Retail Energy Providers and the consumerization of energy

Blog_14

In my last post, I wrote about the business model of utilities and their status as regulated monopolies. Needless to say, I was surprised by this state of affairs, given how much we in the US beat our chests about the power of free markets and capitalism. But, alas, one of the most fundamental services provided in this country is delivered, in many places, by regulated monopolistic entities that have, by virtue of the electricity they supply, incredible power over our quality of life and economic livelihoods. The more I researched and learned, the more it became clear that, in many cases, the incentive structures driving utilities are no longer aligned to produce the outcomes society desperately needs to access clean and reliable power.

Given that misalignment, it would seem like a new model is needed for delivering energy to the masses, and ideally a model that incorporates more market forces and competition. In restructured markets like Texas, New York and others, such entities exist entirely independently of utilities and buy power for and sell power to customers — these businesses are called Retail Energy Providers (REPs) and will, in my opinion, play a growing role in deregulated energy markets.

What is a REP?

At the highest level, a REP is simply a business that purchases wholesale electricity from power producers and then turns around and sells it to consumers in a retail setting. Depending on whether the market is regulated or deregulated, REPs can operate as standalone businesses (in deregulated markets) or as operating units within a utility responsible for power procurement and sale (in regulated markets). A REP acts as the middleman between wholesale power generation and retail power consumption and aggregates the buying power of individual consumers in the market.

Before I dive deeper into the details, it’s important to note there are key activities REPs don’t do. Primarily, REPs don’t build or own any transmission or distribution infrastructure, meaning they don’t own any of the big high voltage lines that move power long distances, nor do they own the distribution infrastructure you see in your neighborhood. In all markets, even the most unregulated markets like Texas, the utility continues to control the distribution infrastructure. REPs can control the power that flows through those lines, but the utility is responsible for maintaining the infrastructure. I will come back to this point in a future piece.

Okay, now onto the fun stuff. With the introduction of competition and the birth of consumer choice in energy markets, suddenly customer service, sales and marketing, and customer acquisition and retention became hugely important to an industry that historically never had to worry about keeping customers happy (since utilities were regulated monopolies, customers had nowhere else to go, even if the service they paid for was abysmal). When consumer choice was introduced, all of a sudden REPs could compete with utilities for the energy spend share of customers’ wallets, and if they did it in a way that provided customers what they wanted with better customer service than the utility provided (read: none), they could start to gain market share and make money.

Generally, the best strategy to win over customers and gain market share is to sell products that people want. Consumers are sensitive to many factors when making a decision to purchase, and as REPs competed against utilities, they offered not just better customer services but also differentiated electricity plans with attributes that consumers cared about. That includes the obvious price factor as well as things like what % of their monthly energy comes from renewable resources, how an electricity plan might factor in the power produced by their rooftop solar panels, and whether or not the REP will offer them cool perks like a smart thermostat or tools to improve energy efficiency.

Depending on what factors a particular customer cares about, the REP has an advantage over the utility by offering a variety of electricity packages at different price points with value props that the customer cares about. In a world in which consumers increasingly expect seamless, tech-enabled buying experiences when making other purchases (see the rise of Shopify, Amazon, and online D2C brands) it’s inevitable that tailoring products and the go-to-market motion in energy markets will start to matter from a customer acquisition and retention standpoint. If REPs offer a clear and compelling purchase experience through online channels, customers may be more inclined to purchase from a REP rather than the bare bones offering of most utilities.

How do REPs make money?

REPs make money on the spread between what they buy power for at the wholesale level and what they sell power for at the retail level. To maximize profit, REPs have essentially two levers to pull: retail load or wholesale supply (retail load is another way of saying customer demand). Recall that supply and demand must be matched in real time on the grid: if there’s too much supply compared to demand, then there will be electricity (and $$) wasted; if there’s not enough supply to meet demand, then the customer won’t have power and the REP will have to make up the difference by buying additional power in the volatile and expensive real-time markets. (Side note, if none of these terms make sense to you, check out this piece I wrote on the basics of the grid and electricity trading.) So, to minimize the chances of these things happening, REPs have to optimize their models of load so they can most efficiently buy supply.

On the supply side, there’s not much a REP can do from a price standpoint. Wholesale power producers bid in their supply at whatever their marginal cost of production is and the REP acts as a price-taker in that situation, since electricity is a commodity and there’s no bargaining power among market actors. Whatever quantity of energy the REP estimates it needs to buy for the next day’s operations, it will go out and buy that quantity at whatever price the market has settled at. As I will touch on in a later section, there are some non-wholesale options for procuring power supply that could help change the quantity of energy a REP would need to buy from a wholesale power producer, but pricing would likely remain the same.

On the demand side, things start to get more interesting. For the REP to have a handle on how much power it needs to buy in the day-ahead market, it needs to have an accurate understanding of what amount of power customers will demand the next day. In other words, the REP needs to model an accurate load curve with relatively small error bars. The more accurate this load curve is, the more efficiently the REP can go procure wholesale power to meet demand, which makes it much less likely to overbuy or fall short and helps avoid having to make up the differences in expensive and volatile real-time markets.

If a REP does all of this well — builds a great load model and efficiently procures the power it needs to satisfy that load — then it makes money off the spread between its energy costs and its energy revenues. If a REP fails to do this well, because maybe the weather changes unexpectedly and the load forecast is no longer correct, or (like in Texas) energy assets that were expected to be online are out of commission, leading to a supply constraint and increased wholesale prices, the REP is forced to scramble to meet demand. Especially in the case of a power shortfall, this means the REP has to wade into real-time energy markets and procure electricity at the spot price, which tends to be highly volatile. And, if a REP has to procure so much supply in the real-time market at sufficiently high prices, it may not have enough working capital to cover all the costs and has to declare bankruptcy and shut down. In the February freeze in Texas, this happened to a number of REPs who could not cover the cost of the (albeit capped) $9,000 / MWh maximum wholesale energy prices.

Data and implications for The Future

Given the importance of accurately predicting load to the profit potential of a REP, how that model is built and the data that flows into it is of paramount importance. Thankfully, the advent of cloud computing and artificial intelligence have greatly facilitated the construction of computational models of complex events, such as a region’s energy demand on a given day in a year. So of course utilities and REPs nationwide are taking advantage of these great advances in computing to deliver better outcomes for customers, right?

Not quite, according to this piece from Utility Dive in late 2019:

“Data management falls into ‘crawl, walk, and run’ categories, and most utilities are crawling in their use of data right now. AI for data management would be ‘running’”,

according to Kevin Walsh, a Transmission & Distribution Principal at OSIsoft, a data management platform for industry and hardware applications. Given the absolute explosion of data available on energy usage from the proliferation of DERs, AMI smart meters, and other connected hardware, it would seem like a great opportunity to improve predictions of customer demand by incorporating that wealth of data into heavy-duty models that can synthesize all of it and make predictions aided by machine learning.

Utilities, which are not known for their nimbleness or innovation, have been slower to experiment with or implement this technology into their forecasting models. And to some extent that makes sense — their incentives are aligned to build more infrastructure because that’s how they make their money, not improve the efficiency with which they serve customers. That differs from a REP, which lives and dies by its ability to predict load and efficiently serve it. As a result, REPs are probably more likely to take risks and incorporate machine learning and other approaches into their modeling in order to gain an edge in the market. My view is that the rising tide of DERs and software connected hardware all but demand that data get swept up and used to feed load models, because not doing so is essentially leaving money on the table in a market with staid utilities and REPs who are too scared to innovate.

In addition to DERs and software connected hardware benefitting the ability to model customer load, REPs can integrate with these devices and enroll them in demand response and, ultimately, treat them as units of energy supply that can dispatch power to the grid during times of peak demand in return for compensation. If a REP can convince customers to allow them to integrate with their devices, several things can occur:

1. First, the REP can enroll those DERs in a demand response program, helping it manage load during times of peak demand or grid stress to prevent shortfalls and expensive real-time market purchases. In return, the customer gets compensated for their load reducing actions in either cash or bill credits.
2. Second, the REP can start to dispatch those DERs to the grid as supply. What does this mean? Take the battery of a Lightning F150, for example. The battery in that car is big enough to power a single home for 3 days during a blackout (source) and is a non-trivial source of energy that could be dispatched to the grid on-demand. Aggregate enough integrations with F150s and other DERs, and all of a sudden the REP is operating a distributed power plant that can supply meaningful quantities of energy to the grid.
3. Third and finally, given a sufficient network density of DERs in a geographic region, if a REP can integrate with and aggregate a critical mass of these nodes it can start to generate meaningful localized network effects that make it challenging for other players to compete with.

The REP of the future will be far more than a buyer and seller of power. The REP of the future will be a highly sophisticated, data savvy business focused on building network effects by aggregating DERs. And while the business model of a REP inherently exposes it to risks ranging from market exposure to regulatory changes to customer attrition, the incorporation of much more data feeding sophisticated models and an explicit strategy to ride the wave of DER adoption feels like a huge opportunity to disrupt the industry. That is super exciting to me, and I eagerly await these disruptions.

With all this talk about buying and selling energy, I would be remiss to ignore the distribution side of this conversation. Utilities are still the only legal builders and owners of distribution infrastructure, which seems odd in a world where private citizens can own their own generating capacity and could in theory sell it to their neighbors or supply a community microgrid. So, to investigate why this is the case, next time I am going to dive into the history of distribution and the hotly-debated topic of franchise rights.

Thanks to Ariel Smith and Michael Lee for fielding my questions.