As pure software opportunities become more scarce and capital becomes more abundant, full-stack startups are in vogue. What distinguishes a full-stack startup from a traditional startup? Traditional startups sell software tools, full-stack startups are services that sell outcomes. It’s the difference between Taxi Magic and Uber. I’ve written about this topic before in The “Full-stack startup” and Jiro Dreams of Sushi, and in this essay I will go deeper and offer full-stack startup operators a framework to help increase their chance of success. The factors to consider are:
- Frequent usage
- Expectation of long-term relationship
- Ability to take a cut of a large transaction
- High perceived expertise
- Linear relationship between tasks and short-term outcome
- Limited variety of short-term outcomes
- Ability to up-sell a long-term outcome
Break down your bundle into tasks, short-term outcome and long-term outcomes
In 2005, Clayton Christensen popularized the concept of “Jobs to be Done” in the Harvard Business Review, which asked entrepreneurs to examine the jobs customers are hiring their products and services to do. For full-stack startups, the “job” their customers typically want done is an outcome, which is delivered by completing a bundle of tasks. The task-level is where traditional startups have feasted by selling software, often to professionals. Full-stack startups are their successors, competing directly with professionals. Here’s a table with a few examples.
When you hire LegalZoom, you are hiring tool that gives you lawyer-approved legal documents to complete the task of “incorporation.” When you hire Atrium, a full-stack startup, you are hiring law firm to help with short-term outcomes like a Series A fund raise or long-term outcomes like legal risk mitigation. You don’t necessarily know or care what goes into the bundle to produce those outcomes.
By thinking of the full-stack startups as selling bundles of tasks, we are applying Netscape CEO Jim Barksdale’s famous quote, “there are only two ways to make money in business: One is to bundle; the other is unbundle.”
Frequency of usage, expectation of long-term relationship and ability to capture a large cut
Any successful business must nail engagement, retention and margin. For full-stack startups, the drivers are usage frequency, expectation of long-term relationship, and the ability to capture a large cut. In the table below I’ve evaluated full-stack startups in different industries against these three criteria from the perspective of early-stage startups as customers.
Most early-stage startups don’t frequently need a lawyer (my first startup went three years without a lawyer, relying purely on standard documents from Y Combinator), don’t always have head count to fill, and don’t often have anything newsworthy to share. This makes it difficult for full-stack startups in law, recruiting and public relations to service early-stage startups because they simply don’t have needs. Unlike a relatively low-priced SaaS subscription, if customers don’t frequently use a full-stack startup’s services, they’ll churn unless they are paying by performance.
Low usage frequency isn’t necessarily bad. It could work if customers have an expectation of a long-term relationship. As an ambitious company, eventually you’d want to retain a law firm, just like you’d want to always have an accountant keeping your books straight. These are powerful defaults that create a lot more room for full-stack startups in the legal and accounting space to get it right.
Finally, if your full-stack startup doesn’t have frequent usage or expectation of long-term relationship, then you better have a business model where you are able to capture a cut of a large transaction. For example, recruiters can take 20% of a hire’s first-year salary and real estate agents can take 3% of a transaction. That business model makes it possible to for full-stack startups in recruiting and real estate to thrive.
Quick lesson #1 from Upbeat: you can’t product your way out of low usage, retention expectation and transaction value
Upbeat was a full-stack, tech-enabled public relations agency I started that failed. Our startup customers had relatively low usage, no expectation of a long-term relationship, and low-priced transactions. To combat that, we hypothesized that we could build useful features to get customers to spend more time interacting with our software, which would get them to think about PR more and, in turn, demand more service from us. This turned out to be insufficient. While our customers spent more time in our product and and the product made them love our overall service even more, we couldn’t materially increase the frequency of their demand.
A full-stack startup has a higher chance of succeeding if customers perceive the service to require a lot of expertise. This is important because keeping customers satisfied is much more difficult as a full-stack startup. The outcomes you promise are sometimes not completely within your control, so you fail to deliver. Other times, your customers can simply feel like they are not properly serviced because customer demand a lot from a service, especially when compared to traditional software tools. Either way, when customers become disappointed, you will have to sell them on the value of the tasks you’ve accomplished for them, and you want those tasks to have high perceived value.
Take legal for example. Legal is the domain of lawyers. If you wanted to draft up your own legal document, you’d really hesitate because you don’t have legal expertise. The expertise commands a high dollar value for each task ($149 for incorporation docs on LegalZoom), and that dollar value will help a full-stack legal startup like Atrium justify the value of their services when customers become dissatisfied.
Compare this to a full-stack recruiting startup. Recruiting has high perceived difficulty but less in terms of expertise. Should a customer of a full-stack recruiting startup become dissatisfied, simply breaking down the completed tasks does not have the same effect because while it takes time to source, contact and qualify candidates, those tasks don’t require the same level of expertise and so the dollar values associated are lower. If it is possible for a customer to say “I’m just going to do this myself,” convincing them to stay becomes a lot harder. This is why TripleByte’s thought leadership on interviewing engineers is so important because it establishes them as unparalleled experts.
Linear relationship between tasks and short-term outcome
Assuming that you execute every task perfectly, is the short-term outcome guaranteed or is it still variable? Do your customers expect it to be variable or are you beholden to the outcome? The more linear the relationship, the better your chance of success because otherwise your full-stack startup would require much more human effort to maintain the customer relationship.
For example, chances are good that if you commit to visiting a personal trainer and doing the exercises, you are going to see short-term gains. If you slip on your diet or fail to go to the gym on your own, you are unlikely to blame your trainer. Personal training has a relatively linear relationship between completion of tasks and successful short-term outcome.
Compare personal training to visiting a therapist. Even if you visit the therapist regularly and complete all the exercises, your mental health may not improve within the time frame you are looking for. We simply don’t understand how our brain works as well as we do the rest of our body. Even if you do improve, there are many confounding variables outside of therapy that you may not attribute progress to your therapist. This attribution problem due to the lack of linearity can bring down full-stack startups because they have to invest significant time managing customer expectations throughout the engagement and fighting for proper attribution when they’re successful.
A way around this would be to absorb the non-linearity by changing the customer. Recursion Pharma has what they believe to be a better approach to drug discovery using machine learning, but convincing drug makers to use the software and be left with the attribution problem puts a ceiling on what they can capture, so it makes sense for them to go fuller stack and become a pharmaceutical company.
Limited variety of short-term outcomes
The smaller the variety of short-term outcomes, the easier it is for full-stack startups. Customers may have different expectations about what your service can do for them, and their expectations can change over time. You risk spreading yourself too thin trying to service a large variety of needs or limiting your market if you don’t.
For example, a startup may want Atrium to help with a Series A fund raise, and then later with a litigation. Compare this to accounting or recruiting for early-stage startups, there isn’t a large variety of things customers can ask for.
In the face of different kinds of potential outcomes and trying to grow without breaking, full-stack startups productize their services in order to scale. In Atrium’s case, they scale by referring their customers out to a vetted network of litigation lawyers. The complexity then lies in ensuring quality despite not having control of the customer experience, managing customer expectations about what can and cannot be done, and establishing a rhythm of working with customers in order to anticipate changes in their needs.
Touching back on having high perceived expertise. The more authoritative you are perceived, the more you can help set your customers’ expectations and advise them on what they should want next.
Ability to up-sell a long-term outcome
If the short-term outcome your full-stack startup provides naturally leads to long-term outcomes for the same customer, then you’d have higher customer lifetime value and an easier time building the business.
Early-stage startups typically can’t afford to think long-term. For example, a startup may hire a public relations agency to launch a product (short-term outcome), but doesn’t have a long-term outcome in mind. This is why PR agencies ask for 6–12 month contracts even though launching a product only takes 2–3 weeks (as we’ve proven hundreds of times at Upbeat). When there’s no long-term strategy to up-sell customers to, agencies have to milk the short-term arrangement for as much as they can. Similarly, you can work with a full-stack recruiter like Hired to hire a few engineers, but you have no intention to continue. This is why Hired sells a subscription and offers volume discounts to get startups to commit to using them more over time. Compare these examples to full-stack legal. Once you retain a law firm as your general counsel, you tend to stick with the firm over the long run as long as they can continue to find ways to help with your entire lifecycle of needs. You treat your general counsel like a primary care physician in that case.
If your full-stack startup is limited to only short-term outcomes, you can move up-market where the average contract value and potential for long-term outcomes are higher. This explains why Gigster pivoted to enterprise.
Quick lesson #2 from Upbeat: you can’t business model or account manage your way out of poor success attribution and lack of long-term orientation
At Upbeat, we suffered from not having a linear relationship between tasks and short-term outcomes. Even if we executed all the tasks perfectly, if the story we pitched didn’t resonate or if Facebook sucked up all the media attention with an announcement, we wouldn’t be able to convince any journalist to cover our story. We also didn’t control how journalists framed the story even if they did cover it. To combat this, we trained our agents to constantly remind our customers that PR, like sales, is ultimately about the number of shots on goal and the relationships you build during the process, but that wasn’t what customers wanted to hear and no amount of account management from our agents could’ve changed that. We even played with our business model and introduced an annual membership with discounts for more frequent usage, but the effects were incremental.
When we succeeded with our customers, that didn’t necessarily lead to us becoming our customers’ long-term PR partner. We required all of our customers to do an on-boarding strategy session with our agents so we can educate them and help them establish a longer-term horizon, and while we succeeded to a degree, it was too blunt of an instrument.
What’s the right level of “full-stack”?
It’s hard to know what is the right level of full-stack. Take the HR industry. You can purchase Gusto and Namely, which helps you run payroll, benefits, and basic HR tasks. If you prefer more service and less control, you can purchase JustWorks and ADP TotalSource for them to co-employ your employees and administer them on your behalf. If you want more service beyond HR, you can buy Rippling, which helps with IT administration and access management in addition to HR. All of these startups offer different outcome bundles and they all have found growth for now. In Rippling’s fundraising memo, they acknowledged that their bundle is “weird” (because customers are buying anyway, so doesn’t matter):
The jury is still out for the HR market, and at the end they may simply appeal to different segments of customers. What customers are willing to pay for changes depending on the best practices of the day and you should look for emergent opportunities for new bundled and unbundled opportunities. As startups are pressured to grow faster and faster, they become more willing to trade control and higher monthly cost for lower upfront cost in order to move faster. Someday, startups will need an in-house counsel, an in-house accounting department, host and manage their own data centers etc., but for now, early-stage startups are willing to hire all kinds of full-stack services to simply do it for them.
Should you build a full-stack startup?
The decision to build a full-stack startup is often based on circumstance. Maybe you operate in an old-school industry that is rigged to the incumbent so selling software is not an option (42Floors, Uber). Maybe you already know there’s an attribution problem even if you were to sell your software that it becomes difficult for you to capture value (Recursion Pharma). Maybe you think the incentives are so misaligned that you have to re-build the entire stack (Atrium, Upbeat). All these are valid reasons to start your ambitious journey, but as early as possible, break apart the bundle to examine the tasks, short-term and long-term outcomes, and use them to evaluate your service against these seven factors:
- Frequent usage (e.g. accountant bookkeeping)
- Expectation of long-term relationship (e.g. general counsel)
- Ability to take a cut of a large transaction (e.g. recruiters take 20% of first-year salary)
- High perceived expertise (e.g. lawyers and doctors)
- Linear relationship between tasks and short-term outcome (e.g. personal trainer)
- Limited variety of short-term outcomes (e.g. recruiting full-stack, front-end, back-end engineers, but not graphic artists)
- Ability to up-sell a long-term outcome (e.g. accountant becoming long-term CFO)
Click here to play with the full spreadsheet. I’ll leave you with some inspiration from Garry Tan (quoting me quoting TechCrunch quoting Kulveer Taggar of Zeus Living 😜).
Thanks for reading! If you like this post, you might like my previous post on this topic, The “Full-stack startups” and Jiro’s Dreams of Sushi. If you’d like to chat more about this, hit me up on Twitter.
 There might also be a generational shift. As products have gotten better and better compared to pure services, younger founders are more likely to hire a full-stack law startup like Atrium that packages their services like a product than searching and working with a law firm the traditional way.