive organizational change is missing out on a huge opportunity.

3) If it’s not useful, it doesn’t get used. This may sound so simple, but ultimately, it’s one of the keys to success. Just as reports often are created, presented, and then immediately shelved, they key to getting the most bang for your buck with data efforts is to make sure that it’s easily accessible and helpful to those who need it. Even if your agency has an amazing set of organizational performance measures complete with dashboards, on a day-to-day basis this is only going to drive decision-making at the organizational level. If you want your supervisors and front-line staff to make data-driven decisions, you have to provide them with the specific data useful to them as well.

4) Taking advantage of data does not require an on-site data scientist and expensive tech tools. Although having these things would be nice, they are not required. Unfortunately many agencies assume that it’s all or nothing, instead of making a concerted effort to do what they can, starting with what they have. The cost for storing data continues to decrease as the technology to display and analyze data increases, making dashboards and other data analysis tools surprisingly reasonably priced.

If your agency is working towards becoming more data driven and wants the ability to take full advantage of what’s possible (even if not now, but somewhere down the road), I would strongly encourage you to consider your specific goals in relation to these four points.

For example, if you are setting performance measures, think about how those would be applied, accessed, and used by individual areas/facilities/supervisors, not just how you’ll be reporting them to your elected officials and public. If you’re thinking of purchasing dashboard software or other technology to display and monitor your data efforts, select one that allows you to customize your measures to those that reflect your current agency strategic plans and goals and can adapt to future ones, not just preset industry benchmarks, or is limited to one area such as operational dashboards or financial dashboards.

If you’re interested in learning more about my experience in leading data efforts at my organization, including the tools we use, how we rolled it out, and how we were able to use data to shift our organizational culture, here’s the recording to a webinar that I recently offered in partnership with iDashboards. Also, feel free to check out some of my other blog posts and resources related to the same subject: