Understanding SaaS: Why the Pundits Have It Wrong | Hacker News Reader

Understanding SaaS: Why the Pundits Have It Wrong

a16z.com

123 pointsmoritzplassnig27 comments

For those interested in getting a good overview of perpetual vs. SaaS business model I highly recommend Dave Kellog's post on the topic. He discusses both the operational and valuation impacts and walks through an example of a hypothetical startup under both models.

kellblog.com/2011/01/26/perpetual-money-vs-perpetual-license-subscription-saas-and-perpetual-business-models

Summary:

1. Wall Street "sees through" the differences in models and value perpetual and SaaS companies roughly equivalently. SaaS companies are worth 1.8x the revenue multiple of perpetual companies (he walks through the math in the post)

2. There are many good reasons for perpetual companies to move to SaaS models but valuation isn't one of them

3. You get roughly twice the EV/R multiple as a SaaS model but building the revenue stream is just about twice as hard. CEOs who have done the transition from perpetual to SaaS say it takes 3 years to makes the transitions and it must be a top 3 company goal for that entire period.

4. SaaS dampens revenue volatility - for better and for worse. Makes it harder to grow the revenue stream quickly and makes it harder to change once established. (This has an impact on investor psychology and reactions to a bad quarter can be very different in a SaaS model vs. perpetual)

5. Sales compensation is a tricky issue with SaaS model. Sales people still want dollar compensation similar to a perpetual sale despite ratable revenue profile of SaaS.

6. The implicit assumption that an annual subscription to use a service should cost less than equivalent perpetual license can be invalid when looking at the product from a customer Total Cost of Ownership viewpoint. (Companies are also outsourcing the capital intensity of having perpetual software)

great add and summary. Hadn't seen this.

On your point 5--amen. If marketing is using affiliates, you see the same problem, plus issues of fake sign ups. Have to make sure affiliates are compensated only on 90 day + signups.

Great summary. Another thing that's small but interesting to me is the discount rate. When calculating LTV, (Annual Recurring Revenue x Gross Margin) ÷ (% Churn + Discount Rate) the discount rate is effectively 0 right now due to historically low interest rates. I think that's helping a lot of SaaS companies get going right now. Or who know, maybe I'm missing something here.

I am also curious on founder stakes in SAAS companies Vs larger enterprise companies. My hunch was that that the customer acquisition cost (before it was paid back in say ~24 Months) are being financed by VCs instead of customers in the previous enterprise license worlds.

Since, 1) VCs are more sophisticated than the typical customers, 2) have more bargaining power - Are founders being diluted more in SAAS companies (Box.net) vs enterprise companies (e.g. Oracle)?

Great post. Superb introduction to the concepts with helpful illustrations. Adding a few ideas to the mix...

Typically a mature SaaS business has about 50% of its revenue from new customers and 50% from renewals. Early on that number is skewed to growth but if a company doesn't invest in retention early on, the company can go upside down, fast. Watching cohorts each month as their subscription expires is a good way to catch this early. Especially if you're in accrual accounting, because you can't recognize new revenue right away.

The 3x CAC for LTV is a good rule of thumb but does require you think about what LTV is. In the late 90s companies acquired customers using a 5 year LTV assumption which led to some really inflated valuations and lopsided businesses. I am conservative in my own startup right now, no more than 1-2 year LTV assumption for now to preserve cash flow until product market fit and retention are more consistent. one great reason for taking investment is to take some risks in acquisition on LTV, Gambling intelligently that you can improve your retention enough to match.

Early on, the ratio is likely to be even higher until you can drive the inefficiencies out of the spend. And as cost per click goes up in a given market you have to think very, very hard about it. If you're in an enterprise SaaS business, especially in marketing, it's really ugly for top of funnel terms. I've seen some terms at $11-$12 a click.

They also did a podcast on this subject. Well worth a listen as well: https://soundcloud.com/a16z/a16z-podcast-valuing-todays-fast...

For those who prefer to read:

I paid for and posted a transcription of that podcast.

http://blog.chrisbarber.co/a16z-podcast-valuing-todays-fast-...

Nice. How does the transcription work, and how much does it cost?

This is a good article and, although many people will know these concepts separately, it's good to have them all stitched together.

For me the question raises how to get through the dip when your cashburn is higher than your revenue because you haven't accurately established your LTV.

The second part of that is knowing how to price your service so that your LTV will exceed your CAC. You can control both parts of the equation (through efficient customer acuqisition) and by modifying the product price. But they don't work separately on each other - an adjustment in one will reflect in the other.

I guess there is no magic answer to that and it depends on your customers, your competitors and whatever precedents you think may be applicable.

I heard A8n on a BBC podcast trying to explain that bitcoin was important not because it was a currency, but that it solved the Byzantium Generals problem, and so was a break with most other ways of paying people.

This is another we-thought-about-this-for-a-long-time-to-work-it-out article. Not A8N quality, but close.

Edit: this repeats pretty much the accepted wisdom as MicroConf et al - Get the customers paying yearly now so you can use that cash to go acquire another customer. Nice to see Rob Walling / Jason Cohen are ahead of the curve still :-)

Most SaaS pricing pages give you the impression that pre-payment is prefered to month-by-month (hence the discount you get for buying 12 months today). With this in mind, isn't the old enterprise model just the ultimate future payment, and thus the ideal pricing model? You get all the cash immediately and there is zero churn risk.

The SaaS model is great for the customer and bad for the business. Maybe that's why the market doesn't like it so much.

that's too simplistic. there are pros and cons to each. for example, when running a saas you have a decent idea of how much revenue you'll have next month. with a one time sale model you start next month at 0 and only make money if you find new customers. (obviously not taking into account maintenance contracts, but still the point remains)

How about if after getting a million dollar check, you just pretend that it represents 100 10k monthly payments with a 0% chance of cancellation. Does that sound more stable/predictable?

Great article! A few comments:

> In the perpetual license model, in-house IT staff managed all software instances and thus could incur the internal costs to switch vendors if they so chose.

This argument is flawed. There are switching costs no matter if it's SaaS or on-premise. This is why big companies outsource IT - it's easy to get rid of 100x Accenture consultants who specialize in SAP and replace them with 100x Accenture consultants who specialize in SalesForce. The switching costs are normally around process change, change management and training.

> Budgets are much more decentralized now, because departments often adopt SaaS technologies and make purchasing decisions independent of the centralized IT organization. In the past, the tops-down technology sales model made it very easy for a CIO to unilaterally replace application vendors.

Which makes them less sticky. Team frustrated using the $150/month PM tool that they chose? Good, they can get rid of it. LTV goes to zero.

> This is why many SaaS companies today invest aggressively in sales and marketing when adoption is high, even though it puts pressure on current profitability.

SAP and Oracle did too at one point. They're milking their ~90% gross margins on support contracts today. BUT, they still spend a shitload on cost of sale. I don't see cost of sale going down[1].

> As a general rule if LTV is 3X or greater than CAC, that’s a good sign that the business model is working.

The fundamental problem today with SaaS is that I have yet to see to that a scalable and sustainable LTV exists (i.e. "winner take all"). As the OP points out, we can insinuate (not completely prove) a 3x LTV for Workday, but what exactly are they doing that SalesForce hasn't done in the last 15 years? Lastly, if this is a winner-take all situation, then shouldn't we stop talking about SaaS, since "SaaS" = "Workday"? Or perhaps, is this simply pointing out that if you're a SaaS your sole position should be to spend donkey-loads of money on marketing/sales, because you'll just end up getting acquired for your customers anyway?

[1] - http://www.techdisruptive.com/2012/11/28/how-are-we-going-to...

these are great questions...I suspect they're looking at erp equivalents with heavy technical integration, not the basic project management app, which have switching costs, but nowhere near the complex, often heavily regulated, issues involved in a multi billion dollar company's data integration situation. Not saying you're wrong, just that in large installs the post is closer to true than not.

As for spending donkey loads of money on marketing, it's really hard to get that model funded today. Investors are wise to it. Not to mention, paid search isn't nearly as cost effective as it was in certain categories. So you're left to grow on only product, SEO, and content, hoping for a viral loop. investors only (should) come in when you can prove that $1 in marketing spend yields $3x in revenue.

Just a note: as far as I know only 20.3% of Oracle's revenue came from the sale of new software licenses and hardware while four times that much, or 79.7%, is from annual maintenance fees and services (basically some kind of subscription). It is very dangerous to assume that these big companies are stupid.

You could argue that those stats are a reflection of a mature company with the license model. When they were in a similar position to many of these SaaS companies we are discussing, their license fee and customer acquisition cost were enormous. That automatically self-selected their customers to be large enterprises, who they were completely dependent on. The initial license fee is such a large proportion of LTV compared to maintenance fees. With SaaS, companies are banking on scalability, personalization (multitenancy), and less sources of friction. Establishing a successful SaaS company allows them to realistically chase the long-tail of SMB that single-license companies cannot scale to do effectively.

Part of the distaste is that many of these SaaS stocks are trading with valuations as if they're already at the far end of the J-curve. The market, evidently, agrees with Marc, so I'm not quite sure what he's complaining about.

I've seen these LTV / CAC ratios before, but what about when CAC is zero? That is often the case for startups that don't spend any money on sales/marketing...

CAC is never zero from an accrual based accounting standpoint. You have marketing software, discounts, staff, etc. that technically count against it.

That said, CAC never remains zero in the long run, unless you're Facebook with huge viral coefficient, and still, it's not zero. If it was zero, you'd be in a massive arbitrage that's likely to be unsustainable. Certainly not in any competitive market.

Ok, so you could do

CAC = (quarterly business expenses [like salaries, rent, health insurance]) / number of customers acquired in that quarter

this is an accounting question. Rent is not a marketing expense and neither is health insurance. But salaries are. As is any SaaS you buy, from hootsuite, optimizely, etc. If you're on accrual accounting (and as a startup, why? no reason to do this until you're much farther along) you recognize the expenses in the month you have them, but the revenue is divided by 12. this is the math the post is describing and why SaaS businesses are so hard--early on the metrics look terrible because the revenue is deferred but the expenses are not.

CAC is the cost directly related to the acquisition of the next customer (ie the variable costs). This is not the fixed costs of running a company (salary, rent, etc). It includes basically anything you spend before you sign up the customer, like answering emails to prospects, implementing new features, fixing bugs found during trials, writing blog posts, etc. Or for high $$ SaaS companies, advertising, phone calls, meetings, proposals, demos, seminars, writing letters, lunch & learns, etc.

If CAC is zero, you are basically saying customers are finding you and signing up on their own, with zero effort from you. This is possible (ie Twitter), but the LTV (life-time value) for these products is usually very low. Usually, the company is expecting a few dollars per year of revenue from each customer from advertising.

did my response below not show up, because that's what I just said.

It's also important to address Customer Success Costs. For single-license model companies, all the costs are front-loaded into customer acquisition. For subscription SaaS companies, the product & overall service needs to be continuously improved in order to retain customers. For this reason, it is not uncommon for these companies to release updates at a very frenetic pace not seen in the older style companies.

You never have a CAC of zero. I mean, you could pretend that all of your costs are R&D or Development, but is that actually true? What about the time you spend posting to the company blog or building the website, or AB testing or any of the other things hat end up getting you customers?

Say you are putting an app in Apple's App Store. Your CAC at the very least is $100 + any time spent to integrate with Apple's store, get approval, etc. because if you aren't in the store, you don't get those customers. You might classify them as development costs or somehow marketing/sales costs but the bottom line is if you're talking generically about what it takes to acquire a customer, plenty of things you do in development are to acquire more customers. How you classify those costs is sort of up to you, but don't be under the assumption that just because you are doing SEO or App Stores or Craigslist that you have $0 CAC.

Your time has a cost. Development time or content creation time has a cost.

It might not be easy to track in the traditional accounting sense, especially in the early days of a company, but never believe your CAC is $0.

CAC doesn't really play into things when you're relying on word of mouth. It helps answer the question "When I go to scale this business, how will it scale?"

Reply on news.ycombinator.com