Sell a second product at ten to nineteen people

1. The stage

The goal of this range is to get past 20 people.

Item	At this stage
Headcount	Ten to nineteen; also put as 10 to 20
Your role	Director: you went from managing people to directing managers who sit underneath you
Leadership structure	One and a half layers: your first two, maybe three, small managers, each running two to three people, so almost two layers, while you still lead some people directly
How many companies get here	2.4 million, down from 3 million at the stage before

2. Each function at this stage

Function	What it covers	What breaks	What graduates it
Product	The goods, services and results you deliver	Customers have nothing else to buy from you, and then they leave	Connect customer service and product, and make something new to sell them
Marketing	Telling people about what you sell	Qualified leads cost too much, and that caps how much you can advertise	Spend more on the narrowed message with the new offer, raise the limit on what you pay to acquire a customer, score the leads yourself, and start conversion-rate optimization
Sales	Getting people to pay you	Customers are sold with unrealistic expectations, and refunds and bad reviews become an issue	Sales aids, a weekly call review across customer service, sales and marketing for quality and feedback, and the recommend score, satisfaction score and refunds tracked by rep
Customer service	Getting customers to refer, retain, review and resell	Customers arrive with unrealistic expectations and get inconsistent answers	A customer service playbook to onboard and train the team, the weekly call review, and an ascension path
Information systems	How you gather, store, analyze and display information	People use their own hardware and software, which causes trouble when they leave, and nothing tracks leads and customers dealing with the business	Company-owned and company-controlled hardware and software, and a dashboard and reporting system
Recruiting	Finding and hiring people	You have trouble filling higher-level roles and manager roles	Promote from within on purpose rather than by default, and pay for employee referrals so the cost to acquire talent stays low
Human resources	Lowering the chance that regulators and employees sue you	You fire people incorrectly, so you are exposed	A termination policy and process
Finance	Managing and reporting your money	You do not have much money to reinvest into growth	A budget that saves for big expenses, simple forecasting, and a record of tax write-offs

3. Product: connect customer service to product, and build the back end

You niched the product down; customers bought the version made for them, found nothing else to buy, and the relationship ended.

The back end. Typically this is where a back-end upsell comes in. Businesses that arrive at this stage are doing maybe 3 million, sometimes 10, somewhere in that range. They need a back end to scale: going broader keeps lowering lifetime value against acquisition cost, while lifetime value stays flat.

Typically five times the core. In the experience here, and as the preferred way to build one, the back-end upsell is typically five times the price of the core offer. It is the 80/20 rule again. Inside the best customers you niched down to, another 80/20 applies, because customers are fractal most of the time: 20 percent of them can spend five times more. Get that 20 percent to buy something five times the price and revenue doubles.

Customers	Price	Revenue
100 buy the core	$1,000	$100,000
20 of them buy the back end	$5,000	Another $100,000

Those 20 might double the business, and might double your lifetime-value-to-acquisition-cost ratio, which lets you keep expanding.

What to sell. More of what they just bought is called one of the favorite upsells, second only to more help with what they just bought. Then comes a better or newer version of it, or the same thing for longer.

Eight ways to make a customer worth more. You should be able to run any business, product or service, through eight ways whenever you need more money from something you sell. A service example is given because 78 percent of businesses are said to be services.

Way	A chair seller	A marketing agency
Raise the price	Charge more for the chair	Charge more for the services
Get them to buy again	Reach out on a more regular cadence, maybe a yearly refresh	Cut churn so they buy again and again
Cross-sell something different	Chair covers: the natural next purchase, like fries with the burger	SEO, creative, content or email follow-up on top of paid ads
Lower your cost	Make the chair for less; not necessarily an upsell, but more lifetime value	Offshore the team for cheaper labor, or raise the ratio so one rep serves ten companies instead of five
Upsell quality	A wood chair instead of a plastic one	A senior or more experienced rep instead of a junior one, or the owner
Upsell quantity	Ten of the plastic chairs	Typically either touch points or duration: a 12-month contract instead of three months, or three check-ins a week instead of one
Downsell quality	A paper chair, or a flimsy lawn chair, instead of a plastic one	Work with one of the reps for less
Downsell quantity	A smaller chair instead of one full chair	Six or three months instead of 12, or less often: maybe every other week, or chat support rather than Zoom calls

Chat support instead of calls is called more of a quality downsell. Downsells do not raise lifetime value; they raise revenue per lead, turning a no into at least some money. SOP 72 sets out the same eight at depth, in its own order.

Which one to build. This is why customer service is tied to product: it can tell you what people ask for, and sales can add what it hears, to guide which of the eight to pick.

4. Marketing: spend more, score leads, test on a cadence

Until the second product exists, lifetime value is not high enough, and leads keep getting dearer.

• Spend more on the narrowed message with the new offer, since the upsell makes you more.
• Raise your acquisition-cost limit, because the second product now counts in lifetime value.
• Score leads yourself. Move from leads scoring themselves, through friction added to the process, to your team looking at incoming leads and scoring them.
• Begin conversion-rate optimization: a regular split-test cadence across every part of the funnel, read on the basic numbers (show, close and offer rates): does this version lift 10 percent? Sometimes lifts of 5, 20 and 15 percent across four or five places can double a funnel's throughput.

Pair the best leads with the best closers. In one practice here, a lead is judged on qualifications (who they are), behaviors (what they did) and source (where they came from). An owner who calls himself a CEO, has 50 people, watched every video, bought two books and came in through inbound organic LinkedIn might be worth more than a smaller business from TikTok that watched one clip before opting in; attribution on the user journey lets you see both. Your worst closer could blow the big lead, so the best leads go to the best closers, called a huge lever. SOP 84 goes deeper.

What to split-test first. One person's priorities, in rough order:

Order	Test	Why here
1	Headline and subheadline	A light lift with high impact: high impact, low investment, so test these hard
2	The offer	More impactful, but higher investment: a new offer changes what you deliver and how the messaging works, while the headline only describes the offer, so it can lift almost as much for far less work
3	The image	Less powerful than the offer but still a lift: medium to low impact, low investment
4	Benefit bullets	Not established on this page
5	Proof	Called a big one
6	Placement: above the fold, and the order on the page	Sometimes a scheduler higher on the page works better, and sometimes the order of proof matters: maybe one testimonial should come before another

Flag: the test order. SOP 41, section 6, puts the call-out, the value elements, the offer and the call to action ahead of the landing page, and the headline, subheadline and image last. Here the headline and subheadline come first. SOP 41's way tests the front of the pathway first; this way runs cheap, high-impact tests before the offer. This page does not settle which reading is right.

Install attribution software for more accurate testing and reporting; at this stage you can afford it.

Recalculate what you can spend. With all of that in place, recalculate the ratio and set a new acceptable acquisition cost; one example is $33,000 where you once could spend only $1,000. Know your 30-day cash collected: whether first-month cash offsets what a customer costs you. Breaking even inside 30 days is called thrilling; beat that and you are paid to get customers, with the back end still to come. Push hard here: as long as you are not losing money in the first 30 days, you are good to go. Some of the companies seen here that reached tremendous scale were willing to lose money for three, four or five months before earning it back, because they knew their numbers. Being willing and able to outspend competitors is how you keep expanding; you sometimes lose money up front.

5. Sales: aids before the call, a call review, numbers by rep

Sometimes new reps set bad expectations, and your reputation takes the hit.

Sales aids. Aids are slide decks, video sales letters, sales letters mailed or sent as PDFs, charts, graphs, data visualizations, calculators: anything that helps reps sell more consistently. A script probably has a stretch where the rep educates and breaks beliefs, and repeating it is tedious. The preference is to put that stretch before the sale, in a video with data and images, and talk it through on the call. Reps stop repeating the same 20 minutes, the prospect is not sat through a monologue (thought to cost rapport), calls get shorter, and expectations are set the same way every time. In one practice here they go into almost every sales process, to steady the close rate: 30 percent every week is preferred to 20 one week and 40 the next, and ideally 40 every week, from repeating the best sales.

A pre-call video, in order. The structure preferred:

Step	What it does
1	A strong, detailed hook built on the pain of your ideal customer, right at the start
2	Proof, promise, plan, with the promise tied to the pain
3	Why the problem has not been solved yet for this person
4	The main impact if it continues
5	The solution, and how it works
6	What makes it better than other people's
7	Proof that it works, woven through the whole video
8	The risks of investing, named: do not hide the risk they are thinking about; they are probably worried it will not work for some reason, or will not fit their tools, so handle each
9	Who they will talk to on the call, then the booking; the booking can also come first, up to you

The call review. Departments watching one recorded call together is SOP 135. At this stage:

• Who. Customer service, product and sales meet, and sometimes marketing too. The head of sales usually arranges 60 to 90 minutes when the heads of sales, product, marketing and customer service can all meet; others can join, and sometimes every department does.
• How often. With a live problem (maybe high churn, missed expectations, customers leaving) you might hold it three times a week or daily until it is fixed, even twice a day: hit it with volume. Without one, the view here is once a week, to prevent problems. Later it can drop to once a week or every other week; at every other week, one sales call and one customer service call make the month.
• Which call. The head of sales gets a recording from a rep ahead of time: one that needs work, such as a call that did not close, or one that shows what good looks like. Start with sales calls, then add onboarding and coaching calls. The sales call is usually the issue: a lot of times the onboarding team has never seen one, and your sales team might never have seen onboarding.
• Assess, analyze, act. Each department asks what it can take: marketing, what it learned about the customer and what the call says that its marketing does not; sales, whether it kept to the script or over- or under-promised; customer service, what sales promises that it does not loop back to or cannot deliver. Each names one thing and acts on it within seven days, before the next review.

It is called, as a promise, one of the most valuable calls you can hold: most businesses at this stage work in silos. Four heads on one call can look like a $10,000 call; it is said to return more.

Flag: the call review. SOP 135, sections 3 and 5, holds it weekly, lists marketing, sales and customer success, rotates the call, stays on a department with the constraint for weeks at a time, and commits the team to one thing. Here the head of product attends, the cadence runs from twice a day to every other week, and each department names its own one thing. SOP 135's way keeps one change in hand at a time; this way puts more meetings and more changes against a live problem. This page does not settle which reading is right.

Numbers by rep. Track the recommend score, the satisfaction score and the refund rate by rep. You will see quickly that refunds are probably coming from one or two reps, and it usually comes down to bad expectations or overpromising: they "sell hot", a polite word for lying. Minimize or end it. The first rule of marketing and sales in one practice here: state the facts and tell the truth, and if the truth is not compelling enough, change reality until it is. By rep, a high refund rate can turn out to be one person while everyone else is fine. The two scores are defined on SOP 149, section 7.

6. Customer service: a playbook for consistent answers

The playbook. Before hiring more people for all the questions, sort every incoming question into categories. There are usually three to five that carry most of the volume, with some stray ones that fit none. Examples might be: change my card; change my address, for a physical product; a different flavor; a different delivery cadence from monthly; more or less of what I get. The playbook is templated answers to the most common questions in each category. A script makes sales consistent; templates make service consistent. Put them wherever you answer: email templates, a QA framework, Intercom, a ticketing system. Reps fill in the name and problem around the same core, so at least the same questions get the same answers.

The empowerment wheel is on SOP 161, and the ascension path on SOP 162.

7. Information systems

Moving people off their own hardware and software onto the company's is SOP 155.

8. Recruiting: promote on purpose, and pay for referrals

Do not promote by default. A known management principle holds that people get promoted until they reach a role they cannot do. A lot of times you promote for convenience, because nobody else fits and you do not know how to find the right person, and get managers who are inexperienced, incompetent, or both, which is often the case. The department under them suffers, whether sales, customer service or finance. You lose twice: a strong individual contributor, and then a weak manager whose whole team does worse.

Protect against it one of two ways:

1. Run them through the normal hiring process, preparing an interview for the role as usual, to see whether they would be hired from outside; process protects you from your emotions. A lot of times insiders are not asked the same questions or graded the same way; treat them as outsiders.
2. Or, if you believe they are ready, try before you buy: a trial of 60 to 90 days in the role. A customer success manager who wants to be director, say, gets a 60-day trial with milestones; the benchmark is what a new hire would do in the role's first 60 to 90 days. In the view here, it invests in someone you have confidence in who likes learning and growth: at worst they gain skills, at best skills and the role. Mark the job description red (weak), yellow (fine) or green (great); by the end every part must be green. Check in, say, weekly or every other week.

Either way you are objective, not subjective: put people in roles on purpose, not for convenience.

Pay for referrals. People referred by someone in the company are said to stay 70 percent longer. As with lifetime value, if you can keep people longer, you can invest more in hardware, software, gatherings and training. Like the cost to acquire a customer, there is a cost to acquire talent; referrals keep it low.

• Anchor on what an agency charges. The average agency takes 20 percent of the role; 20 percent of an $80,000 role is $16,000. Why not pay your team half that instead?
• Make it worth their while. Referral pay went from $2,000 at a first business to $5,000 now, and people try significantly harder. Five referrals at $5,000 come to $25,000, which could be a third if not half of a salary. You could pay up to what talent costs you (SOP 252, section 7, pays a recruiter's whole fee, part at month six; SOP 200, section 8, pays at month six).
• A worked case. An insurance brokerage lived on recruiting productive agents, each worth $250,000 a year to its owner. He offered 10 percent of year one, $25,000 an agent, paid once the agent was productive, a sum the owner would happily pay; the business was transformed.
• Why it compounds. For a revenue-producing role whose output you know, it is the customer ratio again: a reinvestment with an expected return, called one of the most underused ways to compound a business. The referrer champions the new hire, wanting them productive by day 90 to collect the payout, and refers only people they want to work beside. The money stays in the team.

9. Human resources and finance

Letting people go without surprise, with the work covered, is SOP 153. This stage's budget, forecast and write-offs are SOP 151.

10. The bottom line

You had nothing new to sell your customers because lifetime value against acquisition cost was too low, so you build a second product around them. To graduate, make and sell that second premium product to those customers, and keep reinforcing the infrastructure as you scale. The next stage is 20 to 50 people, where the theme turns from productize to optimize.

11. What this page does not decide for you

• The two flags in sections 4 and 5. This page does not settle which reading is right.

12. What this page does not cover

Split-test significance is SOP 41 section 3; sample sizes are not covered on this page.