November 05, 2015

Data is a Girl’s Best Friend

Data Girl Drawing

“A hunch after lunch might be quite consequential, but data is a girl’s best friend.”

While Marilyn Monroe was focused on diamonds sixty years ago, the most valuable asset for today’s utilities and the partners who work with them is data. Looking at aggregate production numbers across an entire customer base does not provide the insight necessary to predict and shape future customer behavior. The ability to capture data at the individual customer level, and use that data to inform strategic and tactical plans, is fundamental in today’s dynamic utility environment.

We recently noted that many utilities WaterSmart serves have varying watering restrictions in place. Some have ordinances for two-days-per-week watering; others for three-days-per-week watering, and others had time of day restrictions but no limit on the number of total days. Fire safety, plant health, revenue and water supply conditions all played a role in these differences.

As we began comparing data from our partners, Board Chair Felicia Marcus and her hardworking peers at the California State Water Resources Control Board issued regulations requiring each water provider in the state to report on the number of days they allowed watering (and to institute two-day-per-week per week watering if they had no other limitations already in place). Now there would be comprehensive data, easily available.

We reviewed the data from each of the ten hydrologic regions in California to examine if the difference in the number of watering days allowed correlates with the R-GPCD (Residential Gallons per Capita per Day) and the ability to meet conservation goals.

For the first region we investigated, the California Central Coast, a higher number of watering days was not strongly correlated with an increase in R-GPCD. How about a region comprised of more utilities, such as the South Coast region? Again, the data, as shown below, indicated no strong correlation.

Chart 1 Central Coast

Figure 1: Watering Days versus Per Capita Consumption – Central Coast Region.

Chart 2 South Coast

Figure 2: Watering Days versus Per Capita Consumption – South Coast Region.

We then turned to look at some of the interesting data that was being surfaced at WaterSmart. To help utilities focus their attention on areas where staff could see the highest returns on their investment, we developed a set of consumption analytics specifically focused on drought reductions. One of the interesting charts we present is a History Chart, as shown below. This shows the trend over time in customer conservation, as compared to the utility’s state-imposed reduction goal (8-36% as compared to the baseline year of 2013). We mapped, over time, the percent of accounts in each of four groups.

  1. Accounts which already are efficient – below a baseline R-GPCD level, as defined by the utility;
  2. Accounts which have reduced their water use by the percent required by the state;
  3. Accounts which have reduced, but not by as high a percent as required;
  4. Accounts which have not reduced their water use.

This account-level information adds another dimension to the utility-level data collected by the State Board. It enables utilities and their partners to delve deeper into what is going on with their customers and develop appropriate strategies for different usage segments.

In August of 2015 water consumption reductions ranged from 5% to 17% and were highly variable across each utility. The distribution of households that had met their reduction goals varied even more. The proportional size of this group varies from 33% at one utility, to 57% at another. For a utility that is trying to achieve specific reduction goals, knowing what these usage curves look like is imperative.

Figures 3-5: Sample Data – WaterSmart utility partners’ Drought Reduction Data

Chart 3 Drought Reduction Sample Data

Chart 4 Drought Reduction Sample Data

Chart 5 Drought Reduction Sample Data

While the difference in these usage patterns across utilities was informative, the data prompted further questions. For utilities with interval (AMI) data, are those accounts that have not reduced their usage the same ones which are irrigating more than the number of watering days in the ordinance? Although our analysis is not yet complete, early research indicates a positive correlation.  The answer will help guide utilities, with or without AMI data, to a desired course of action. WaterSmart can provide personalized messaging to this group of accounts through our Water Reports, drawing attention to the impact and consequences of these irrigation practices and motivating change.

Another customer attribute that is important for utilities to understand relates to reductions in use. Of all the accounts that have not reduced their use, how many habitually have not reduced? Is this percentage constant across utilities?

Again, the presence of data allows us to answer one question and then explore the impact of that answer. The answers again have wide variability across utilities. In one utility, over 80% of households are non-compliant with their reduction goal at least half of all periods: There is a habitual lack of attempts at reduction, indicating that existing messaging is not having an impact. In another utility, only 10% of accounts repeatedly miss their goals: This indicates a more episodic type of high water use, perhaps due to less controllable events, like an irrigation valve leak. Utility intervention will be different in each of these circumstances.


Ultimately, the ability for state regulators and individual utilities to effectively address water stress through demand management programs necessitates a nuanced understanding of customer behavior patterns. Acquiring this understanding requires access to quality and timely data at the individual household level. Just like diamonds, data can be mined and will provide prosperity and gems (of information). Data represents the operating capital of the modern utility. We look forward to working with many more utilities, regulators, and communities to help them gather and interpret their information to help ensure reliable and resilient water systems long into the future.

What other questions would you like to explore? Which data insights are still “a diamond in the rough,” and need further polishing to be helpful? Which will be most valuable to answer? Please share your comments!

Leave a Reply