To determine which sweaters and slacks to stock, Spanish retailer Zara doesn't guess. Nor does it ask the editors of Vogue or sleuth the streets of Paris and New York. It lets its customers decide.

Zara lines its store shelves with garments in different styles and colors – say, some red pants and some blue, some with ankle zippers and some without. Then it monitors purchases and comments. If red pants with zippers outsell blue ones without, Zara stocks more of the former and less of the latter.

Zara is a leader, among retailers, in using customer feedback and data analytics – it has even streamlined its supply chain so that it can respond quickly to the latest data. Most companies aren't so advanced, but many of them work hard on market research using techniques such as focus groups, pilot-market tests and surveys.

In another part of the economy, however, billion-dollar investments are often sent to market with nary a focus group nor survey. That's the U.S. public sector at all levels – county, state and federal.

Take energy-efficiency programs. The federal government funds efficiency retrofits to low-income homes and federal office buildings. Many state governments, including California, offer low-interest-rate loans and rebates for energy-efficiency upgrades. And utilities, at regulators' behest, spend billions of dollars offering customers rebates on energy-efficient gear such as better hot-water heaters and furnaces.

We know very little about how well these programs really work. We don't know how many more customers would buy efficient clothes dryers if they received rebates of $60, instead of $50, or whether rebates help much at all. Likewise, we don't know how much energy is saved with home efficiency upgrades such as programmable thermostats and better windows.

Engineers make predictions, but customers don't always behave as expected, and contractors don't always install upgrades as an engineer might suggest. Some people never get around to programming their thermostats, and contractors sometimes forget to caulk around electric outlets. Utilities file reports about their programs with regulators, but do not generally use state-of-the-art data analysis tools.

Californians recently closed a tax loophole with Proposition 39, and it appears that a big share of the increased state revenue from the change will go to improving energy efficiency at K-12 schools. This makes sense – the money will likely help schools cut their utility bills, and the lower energy demand should mean less pollution.

Rigorous evaluation could tell Californians, for sure, which of these investments work well and which don't. Early indications are that the program will focus on schools that consume more energy than average. But maybe the retrofits would be more valuable at older schools: A school in the Sierra necessarily spends more on heat than one on the Central Coast, but if it's new, it might already be well insulated. In any analysis, it's important to control for these kinds of confounding influences.

That's why companies in the private sector are increasingly turning to controlled experiments to be confident in their evaluations. These are sometimes called randomized control trials or A/B tests, and drug companies call them clinical trials. The manager, for example, who proposed that Amazon suggest related products at checkout was initially rebuffed by a marketing executive. So he developed a test, diverting a random subset of customers to a page that suggested additional products. He showed that customers who saw suggestions spent significantly more, and Amazon adopted his idea.

Online retailers aren't the only experimenters. Several developing countries, from Mexico to Rwanda, have tested government programs with experiments. In Mexico, a program called Oportunidades has succeeded at reducing poverty, and the initial beneficiaries were randomly selected from eligible households. Hundreds of studies have analyzed the program, and officials continue to tweak its details based on the results.

California lawmakers should do Zara- and Amazon-style evaluation with Proposition 39-funded programs to help the state pinpoint which ones produce the most benefits. The Alfred P. Sloan Foundation recently funded a center, operated jointly at Berkeley and MIT, to help companies and governments bring just that sort of rigorous assessment and feedback to energy efficiency programs.

Californians have long been the country's most forward-thinking folks. We've led the country on energy efficiency and environmental protection. We now have the opportunity to be a leader again – and improve our government in the process.

© Copyright The Sacramento Bee. All rights reserved.

• Read more articles by Catherine Wolfram

Order Reprint