(This is the second part of a four part article on running sales at a startup. You can find the first part here and the third part here).
II. Strategy
Strategy means deciding who your best potential customers are and how you’ll reach them. Strategy development requires that you think before you act. New hires often try to hit the ground running by generating a lot of activity from day one. For them, strategy development takes the form of throwing stuff at the wall and seeing what sticks.
That’s a mistake. You’ll discover that some things worked and some things didn’t, but you won’t know which is which because you weren’t matching activity with outcomes and systematically measuring the results. When you ask yourself what you should double down on, your decisions will be based on anecdotes rather than evidence.
Establish a process to discover your strategy. The goal is to find customers who fit three criteria: (1) best fit your product, (2) are valuable to serve, and (3) are cost effective to acquire. You start with many plausible hypotheses. You narrow these through analysis and testing.
Here’s how it works:
Strategy development is perpetual but means different things depending on your company’s stage. Discovery stage companies are focused finding best fit customers who are valuable to serve and run binary tests that succeed or fail. Scaling stage companies also focus on reaching customers in a cost effective way in addition continuing their work on issues of best fit and value. They run optimizing tests that show that method A is 10% better than method B.
A. Hypotheses
1. Best Fit for Product
Fight on ground where you have the advantage; sell to customers whose need for your product far outweighs their barriers to interest and adoption.
Needs are high when:
- Your product addresses a major pain point or creates a major opportunity.
- The need is felt by a powerful person or department.
- The need impacts their entire organization.
- Your product is vastly better than competitors and the status quo.
Barriers to interest are low if:
- The customer knows they have a need.
- The customer is looking for a solution.
- The customer is looking for your type of product.
Barriers to adoption are low if:
- Your product is easy to try because it is simple and low risk.
- It is easy for the customer to decide to buy, such as when the decision can be made by a single low ranking person.
Generate a lot of hypotheses about what segments would best fit your product. Define segments based on externally visible characteristics, such as 200-500 person accounting firms in Southern California that are at least five years old. Don’t include traits like “risk tolerant” or “innovative” unless you have some way measure them from the outside.
You’ll winnow your list of ideas using logic as you exclude customers who aren’t plausibly valuable or efficient to serve and as you test your hypotheses experimentally.
2. Value
Are any of these segments valuable enough to pursue? This depends on how many companies there are in the segment and how much each customer is worth.
The customer’s value depends on the amount of money it will pay you over the life of your business relationship, the lifetime value (LTV). You calculate LTV by taking the customer’s monthly recurring revenue (MRR) and dividing it by the percentage of customers that you expect to leave each month (churn). Bigger companies have lower churn, so segment your potential customers by size. You might use these churn benchmarks.
Make some rough estimates of the number of companies in each plausible best fit segment and their LTV. The way you use these estimates depends on your company’s stage. Discovery stage companies use these estimates to set long term goals. They may not be able to serve high LTV customers today but they need to make a plan to do so in the future. Scaling stage companies need to go after valuable customers today even if they don’t yet have a cost effective way to acquire them. They can refine their acquisition methods to make them cost effective.
3. Cost Effective to Acquire
The importance of optimizing your cost of acquiring a customer (CAC) depends on your company’s stage. Discovery stage companies are working toward cost effective acquisition in the future and should have a plan to get there, but it’s not their top priority right now. Scaling stage companies need to focus on CAC.
Think of CAC as the cost you pay to overcome the barriers to obtaining a customer. There are two types of barriers: barriers to interest and barriers to adoption. There are also two types of costs: marginal cost and zero marginal cost.
Marginal cost acquisition methods require more resources for every customer you acquire. Their advantage is that they scale quickly and typically show results in the short run. Here’s an example: Sit down at your computer. Pull up google. Take out your cell phone. You are ready to make cold calls. If you can connect with your target customers and there is best fit, you’ll start generating interest quickly. But this is an expensive way to sell and there’s a limit to how efficient you can make it as you scale.
Zero marginal cost methods don’t require more resources for every additional customer you acquire. That’s their key attraction: They’re a license to print money if they work. Their drawback is that they can consume a lot of time and resources before they hit critical mass, so you won’t see results for a while. Worse, you may not know if they will work until after you’ve put the effort in. For example, if your goal is to generate interest with blog posts, you’ll have to spend months at it to build content and name recognition. Once that’s done, you may have created a superbly efficient way to get new business or you may just have wasted your time.
Ideally, you’d use zero marginal cost methods to overcome both barriers to interest and barriers to adoption. But zero marginal cost methods work best when the barrier is low. The degree of each barrier depends on the interplay between the product and the buyer.
For example, if you’re selling a gmail plugin that lets an individual salesperson send mass e-mails, you face low barriers to both interest and adoption. Salespeople are looking for these tools, making the barrier to interest low. They also love to try new methods, making the barrier to adoption low as well. So write some blog posts to get people interested and have an online order form to let them buy. In other words, you can use zero marginal cost methods for both interest and adoption.
Selling new marketing automation software to mid-market companies has a low barrier to interest but a high barrier to adoption. Marketers are scouring the internet looking for tools to give them an edge, making the barrier to interest low. But switching software for a whole company is a big ask, making the barrier to adoption high. So you might use blog posts to get them to contact you but you’ll probably have to do personalized software demonstrations to get them to buy: zero marginal cost for interest, marginal cost for adoption.
Selling software to recruiters has a high barrier to interest but a low barrier to adoption. Recruiters are tenacious about screening out attempts to contact them, making the barrier to interest high. But they’ll jump at your product if you can demonstrate that it provides an advantage, making the barrier to adoption low.
Selling safety software for oil rigs has a high barrier to interest and a high barrier to adoption. Customers aren’t looking for a new solution because they think current methods work just fine, making the barrier to interest high. Implementing a change requires 50 people to sign off because of the risks, making the barrier to adoption high as well. You’ll have to go to events, network, and cold call to get customers interested and you’ll have to do demos for every department in their company to close them: marginal cost for interest, marginal cost for adoption.
You can visualize this in the 2X2 grid below.
There are matters of degree. Even if zero marginal cost methods work for some of your customers, it may still be profitable to use marginal cost methods to acquire the rest. For example, Hubspot sells marketing automation software, which normally has a low barrier to interest. Indeed, the purpose of Hubspot’s software is to automate zero marginal cost interest generation. Yet they still do a lot of marginal cost cold calls for their own interest generation.
Why? In any segment, barriers exhibit variance. Many of Hubspot’s potential customers have low barriers to interest while others have higher barriers. The LTV for the higher barrier to interest customers is enough to justify marginal cost interest generation, so why not pick up the phone and sell them all?
Marginal cost methods become more efficient with time. You might start out having core members of your team make cold calls to generate interest to see if cold calls work. Once you know they work, you can hire specialized, junior salespeople known as sales development representatives or SDRs to make cold calls. You’ll have more senior staff, known as account executives or AEs, focused on closing deals. Later, you’ll have a sales operations team that systematizes every interaction and a marketing team that supplies collateral to answer common questions and provide proof for commonly used value propositions. Specialization increases efficiency.
But there are limits to how efficient you can make marginal cost methods. For example, if you need to do demos to close deals, you probably can’t have annual contract value (ACV) of less than $1,000 and still profitably acquire customers.
Here’s a shortcut for the math: Ask if it’s profitable to hire salespeople at a given price point. Salespeople are usually paid 20% of the first year’s contract value, so $200 for each $1,000 ACV deal. Say your product is simple so you hire junior AEs and pay them $80,000/year between salary and commission (the low end of the range). If your AEs work 220 days/year, only close customers through demos, and close 30% of the customers they demo, that means they need to do about 6 demos per day to earn their keep. That’s a lot.
B. Experimentation
1. Knowledge Inventory
Start by figuring out what you know and what you still need to find out. If you do this rigorously, you’ll discover that you know fewer things and different things than you thought.
Analyze your current customers by creating a spreadsheet that breaks them into segments. For each customer, answer these questions:
- Why did they buy?
- How did they buy?
- Who made the decision and what was their title?
- How long did it take to make the decision after they became interested?
- How many calls or meetings did it take after they became interested?
- How much are they worth?
Then pick a few customers and take them to lunch. Even better, spend a day shadowing the person who uses your product. Ask open ended questions about both your product and their jobs generally. Write summaries of these conversations and circulate them.
Use this information to narrow the range of hypotheses you’ve developed above. Emphasize the information you gather from unaffiliated customers. You’ll probably discover that you need to start pursuing bigger customers than your current customer base.
2. Running Experiments
You need to have a plan for your experiments and stick to it. It’s easy to lose your bearings once you’re in the thick of testing. You’re rooting for your team to succeed, you have preconceived notions about what is right, and you’re often short on sleep. Unless you impose the discipline of a plan, you may waste a lot of time and learn very little.
These are the elements of of a successful testing program:
Your testing plan should contain a list of your hypotheses, ranked according to their importance and the level of doubt. Prioritize the tests so that you address the critical issues first.
What sounds like one hypothesis might actually be two. For example, let’s say you believe that 200-500 person companies best fit your product and can become interested in your product through blog posts. This belief comprises two hypotheses: one about best fit (200-500 person companies) and one about the efficient way to generate interest (blog posts).
Testing hypotheses one at a time is usually best. In the example above, you might decide to test your best fit hypothesis by cold calling the 200-500 person companies rather than writing blog posts. Why? Cold calls can start right away and, if you connect with your potential customers, you’ll quickly get a read on whether they really fit your product. Testing both hypotheses at once would take months as you scaled up your blog.
Predefine the volume of activity that constitutes critical mass, the amount of activity necessary to fairly evaluate the hypothesis. If you don’t, you might find yourself persisting with a test that has already failed because you feel that success is right around the corner. Critical mass varies depending on the method you’re testing: Two weeks of cold calls might give you critical mass but two months of blog posts might not.
Similarly, predefine success and failure. This means picking a single metric as the yardstick, such as the number of inbound leads generated by a blog, and measuring it against a target. Defining success is a judgment call. Don’t expect too much because the quality of your activity while testing won’t be as good as when you scale. But don’t expect too little because you don’t want to think that a test is succeeding because of a random variation or a fluke.
Finally, determine the exact content of the activity you’re testing. The more specific the hypothesis, the more tightly defined the activity. For example, if you’re testing a new cold calling script, everyone needs to follow it. If you’re testing cold calling in general, you can allow methods to evolve a little over the course of the test as your team learns. But the less specific your definition of the content, the greater the activity level needed to reach critical mass. Strictly speaking, you’ve been testing multiple activities and therefore need several times the volume to know if you’ve been successful.
After you’ve reached critical mass, take a breather before you evaluate the results. You probably lost a lot of sleep to make the test happen and it will be tempting to find rationalizations for failure. So get out of the office, have a few drinks, and get some sleep before you analyze the data.
Test results always pose a question and you should always look back at your testing plan before you answer. If the test succeeded: Should you optimize the method you tested by testing a variant or move on to the next test in your plan? If the test failed: Should you tweak it and try again or move on to the next test in your plan?
Usually, the right answer is to move to the next test in your plan. There are so many important issues you know nothing about; spending additional resources on the issue you just learned about probably has less marginal return than moving on to the next issue.
But not always. This is why you got some sleep before evaluating the results.
