Four Principles for SaaS Pricing, Part II

The Endowment Effect, a behavioral economics term coined by the University of Chicago economist Richard Thaler, contends that people attribute greater value to a good that they already own or have experienced. In 1990, economists Kahneman, Knetsch and Thaler "found that randomly assigned owners of a mug required significantly more money to part with their possession (around $7) than randomly assigned buyers were willing to pay to acquire it (around $3)." This notion guides the pricing strategy of many start-ups: founders believe that once customers experience or own the product, they will ascribe a greater value than if they know about the product but have not yet experienced it.

While the endowment effect is real and observable, it can lead to a lot of mispricings in the market if used insouciantly. The A16Z podcast on Pricing Free elucidates the dangers of pricing using a "bottoms-up" strategy where companies use a low or "freemium" pricing model, thinking that it would obviate the need for a sales team. "Companies that take that approach can become fairly dominant on the bottom-end of the market, but you may have another competitor who comes in with a sales team, and maybe even an inferior product but a couple additional layers of functionality" which will lead to a price war in an already low-price environment, according to Mark Cranney in this episode. So, the moral of the story is, don't price yourself at the bottom of the market just to get customers to try your product because you might get stuck there.

This leads me to the crux of my post, which is delving into the next two of the Four Principles of SaaS pricing (the first two are covered here).

To review, the Four Principles are to price so that you:  

1. Cover Your Costs
2. Increase Your Valuation 
3. Grow Customers 
4. Signal Appropriately

3: Grow Customers: I want to immediately point out that this tenet could be in conflict with #1, cover your costs, and I talk about covering Customer Acquisition Cost in my previous post. Expanding your business, however, requires expanding your customer base, and it is important to acquire those customers at the right price. And while it isn't the primary impetus for my post, I do briefly want to touch on pricing strategies as they relate to pricing "correctly".

Using the Endowment Effect theory, you'd want your potential customers to experience your product so that they can get attached to it, perhaps even to the point of feeling Loss Aversion at the prospect of not subscribing to it. There are many pricing strategies that achieve this by allowing customers to try out the product in some way, and I won't rehash them here. Instead, I'll direct you to this seminal white paper from Bessemer Venture Partners called "Software as a Service Pricing Strategies." Some of the strategies recommended here are:

• Capacity-based freemium models such as those offered by Dropbox 
• Feature-based freemium model such as that offered by Skype 
• Time-based freemium models, meaning that the company gives potential customers a free trial
• Use-case freemium models such as that offered by Adobe to just the readers 

I like time-based freemium models (with the caveat that it really does depend on your business model) because it allows you to achieve the Endowment Effect with the customers and even lets them feel a little Loss Aversion in case they don't sign up. Once they do sign up, however, you need to make sure that you are pricing appropriately to the rest of the market, something I'll discuss more in the Signal Appropriately section.

For more on pricing strategies, be sure to read Tunguz's posts on How to Price Your Product and The Three Part Tariff.

Using Opportunity Costs in CAC:

One twist to calculating CAC that I recommend is that while traditionally you would use just the paid sales and marketing costs to calculate customer acquisition costs, I would consider forgone income, or opportunity costs as customer acquisition costs as well. So if you would normally charge $100 a month for a product and you give away 2 months of it for a free trial, include those two months of foregone revenue ($200) in your fully-loaded CAC calculation. Obviously, your revenue payback period will be longer when you include the opportunity costs into the calculation. Maybe it'll take you 12-15 months to recover the CAC instead of the benchmark 11 months noted by Tunguz. But I think that this is a good metric to measure so that you don't become cavalier about handing things out for free. It will help you quantify the costs of letting your customers try out a portion of the software, and it will be important to take a pulse of where that metric stands over time to make sure you're headed in the right direction.

4: Signal Appropriately:

In the late 1960's George Akerlof wrote about asymmetric information, an idea that later won him the Nobel prize, in a paper called "The Market for Lemons". In a used car market, for example, buyers may be unable to tell the difference between a good car, worth $1,000, or a "lemon", worth $500.  To account for the possibility of a lemon, buyers cut their prices. Astute buyers may offer $500 to start and sellers, knowing the quality of the car, may or may not accept the offer. And so, Signaling Theory was born, asserting that the quality of an item is endogenous, and that the price of a product signals the quality of it. Hence, it is imperative that the product be priced appropriately. Here, I will elaborate on three types of signals: Market Segmentation Signal, Competitive Landscape Signal, and Customer's Return-on-Investment Signal.

Market Segmentation Signal: 

First, determine to which segment of the market your product will be sold. Who will your customers be? If you want to sell to large enterprises, then don't price your product for small businesses. That's giving the wrong signal, and may actually turn off some enterprises who think your price is indicative of the quality. (There are exceptions to this, of course. If you've figured out a way to bring costs down in a way that others haven't and are passing on those savings to your customers, then by all means, price lower. But you'll have to be very clear in your messaging as to why your quality is better AND your price is lower).

Competitive Landscape Signal: 

Second, price relative to your competitors. Here's where a quantitative model called value-based pricing can prove to be useful. Value based pricing works like this:

1. Determine the cost of your closest competitor


2. Outline the functions that your competitors have that your product does not have. What are those components worth. Try and be objective here, and use market based data where possible. Are there two similar products in the market, one with some functionality missing? What is the delta between the two price points? 


3. List the additional functionality that your product provides, and ascertain the value of that functionality. Again, try to be as objective as possible, and use market data where available. 


4. To arrive at the price of your product, take the price of your closest competitor, subtract what you calculated in #2 and add what you established in #3. 

Especially when you have competitors, value-based pricing presents a great way to calculate the differentiated worth of your product and and set your price accordingly. For more on value-based pricing, check out this Harvard Business Review article. 

Customer's Return-On-Investment Signal: 

Lastly, the price of your product can be a signal how valuable your product will be to your customers. Each customer makes an investment when buying your product, and if you can show him that the return on investment (ROI) will be substantial, it will make his decision to purchase the product a lot simpler. That return can come in many shapes, according to Bessemer, such as increased revenue, time savings, efficiency gains, resource or cost savings, or reduced errors. 

I really like this article by Lincoln Murphy of Sixteen Ventures on pricing. In it, he reveals the 10x rule. "If I sell something for $100, I want to provide at least $1,000 in value to them... at least," he says. 

"You do this by understanding your customer’s Desired Outcome, goals, opportunities, problems, etc. You do this by offering Price Anchors that are not competitive services, but what it would take to replicate this in-house, with low-efficiency, high-cost human beings, what you (or the industry) has paid to create this solution, or the fear of not meeting some level of compliance and the costs associated with that."  

So this may be an exercise in quantifying both the tangible and intangible benefits to arrive at an ROI. When you can quantify the ROI for your customer and compare that against your price, which will be 1/10 or lower than the return, you are signaling your product's value in a very quantitative manner, making it very easy decision to purchase your product. 

The primary take-away from my two pricing posts should be that pricing for SaaS involves both art and science. There isn't one equation for it, and you can't just say, "let's price at where our marginal costs will be." At the same time however, there are plenty of quantitative tools that will help guide and set benchmarks for pricing considerations, and those should be utilized in conjunction with the "art" of knowing your customers, your competitors, and the market.

This post was edited by Andrew Amos.

Four Principles for SaaS Pricing, Part I

"I want you to be very careful, because no single decision will impact your company's valuation more than the one you're about to make on pricing." That's Ben Horowitz as quoted on the A16Z podcast on pricing. As the podcast explains, pricing isn't intuitive, and the partners at funds constantly push their entrepreneurs to raise prices. 

Unfortunately, higher prices don't automatically lead to higher valuations. There will always be a price/quantity struggle especially as you start having competitors, as I explained in my previous post on auctions. And if you don't get enough people to buy at the higher prices, your revenue, and hence, your valuation suffers. 

In that vein, I've designed a framework for deciding on how much to charge for your SaaS product. This framework could work for any product, but the first tenet deals specifically with subscription services. Another caveat: this framework applies well to companies operating in post-chasm industries. That is, once the early adopters have embraced a new technology and competitors have entered the market, this framework can be used to set pricing in relation to your own costs and prices of your competitors.

The Four Principles are to price so that you:

1. Cover Your Costs
2. Increase Your Valuation 
3. Grow Customers 
4. Signal Appropriately 

In Part I of this blog post, I'll cover principles 1 and 2, Covering Costs and Increasing Valuation.

1. Cover Your Costs:

Sure, this sounds backwards, but stay with me. Most start-ups worry about how much they can burn given how much cash they have. Covering your costs, however, proves to be an interesting thought exercise, especially when you look out into the future. And the exercise can yield an excellent benchmark. In two years, when you've acquired the number of customers you've predicted, what will your costs be? How should you price to cover those costs? Let's dive in a little deeper.

When the early majority begins adopting a certain technology and slew of companies begin offering solutions to a specific problem (even if the solutions are differentiated), the market begins to resemble monopolistic competition. I wrote about monopolistic competition in my post on newspapers. With low barriers to entry in starting a software company and product differentiation, each technology can set its own prices (as in, they aren't price-takers), but the industry provides a guideline. 

Economic theory in relation to monopolistic competition would tell us that firms need to price so that average revenue is greater than both their average total costs and marginal costs in the short-run to make an economic profit. In the long run, average revenue will equal average total costs. Both graphs are shown below. 

But when speaking of revenue covering costs, pricing subscription products gets tricky. For the most part, costs related to selling the product will exceed the first month's revenue if you're expecting the lifetime of the customer to be several months to several years. So here's how I would think about covering costs. 

COGS: The first thing I would say is that your price every month NEEDS to cover costs of goods sold (COGS). Those are direct  incremental costs that are incurred every time the technology is sold. Usually, the biggest component of COGS for a SaaS business is hosting costs, but can also include set-up fees or platform fees. So, never price so low that the respective monthly hosting costs aren't met by what the customer is paying you every month. I've seen too many companies fall flat because they have negative gross margins; it's hard to price yourself out of that kind of a hole. 

CAC: When considering how much to price in relation to your customer acquisition costs (CAC), which is more of an indication of your sales and marketing costs, a payback period is a good way to think about it. Tomasz Tunguz at Redpoint points out that a median start-up has a payback period of 15 months on a gross margin basis. That is, it takes 15 months of gross profit dollars from a particular customer to cover the sales and marketing costs of a new customer. In terms of revenue, he notes that CAC is typically 11 months of revenue. I think those are good benchmarks to work against, and as Tunguz points out, the faster your payback period, the more working capital you free up for other initiatives. For more on the right way to calculate CAC (using full cost of customer acquisition), check out Andreesen Horowitz's "16 Startup Metrics". 

Total Costs: Lastly, I don't hear much discussion about total cost of operations in relation to revenue. The question many managers ask is, "How much can I burn given the amount of cash I have raised and have on the balance sheet?". And while that mentality is important to insure against insolvency, that burn number usually doesn't factor into how much to price your product, and I think it should. 

When it comes to covering total costs, it is helpful to project yourself into the future, as I mentioned above, and think about what your costs will be when you reach a certain number of subscribers. For example, if you forecast having 100 subscribers over 2 years, how many sales people will you need to get there? 5? 10? 20? What will you have to pay each one? Will you need a Chief Revenue Officer? What will your marketing division look like? Who else will be on your payroll, and what will your rent be? Build a financial model for just the costs, which will help edify how much revenue you will need to cover those costs or make a profit. Once you have the revenue number, you can divide that by the number of subscriptions you anticipate having (in this case, 100), to get to your price. Of course, you don't have to price exactly there, and pricing may actually differ from customer to customer depending on negotiations and how early you are in the process (your earliest customers will likely get a great deal), but this will give you one benchmark to enlighten your process. 

A quick aside on measuring total costs using financial models: I've had some cognitive dissonance here. My accountant side likes Statement of Cash Flows, my investor side likes EBITDA, and my economist side prefers Net Income (to include non-cash costs). I've reconciled all of those to find that EBITDA minus Capex minus Interest Expense is the best proxy of how the core business is doing. It takes out the noise of non-cash items, such as depreciation, and exogenous cash items, such as capital inflows/outflows for debt or equity. If you want to measure just the expenses, take out the revenue line on the EBITDA, subtract capex and interest expense, and you'll be left with just the costs of running the business. Those represent the total costs you'll need to cover. 

2: Increase Your Valuation:

In the second tenet of my framework, I am deliberately eschewing saying "Increase Your Revenue", knowing full well that revenue serves as a large component of valuation. But as an example, Company A with revenue of $100MM with 80% gross margins is very different from Company B with revenue of $100MM with 50% gross margins. Company A deserves higher valuation. 

A step further, Company C with revenue of $100MM with 80% gross margins and 80% annual retention rate is very different from Company D with revenue of $100MM, 80% gross margins and 50% annual retention rate. Company C deserves higher valuation as it likely has lower sales and marketing costs needed to replenish its churn. 

Bill Gurley wrote an article called "All Revenue is Not Created Equal: The Keys to the 10X Club", and I think it is worth a read in its entirety. He discusses the distinguishing factors between high quality revenue companies and low quality revenue companies. Some of the characteristics that warrant a higher revenue multiple, in his opinion, are:

• Visibility of revenue
• Defensible nature or sustainable competitive advantage
• Customer lock-in or high switching costs for customers
• Gross margins
• Scalability or marginal profitability 
• Organic demand versus high marketing spend

So when it comes to pricing for valuation, you want to ensure that you're not squeezing gross margins in order to obtain higher revenue or market share. That is, don't price so low that you're not making any money. Some startups think that once they dominate the market, they will be able to raise prices. That works in some cases (Opentable for example), but you really have to understand the elasticity of demand of your customers, as well as your competitive landscape.  If, in the process of gaining market share you also build up solid network effects, then you are more likely to succeed in raising prices later.

On the other hand, don't price so high that your customers can't afford the product in the long run, and you're constantly spending more on sales and marketing to replenish your customer base. It's a balance.  

In Part II of this blog post, I'll explore the other two tenets of this framework, Grow Your Customers and Signal Appropriately. As you'll see, growing your customer base can come in conflict with covering your costs. Keeping in perspective (i) maximizing valuation and (ii) signaling appropriately to both your customers and competitors, however, should help define your customer acquisition strategy in a cohesive way.  

Marketplace Transactions: Auction Pricing

Platform's Rake: 

If you're in the process of starting a virtual two-sided marketplace, there's plenty of information out there on how much to charge and who should pay. For example, Bill Gurley's article called "A Rake Too Far" illustrates how much existing platforms skim off the top (termed "rake" to follow casinos' syntax) to formulate their revenue. Marketplace Academy's Juho Makkonen expounds on Gurley's findings with his own data to contend that the average rake among virtual marketplaces is 9.2%. Makkonen then gives guidance on figuring out how much rake you should charge for your own marketplace, using variables such as marginal costs, competition, network effects, and transaction size.  

Secondly, there's also plenty of advice on which side to subsidize. Eisenmann et al demonstrate in their paper "Strategies for Two-Sided Markets" that in a two-sided network, there is typically a "subsidy" side, which is a group of users, who, when attracted in volume, are highly valued by the "money side", the other group. Because the strength of the network hinges on the subsidy side, that side ends up paying little to nothing. Eisenmann et al suggested that factors used to determine which side ought to be subsidized should be figuring out which side is more sensitive to price and quality. The following table, from the paper mentioned herein, depicts different two-sided markets and highlights the subsidy-side of the market with a star. 

Marketplace Transactions:

That said, I'm interested in exploring pricing structures for individual transactions within a marketplace. I don't think there's as much thought given to structures of individual transactions, and for that reason, there may be room for optimization. Structures of individual transactions, as I've observed, fall into the following four categories:

1. The platform sets the transaction price. Ride-sharing apps such as Uber and Lyft, as well as Handy, practice this method of pricing their transactions. I view these platforms as psuedo-resellers: they buy a service/product from provider and sell it at a higher price to users. These platforms facilitate the transaction and the meeting of buyer and seller.

2. The seller of the product or service determines the price. Airbnb, Kayak, and Amazon Marketplace fall into this category. These marketplaces mimic a traditional physical market where sellers sell their goods at their prices, and buyers choose what to buy based on prices and other factors.

3. Auctions establish pricing in this case. Ebay is the prime example of this, but other marketplaces make use of auctions as well, such as Priceline for travel, Beepi for cars,and Ubid for electronics. 

4. No dollars are exchanged in this marketplace. Marketplaces such as Opentable and Yelp have no pricing at the transaction level, but there is still a "money side" who pays the platform fee.

Avoiding Auctions:

If we take a look at Gurley's list of how platforms charge, it becomes apparent that not many marketplaces use auction style pricing for their transactions. There could be a few reasons for that: 

1. Friction caused by timing. Auctions may take a while to conduct. If we have to wait to bid on a a car service before booking it, we may forgo taking that time to negotiate and go with another option that's quicker. Therefore, auction pricing may not be the best idea where time is of the essence. 

2. Platform's marginal costs. The platform wants to make sure that it covers its costs, and the easiest way to ensure that might be by setting the transaction prices and basing its commission off the transaction price. For example, if Uber knows that it's marginal cost for a particular ride is $1, it might charge $10 for the ride and take a 20% commission. That way, the company would make a $1 profit ($2 commission -$1 marginal cost). If there were an auction instead and the ride only went for $5, Uber's commission would only be $1 and profit would be $0. 

3. Consistency. The platform may not want to have dynamic pricing for the sake of offering consistent prices and quality. Handy, for example, may want to make sure that you get a consistent quality of cleaning done for the same price, hence setting the transaction price each time. 

4. Sellers' marginal costs. In this scenario, the platform may permit sellers to elect their own prices to protect the sellers' marginal costs. Etsy serves as a good example here since craftsmen on the site may have vastly different costs, requiring them to charge varying amounts. They may also have varying levels of quality of workmanship, allowing them to charge higher or lower prices than their competitors. The Etsy platform wants to cater to that kind of variability in quality and competition by sanctioning sellers to establish their own pricing. 

Types of Auctions:

Auctions have the benefit of allowing price to be close to where supply meets demand without the platform's intervention (more on that below). There are lots of different types of auctions, and I won't go through all of them, but they are English, Dutch, First Price Sealed, Second Price Sealed, and Reverse Auction. You can learn more about all of them here. The two most relevant ones for our purposes are the Second Price, Sealed Bid Auction (used by Ebay) and the Reverse Auction (used by Priceline).

A Second Price, Sealed Bid Auction is where bidders submit what they are willing to pay. Their bids remain undisclosed to other bidders. The highest bidder wins the auction, but pays the price offered by the second-highest bidder (in Ebay's case, a premium is added to the second-highest bid). Therefore, the buyer still ends up having a consumer surplus. It is important to leave some consumer surplus on the table to keep customers coming back to the platform. The seller can set a minimum starting point to cover his costs, so issue #4 from above is solved.

A Reverse Auction requires that the buyer name his own price and the seller either accepts or denies that price. In this case, the buyer starts with a price that is much lower than the maximum that he is willing to pay and then gradually moves higher. Priceline uses this for its travel, which makes the buyers to feel like they got a great deal, but also ensures that sellers are able to get some producer surplus.

Why Auctions Are Important:

Even though there's a myriad of reasons for eschewing auctions, they might be the best way for sellers to discover the right price, and therefore, for a platform to build momentum over its competitors. In my opinion, the auction strategy works best when the outcome of the transaction is not time-sensitive.

As I discussed in my previous post on Ride-Sharing Apps, it is difficult to know how much a customer would be willing to pay for a product or service. In economics terms, it is difficult to draw a demand curve (how much of something would be sold for each price level) in real life. Therefore, it is difficult to price where supply and demand would meet. What happens if you can't price where supply meets demand? Let's look at the following graph:

At P1, the price is too high and supply exceeds demand. For example, you are setting prices for renting out apartments for a night, and you set the price at $100 (P1). You get 10 owners willing to lease out their apartments for the night, but you only get 5 renters. This results in excess supply. The number of transactions is 5, revenue is $500.

In the second scenario, you see the excess supply and decide to lower prices to increase quantity demanded. The next night, you set the price at $50 (P2). The scenario flips, and you get only 5 owners willing to lease out their apartments, but 10 renters. The renters get the apartments on a first-come-first-served basis. Pricing at P2 results in excess demand. The number of transactions is 5, revenue is $250. For this reason (lower revenue), most platforms are more likely to overprice than underprice.

In the third scenario, EP, which stands for Equilibrium Price, is where supply would meet demand. In this case, the right price could be $75, and you might end up with 8 lessors and renters. It's tough to know where to price, however, and it could become a guessing game. Here, supply meets demand, the number of transactions is 8 and revenue is $600 (higher than if you had over-priced).


Let's look at a fourth scenario, where price isn't set by the platform. Each individual apartment owner puts up his apartment for a reverse auction, sort of like Priceline, where the prospective renter submits the daily rent she'd be willing to pay. She'll start off lower than the maximum that she's willing to pay, and he will either accept or reject. That is, she'll submit a bid closer to P2 and move higher if he rejects the bid. There's no guarantee that they'll end up at EP, but if they agree on a price, it will likely be lower than the maximum that she's willing to pay and higher than the lowest amount that he's willing to accept. In large quantities and with free-flowing information (how much did similar apartments rent for on the same day?), prices will move towards EP naturally, without the platform having to do the guesswork.

Auctions can be a very powerful way to grow the marketplace precisely because they allow prices to be set dynamically through supply/demand. Especially for commodity services and products, buyers may be able to pick up what they need for very cheap, giving them the incentive to keep coming back. For example, Handy's cleaners may be willing to work for less on weekdays during school hours, which, if Handy were an auction, would be reflected in the pricing. Or, someone who only needs a sub-par writer should be able to bid a much lower amount for my writing skills on Fiverr than for an excellent writer. Were some of these platforms' prices auction-based rather than pre-set, I'd surmise that prices would be lower in many instances, resulting in higher consumer surpluses and a higher number of transactions. 

From sellers' point of view, price discovery becomes easy since they can price discriminate. Apartments will demand a higher rent price on July 4th weekend, for example, but owners shouldn't have to guess how much higher. Moreover, apartments on July 4th weekend can then be allocated to those who are willing to pay for them instead of on a first-come-first-served basis, which is what would happen if rents were set arbitrarily than through an auction. 

While platforms may fear that dynamic pricing may result in variable or lower commission (rake), I believe that the higher growth (from greater transactions) and the network effects that come with it will make up for that. Furthermore, Gurley gives good reasons for keeping the rake low, which is what may happen if overall prices are low and the platform takes a low percent of transaction as its rake. He states, "If your objective is to build a winner-take-all marketplace over a very long term, you want to build a platform that has the least amount of friction (both product and pricing). High rakes are a form of friction precisely because your rake becomes part of the landed price for the consumer. If you charge an excessive rake, the pricing of items in your marketplace are now unnaturally high (relative to anything outside your marketplace). In order for your platform to be the “definitive” place to transact, you want industry leading pricing – which is impossible if your rake is the de facto cause of excessive pricing. High rakes also create a natural impetus for suppliers to look elsewhere, which endangers sustainability." 

And while having high consumer surplus and low prices is imperative for building the demand side, Mike Russell of Paintzen talks about building the supply side of the marketplace on the Traction podcast. A good marketplace will strive to know and solve for the sellers' pain points in a specific industry, but reducing friction in terms of pricing (rake), as well as providing an avenue for sellers to price discriminate and enlarge producer surplus will make the platform sticky for the sellers. Therefore, auctions can be an elegant way to solve for price discovery where time is not of the essence, and they can help expand a new marketplace by attracting buyers due to higher consumer surplus.