Many successful software product startups begin life as already successful business service providers. They start out as teams of experienced human beings, paid by the hour to do complex and difficult things.
Let’s call it Software from a Service — SfaS.
Once service providers reach the expert level in executing their service and begin to refine their methods and approach, they often conceptualize a mostly automated version of the perfect execution of that service: Trimming down to the most valuable parts of the offering, delivering those parts in the most efficient and cost-effective manner, and relying on the least amount of human interaction possible.
6 Questions to Ask Before You Press the Automation Button
- How do we find the most-ready market?
- How viable is the automation?
- How valuable is the automation?
- How expensive is the automation?
- Can we develop the UX?
- How much manumation can we do?
This allows them to develop a narrower – but still robust – core version of their service offering, aided by automation via technology, as a standalone software product.
In fact, with all the recent advancements in technology, including no-code and even AI, this type of automation makes it possible for a growing number of vendors to turn more types of services into even more DIY software products.
The problem is, once the service providers get their ideas together and maybe some of the technical components sketched out, they realize that the development of those software components will result in a lot of expensive work. Then the venture becomes somewhat of a gamble: How sure are you that the targeted industry is ready for a totally do-it-yourself version of the service?
Expensive development without corresponding demand often leads to expensive failures. So how do you decide what to automate and when?
Where Do You Start?
The strategy for turning a service into software isn’t like developing a net-new software product. Well, I mean, the endgame is the same, but the development doesn’t follow the same playbook that a brand-new software solution might. You get to skip ahead a bit for three reasons. You already know the problem, intimately. You know the business needs of the customer. You know the key benefits that the product should offer to address those needs.
There are essentially two things you don’t know: Who is going to buy a do-it-yourself version of an established do-it-for-me service? What does the automation roadmap look like?
To answer the whole question, you need to get the first part absolutely right.
How Do You Find Your Most-Ready Market?
You’ll need to segment a market out from your existing market, and again, this is much easier than trying to identify a brand-new market for a brand-new product. You simply analyze your current target market, and from that you develop a new ideal customer profile.
The key to successfully automating a service is being able to offer most of the value of the manual service, let’s say 80 percent as an example, for a much lower cost, let’s use 20 percent.
I love a good 80/20.
Your ideal customer is going to be one who can live with the 80 percent of the value of the manual service, and would easily trade the remaining 20 percent of value for the savings they get at 20 percent of the cost. Ideally, you might look for new customers who could now enter the market for your services because of that cost reduction.
I’ll give you an example. At Automated Insights we were the first publicly available generative AI engine, and we worked mainly in sports. We started covering all the pro and college sports teams with our automated content, but we quickly realized that they were not our ideal customer. Then we moved to Yahoo and NFL fantasy football, who fell in love with the idea of 80 percent of a traditional game recap for their fantasy players for something like 0.0001 percent of the cost of theoretically having humans write those 13 million recaps.
80/20 is great. When you can go 80/0.0001, you’re doing the heretofore impossible, so to speak. But believe me, 80/20 is more than good enough.
That said, the market you choose is critical. Even if you automate everything, even if you automate it well, and even if that automation can serve your current service target market with the automation, don’t start there. You want to own the new market you’re creating and perfect the solution for them before you try to convert an entire existing market, and go for the billion-dollar play.
How Viable Is the Automation?
A word of warning. Not everything should be automated.
I used to say this all the time at Automated Insights. Could we automate player quotes in our post-game recaps? Possibly, potentially, maybe by finding the post-game audio or whatever and using speech-to-text and so on.
Would it be worth it? No. Would the customers in our new target market miss that? Absolutely not, not at the price point we offered.
Teaching Startup — my project to bring startup advice to all entrepreneurs by providing expert answers to real entrepreneur questions at a fraction of the cost — is another example, with different reasoning. I will try to automate everything except the answers to the questions themselves. No ChatGPT allowed. Because that’s where the true value is and automation would destroy that value. Instead, I’ll automate everything else.
So keep this rule in mind: If automating a part of the service doesn’t make sense or wipes out all the value or doesn’t significantly decrease your costs, eliminate that part of the service altogether and then determine whether or not your core product is still viable.
Now let’s play those 80/20 percentages.
How Valuable Is the Automation?
In other words, how much time and money are you saving yourself, and thus, your customer, by automating a component of the service? Once you figure that out, start developing the components that offer the biggest time and money savings.
This is about increasing the X percent of value of the product towards that 80 percent.
A lot of the time, the largest potential value conversions lie in the margins, the logistics and the outliers. For example, the entire streaming video model is based on the elimination of physical media to distribute the final product. Netflix changed its whole business to switch to this model.
On the flip side, if an automation doesn’t save a ton of time or money, deprioritize it for now.
How Expensive Is the Automation?
I can’t stress enough how important it is to build the quick, inexpensive automations first as you prove the viability of this product. And by viability, I mean “does it deserve to exist” in this new market. This is how you avoid making expensive mistakes.
This is about decreasing the Y percent of the cost of the traditional manual service towards that 20 percent. Now you’re getting closer to your 80/20.
Can You Develop the UX?
There are two ways to test your new product’s viability in the new market.
The first is to use these automated components, as you create them, with your current customer base, and test them as you would expect them to use those components in an automated self-service scenario. This will help you get closer to automating the existing user experience.
But more importantly, you’ll want to test the viability of an automated, self-service, DIY product with your new target market. This means the product doesn’t start out as just a bunch of disconnected automated parts of the service, but takes them all the way from the beginning of your service, with all the options and flows, to the successful end of your service, with the reporting you need to prove ROI of your product to your customer.
And then everything you can’t throw in there as an automated part of the service, you manumate.
How Much Can You Manumate?
Manumation is the process of manually executing what looks to the customer like an automation. I gave this example years ago when talking about how Uber might have been manumated to prove viability:
The app starts out as just a button to get a ride. When the user presses that button, it sends a message with their location to Uber Central, maybe as a text or converted to Slack. Then someone at Uber Central picks up the phone and calls one of their friends to go pick up that user and take them where they need to go.
This is manual, it’s expensive, it’s slow sometimes and breaks sometimes. But it doesn’t look like that to the customer. The customer just sees and uses an app that looks like it’s all automated. It’s a minimum viable product in its most minimum sense.
Little mistakes are okay. The question is how much business can you take on and how long can you do this before you make really big and expensive mistakes?
Once you’re at this point, you can release an MVP to your new market, then keep increasing value, cutting time and money costs, staying upright and approaching 80/20 — the robust core of your new DIY software product.