SaaS Economics Blog

HubSpot has a unique culture.  We hire people who are thoughtful, smart and fight for what they think is right.  And we try to hire as many as possible of them.  Our core tool for deliver great value to our customers, is to have great leaders in the company. 

It turns out (not unsurprisingly) that these type of leaders don't like to be told what to do.  They want direction on where the company is headed, but not directions on how to get there.  As Dharmesh likes to say - they want to be inspired, not managed.  

To that end I find the best leaders at HubSpot do 3 things really well:

• Obsess over creating a vision (with a matching set of high goals)
• Obsess over driving alignment around those goals
• Empower people with the freedom to achieve those goals

And they do this regardless of where they are in the company - not just 'managers'. In fact, some of our best leaders are not 'managers' at all and have zero people reporting to them.  

• Anyone at HubSpot can set and sell a vision at a certain level - there are no offices, everyone's calendar is open to be booked for coffee, and we have an active wiki where ideas are shared and debated.  It takes a lot of selling, but the opportunity to be heard is there.
• Many people drive alignment through direct reporting relationship (and may complain when its not there) - but there are other (often better) tools to drive alignment including: Managing meeting agendas, creating regular reports, adjusting commissions or compensation structures, and running internal contests
• Finally most managers can empower people to achieve their goals by getting out of the way.  Often its as simple as asking 'who owns this project' - and once that person is identified, funneling all questions through that person.  If they see a roadblock ask again 'who owns this project', and make a decision. At HubSpot we call this person a DRI - Designated Responsible Individual.

One truth I've found about startups as they transition to become growing companies is that decisions tend to get tougher - not easier.

I attribute this to the fact that for any sufficiently novel or innovative idea - someone is going to think it's terrible.  If no one thinks it's terrible - then it's really not that good.  And the math behind that at even a small scale can be astounding.

Lets say for your brilliant idea there is only a 20% chance someone in the room thinks it's terrible.  So if you tell one person - there's a 80% chance you'll move forward with.  Pretty good.  But as the number of people in that room grows, the odds of everyone agreeing to the idea gets bad. Fast.  By 5 people there is only a 33% chance that brilliant idea will move forward, by 20 there is only a 1% chance everyone in the room will agree it's awesome.  1% - that's it.

What happens then, at most companies, is that to get any idea pushed through they settle on uninspired compromises.  They dilute their brilliant ideas down to something everyone can agree on, and lose all the magic and inspiration from that original idea.   It makes sense, it's the only way to get everyone in the room to agree.  It's just also path to bad management.

So at HubSpot we have a policy:  No uninspired compromises. 

We do our best not to let ourselves fall into this trap, and we do a couple of good things to avoid that mistake.

1. Give teams goals, and the autonomy to hit those goals.  Don't legislate the process - only the end results
2. Keep teams small. Don't let the room get too large so people can make decisions and move forward.
3. Identify a DRI (designated responsible individual).  That person owns the decision, not the entire room.  This is critical for cross functional decisions.
4. Require transparency. Hold people accountable to these decisions by reviewing the success of the decision, and then keep iterating.

Nate Weiner posted a brilliant article today articulating why his company - Pocket - moved from an upfront paid model to a freemium model.  The insightful post had lessons for all software companies looking for a simple framework to drive developing their pricing strategy.

Nate's argument boiled down to a failure of his previous (upfront) model to match the value the customer received over time to the price charged for that app over over time.  Pocket (formally ReadItLater) was charging upfront for an app that didn't deliver value until much later in time.  Nate's insight was that when the value is received upfront - charge upfront; but when the value is received over time, charge over time.

Here is a copy of the graphic from the post that neatly summarizes the different models for how value is delivered over time:

Expanding on Nate's post - with these three types of value-over-time models, there are a matching set pricing models available to most digital companies:

1. Value upfront >> Charge upfront (e.g. video games)
2. Value over time >> Subscription fees (e.g. Carbonite - desktop backup software)
3. Value grows over time >> Freemium (e.g. Evernote - note taking software)

In each cases, the pricing model allows the company to match the timing of the preceived value to the price charged.  Additionally there is a 4th type - when the value grows exponentially over time - specifically when there are network effects involved.  In these cases the matching pricing model is to be free, and then to find alternative ways (e.g. advertising) to monetize the value created.  That's why social networks like facebook or twitter don't charge the heavy 'users' of the platform that are actually helping them create value.

While this is a bit simplistic the core philosophy of matching the timing of the value delivered with the timing of the price charged is very powerful.  Taking this one step further - the best companies in the world have found out how to not only match the timing of value delivered as described above, but the amount of the value delievered.  This is where models like Google's Adwords, or Salesforce.com per user pricing are some brilliant. These comapnies have mastered the ability to match the timing and amount of value delivered - therefore creating successful, retainable customers and build remarkable businesses.

I recently posted an article on OnStartups about HubSpot's Experiments Framework.  I've reposted some of the relavent parts below.  The full article can be read at OnStartups.com 

The foundation of this framework is based heavily on Clay Christensen’s work inThe Innovators Dilemma.  We’re huge Clay Christensen fans at HubSpot (even have a conference room named after him) and have been life-long students of his work.  In his work Clay asks a very straightforward question without an obvious solution - specifically: Why do well managed, successful companies repeatedly fail to create new disruptive innovations?

This framework was developed fundamentally to combat that challenge and create a lasting culture of entrepreneurial exploration.

HubSpot’s Experimentation Framework

The framework has 3 stages, each with a distinct goal and approach.

Alpha – Lowering barriers to experimentation

No bureaucracy, no red tape, full access to information.  This stage is simply focused on enabling anyone with energy and an idea to try a new solution.  Tests are run by everyone and anyone – but are generally done in spare time (nights and weekends) and with few resources. You don’t need to ask permission to run these tests - and by design no one ever knows all the alpha stage experiments actively being pursued.  It's open and distributed.

Beta - Determining proper funding

When an experiment reaches Beta stage the ‘founders’ are fired from their day job and work on the experiment full time.  While founders determine their own goals and metrics – these leaders are encouraged to be patient for growth but impatient for profitable economics. Like many founders these people also report to a 'board' regularly and are subject to evaluation on future funding.  At its core this stage is about providing access to funding for entreprenurial folks with new ideas and transparency/accountability into the success of those early tests.

v1 – Scaling successful experiments

v1 projects have proven economics and now are looking to scale the success.  Often this requires growing the team beyond the founders, building dedicated systems and developing regular tracking of core metrics.  Founders with experiments graduated to v1 are now considered 'mini-CEO’s' and are tasked with running their project as a start-up within HubSpot.   

We established this framework in the hope of driving innovation and empowering the entrepreneurial edges of our organization to create change.  It seems to be working - we’ve had several successful founders graduate from the program (Pete Caputa with VAR program, Jordyne Wu with the Services Marketplace) and we created a culture to be proud of.   It's enabled us to focus on the core business without foregoing the entreprenurial engery and creativity of our team.

At a high level calculating the cost of customer acquisition (COCA) is relatively straightforward to calculate for a SaaS company.  It's simply the total Sales and Marketing expense for the period divided by the number of customers acquired in the period.  Easy.

Unlike other SaaS unit economic metrics, unfortunately, this simplicity breaks down rapidly when calculating COCA for a specific segment.  How do you break down the marketing spend aimed at the software industry vs marketing spend targeting manufacturing?  How do you break down the sales spend for SMB segment vs enterprise segment?  Its often very difficult to calculate reliable numbers within these narrow bands.

At HubSpot we've looked at several different calculations over the years and have come up with one simplifying approach - that although imperfect - can drive COCA when applied consistently.  

Measuring COCA by Segment:

At HubSpot we calculate COCA by segment as:

Marketing COCA =  Total Marketing Spend * % of Total Leads in the Segment

Sales COCA = Total Sales Spend * % of Sales Reps selling into Segment

Total COCA = Marketing COCA + Sales COCA

A few notes on this.  First, the total marketing and sales spend includes everything (managers salary, overhead, etc.) - not just direct program spend or commissions.   Some people do this when calculating COCA and I think its wrong and ignores some of the real costs there.  Second, this assumes all leads are the same price.  It's a simplifying assumption in the wake of more information around specific channel spend, we find works well for most companies.  Finally, this also assumes all sales reps have the same costs associated with them.  While this isn't always accurate depending on the mix of inside and outside sales reps - much like the marketing spend we find works well for most companies and is relatively robust.

With these assumptions and this approach we find that we can segment our SMarketing (Sales + Marketing) funnel down to very refined levels and measure the health of our business in specific markets.  

When we conduct this analysis we do not only look for segments with the lowest cost however - but instead try to identify segments that have the highest return on our sales and marketing spend.  So although COCA is valuable, the real insights come when compared to other unit economic measures such as the lifetime total value of the customers in that segment.  (lifetime total value, LTV is measured as avg. Monthly recurring revenue * gross margin / monthly churn rate)   

By most VC standards, SaaS companies should target a LTV to COCA ratio of 3, and a COCA payback of <12 months.  Many public companies are well within those ranges or much beyond them.  However - these numbers heavily dependent on growth rates - its much more difficult to maintain high unit profitably and solid economics while growing 100% year-over-year than it otherwise could be.  Said another way - in most companies if they fired 1/2 their sales team these numbers would improve, but the growth would stall.

For many of these reasons its essential for companies to understand the economics and COCA of the markets they target and the segments their business sells into.

So what do you think?  Are there better alternatives for calculating COCA by segment?  For some additional information also check out David Skok's piece on the Startup Killer: Cost of Customer Acquisition.

Harvard Business Review Daily Stat reported recently that for software firms operational skill is the key to avoiding corporate failure.  In the short post the reference a study done by Why Software firms Fail.

From the HBR snippet (accents are mine):

The study shows the importance of operational capabilities and the need to for management focus in this area in order to sustain its competitive advantages.  

As always though, the first step to improving Operational Capabilities is to first measure it. But how do you do that?

One specific potential measure the researchers hit upon (though they roll it up with other inputs) is gross margin.  That is - the difference between your operating revenue and the cost of goods sold.  I actually think this makes a lot of sense.  The goal of operations, as suggested in this study, is to  "use resources cost effectively" and that firms gain "an operating advantage byefficiently leveraging their resources to create operating income".

So what do you think?  Do you have someone at your firm whose job it is to focus on improving gross margin? 

HubSpot, like most companies, focuses much of its business operations efforts of making processes more efficient and collecting hoards of information to report on the success of those initiatives:

• Marketing and sales collect information about our leads - to help engage and convert these leads
• Customer team collects information on what marketing actions our customers have taken how successful they have been - to help guide the onboarding process more effectively
• Product team collects information on how our customers are using the product - to gain valuable feedback that drives the product development process

In most cases these collection points are completely self serving for the organization at hand.  They need this information to run their division more effectively and share it regularly within their organization.

This is common for many companies.  Using business intelligence to drive internal processes is nothing new.  But where many companies fall short is answering:  How can all the data I collect help our customers?

At HubSpot we've started down this path. 

• Marketing leverages data from the product team about how our customer do marketing to create content for the blog (often its our best performing content)
• Sales team leverage data from the customer team to create a 'inbound marketing calculator' that can help prospects understand the potential for inbound marketing in their business
• Customer team leverages data from the product team to create a monthly newsletter informing customers of what parts of the application they should explore next
• Product team (soon) will leverage to demographic data from the sales/marketing team to create a benchmarking tool specific to a customers industry

In each of these cases we've tried to look at the pile of data we collect at HubSpot and understand how we can re-purpose it to the benefit of prospects, leads and customers.

Is there information your organization already collects that you're just sitting on?

At HubSpot we've recently started a series of projects aimed at defining some new metrics for running our business.  We do this because (admittedly) we are dorks and love data - but also because we need to solve some real problems in the organization.  These problems generally range from aligning incentives, evaluating project success, or managing day-to-day operational flow.  In each of these cases, however, a common challenge emerges centered around finding the right balance quality and quantity.

Below I have identified a series of touch points between organizations where this has been particularly interesting for HubSPot. As you'll see we've tried a couple different approaches to identifying the right balance.  Nothing is ever perfect - but I think there are some take-aways here for others, 

Marketing and Sales: Sales 'Points' SLA

• Balance Need: Volume and Fit
• Tendency: Simply count leads because its simpler
• Solution: Score each conversion event (not simply lead) and hold Marketing accountable for a sum of those points

Sales and Customer Team: LTV base compensation

• Balance Need: New Customer Velocity and Retention
• Tendency: Compensate only for new units
• Solution: Measure sales reps by Total Lifetime Value generated each period - not just new units.  For SaaS this strongly influenced by a sales rep individual churn rate

Onboarding Consultants and Account Managers: Customer Happiness at handoff

• Balance Need: Pace of Consulting and Success of Consulting
• Tendency: Don't measure anything because its hard
• Solution: Determine 'customers happiness' objectively (frequency of use, features set-up, survey results) and hold consultants to a onboarding schedule

Product/Engineering & Customer Team (specifically services): Daily Pain

• Balance Need: New Innovation and Fixing Existing Innovation
• Tendency: Allot some % of time to bugs and hope it works
• Solution: Score open bugs by severity and hold engineering to a cap - and allow time for new development while below that limit

In each of these cases we have tried to understand the underlying motivations that lead to imbalances and identify objective metrics that when monitored regularly can keep us from tipping too far in either direction.