It's even harder to be open and honest with yourself, your employees, your customers and everyone else when shutting down. Perhaps somewhat weird, but I find it refreshing to see posts like these. I think it's the sign of a good and healthy entrepreneur/team (albeit hard to write).
I'm sure you and your team learned a lot, and I love the gesture to open source your software.
Scheduling is hard too. Every company has different requirements around how their scheduling works. Lots of businesses need their employees to clock-in and -out of their assigned shifts for accountability reasons, not sure if Staffjoy supported that. The shifts can have requirements too, e.g. the maintenance shift can only be scheduled for 8 and 12 hours, but nothing longer/shorter. There are usually payroll and billing implications, so adapting to an existing payroll/billing system would be an additional challenge, and not doing so would be a non-starter for lots of businesses.
If anything, the product surface area they were trying to tackle is really broad. Probably could have focused on a particular vertical and provided more value-adds.
I'm excited to see what gets open-sourced. Schedule visualizations are also hard. :-)
We looked at the Microsoft tool [1]. I honestly think it's more of an upsell play for existing customer than a new customer acquisition play. The pricing is confusing and not competitive [2].
I also tried a startup which was largely premised on having better scheduling, and solving obvious operational inefficiencies with competitor businesses.
In hindsight, that was dumb, because we never got to anything near the scale where the scheduling would matter.
We didn't have a better way of acquiring customers, so efficiency never came into play. Conversely, if you have enough customers, operational efficiency doesn't matter that much.
Anyway, no profound insights here... just pointing out to people that if you have a great startup idea based on using a more elaborate scheduling algorithm than your competitors (an attractive trap for programmers, no doubt), then think again.
Business execs are often taught to distinguish between cost centers and profit centers, and to focus their attention and resources on the profit centers.
Any product that's premised on making a company's cost centers (e.g. HR scheduling) marginally more efficient is going to see very slow adoption just because it's not something execs will focus on -- they have so many other priorities that are higher on the list.
Also, SMBs almost never want to convert to pay cash for anything because they are usually so cash crunched.
Yep, it totally makes sense when put in those terms.
Also, any attempt to make the cost centers more efficient (by introducing elaborate software and algorithms) will probably make them quite a bit less flexible, require higher skilled staff, etc.
I sincerely feel your pain, we also had to fight old paper and excel habits with a digital tool. I am also in a startup working on a Scheduling product with math optimizations (absence management, work constraints, IBM Cplex solver)...! I will be happy to review your open source code and see if anything can be reused on our side. PetalMD has a freemium approach in a niche market (Canadian healthcare providers), and we burned lots of money before we finally focused on the core value for our user base. The freemium gave us the critical mass and leverage to have discussions with partners and we learned through the years what had to be enhanced in our product to offer several pricing plans.
Our niche is really helping focusing the product enhancements and the marketing efforts. Scheduling needs are broad and we constantly have to refuse opportunities.
I use CPLEX at work. I've also been told that Gurobi has a lot of people that left CPLEX and is generally up for trying more advanced albeit risky advancements.
Thank you for the pragmatic writeup rather than the usual "fantastic journey" blather. Also, returning investor money, rather than wasting it, demonstrates leadership.
Had this discussion with the investors. I plan on exploring a different idea in a different space with a new founding team. The investors suggested restarting the corporate structure, and some have committed to reinvesting the money into the new company.
This isn't the case, and there is no newco at this time. Unfortunately, I had to lay off all of employees today. In addition, we are liquidating any IP, closing customer accounts, paying out our cash, and dissolving the corporation.
HR is a hard space. Sorry you had to go through all of the downside of a business not working out. It probably means next to nothing today but this experience will make you more successful in you future endeavors.
This is how your side would frame the issue. The opposing side would do their best to portray events as if you and the investor had identified a billion-dollar idea, and done shady manouvering to clean the cap table.
Agree with this. Question, it looks like this company just raised a seed round in January 2017 according to crunchbase. How can you raise a seed round in one month and then completely shut down the next?
We closed the round in July of last year, prior to starting the V2. We sat on the announcement until we needed a press boost on a product announcement.
Gotcha, thanks for the clarification! As an entrepreneur myself, greatly respect the decision to shut down and return money to investors instead of just spending until theres nothing. Good luck in your future endeavors!
I'm moving onto a new idea in a different space. With our lawyers and investors, we decided that a restart was a better idea than a pivot for a variety of reasons.
What percentage was given back to the VCs? (just curious)
I, too, found the post easy to read and straight-forward. If there was ever a great way to answer the hiring question of "Tell me about a time where you failed and what you did about it?" it would be this. Good luck in your next gig!
Thanks for committing to open source your code! Nobody else has highlighted this in the comments yet, but this helps ensure the money spent on your company didn't end up wasted: At the end of it you will be leaving behind a valuable contribution to the open source community.
Thanks, most of the code for V1 and V2 should be online by the end of the week. Two of the repos use applied math libraries that cannot be open-sourced. For one, I think I can drop in an open-source alternative. The other I may not be able to open source.
I'll also be publishing our YCF application and our pitch decks in the coming days.
Is it possible for the library simply be removed, and the code posted with the notice that they won't work without either the proprietary library or some code work to swap it out?
I like to believe all open source code has value to someone, even if it's someone stumbling through working with said proprietary library, and trying to find something else that connects to it. ;)
The pre-V1 algorithms used the Julia JuMP package [1], which allows swapping of the backend solver easily. I don't think some of the functional tests would pass on the open source solver though (precutting and heuristics in gurobi caused a huge speed gain).
The later iteration of the algorithm used the Gurobi python library directly, which is closed source.
would love to see the solver application (if you are planning to swap an open source solver for gurobi). These class of problems are something I have wondered about, but never truly tried ny hand at.
I use WhenIWork for this area, which suits our needs. Not sure what would make me switch, probably not a text messaging. Most of my employees have smartphones, so a mobile app is fine.
Biggest problem for me (as a business owner) is managing schedules to a budget, and making sure we are being as efficient as possible with our staffing spend.
Most tools I've seen have a pretty weak budgeting feature. (You just put your budget in and it tells you how you did)
I know some companies are using ML to do this internally - predicting staffing needs, optimizing costs, etc.
It strikes me as a very individualized problem; might be hard to generalize - or the markets you could generalize too aren't that big/don't need a solution all that much.
ML is more statistical - good for forecasting. We were doing optimization mathematics with hard constraints for the scheduling.
One of the issues that we encountered was that a generalized algorithm was tough. Different companies had different requirements for an algorithm, and it started to feel like a consulting company. For instance, if a big box retailer would rather be slightly underscheduled at peak time rather than adding additional employees (and being over-scheduled near the peak time). A high-end boutique retailer (like a diamond shop) might be the opposite - each sales brings in so much money that they overstaff at non-peak hours so that they can service customers during peak hours.
Above is a nice example. There has been some nice contests with the above problem and the ILP solvers work extremely fast and great and solve them to optimality. Although Staffjoy constraints might have been more general.
Either way you could easily attack any custom problem with an ILP solver. It depends how long it would take to get a feasible solution and then how long to minimize the costs or fit the budget.
In order to speed up the solver you might use ML but that would require previous data. Probably the only way to speed up the solver is to learn it through reinforcement learning on a batch of data. Takes time and time and time. Not to mention that your ILP solver has to be equipped to merge with any ML machinery you are using.
In the old days, everything was a hard constraint. However, free trial users don't always put in problems that are solvable. I would get paged at 4AM about infeasible models, only to discover things like "this employee has zero availability but a minimum of 40 hours per week of work".
So, we turned into more of a "scoring" system where all minimums were soft constraints (e.g. minimum hours per week) whose violations caused a large point decrease [1]. Maximums were a hard constraint. The system made it impossible to input an infeasible model, and worked fairly well.
Splitting the scheduling problem into separate steps (forecast->unassigned shifts, then unassigned->assigned shifts) also sped up the algorithms (and allowed for people to create unassigned shift templates).
But deciding how many nurses to have at what time, based on the probability of how many and what kind of patients and the likelihood of a good medical outcome is a different problem.
It is, but saying it's machine learning hides the combinatorial issues and combinatorial search that has to happen. Machine learning model can't replace the necessity for search.
Scheduling problem (the general one) maps easily to the vehicle routing problem. Vehicles are routed and service customers. On-demand requirements and scheduling are probably easier to conceptualize in that framework (think Uber but with ride-sharing bus sized cars, or team picking up and delivering food from restaurants to locations, or repairmen doing stuff at people homes).
All needs combinatorial search and machine learning is only one little piece of the puzzle.
I usually say it the other way -- optimization is just one part of machine learning.
ML includes optimization/search, dimensionality reduction (aka unsupervised learning), prediction (aka supervised learning), and reinforcement learning. And I'm probably forgetting a category.
ML does not include integer linear programming (nothing learnable there) and similar mathematical optimizations. ML can be a part of it but is not integral to it. Just like finding the shortest paths between two points does not include ML. ML can be used to learn the weights through time (traffic sensitive routing) but the search part is separated from it.
IMO, ML is nowhere close to useful in these problems when most of the clients just use pen and paper. It should be fairly easy to beat that with some simple search heuristics.
But once you get into the modeling of how many customers come in at what hour and how likely they are to purchase how much conditioned on the staff present, it's a learning problem.
It's a pipeline. Check out this [1] paper about the Taco Bell scheduling system. It's forecasting (stats, but would probably be called ML if paper wasn't in 1998) -> queuing theory (markov chains, sometimes formula-based, sometimes simulated) -> scheduling (normally mixed integer linear programming)
Eh, what the heck, with enough layers (and enough data) a neural network will eventually create a neuron for each of those tasks. Just let it train for a few years.
"small businesses still use paper or spreadsheets"
that's usually your first red flag. when management starts going after this market then things aren't looking good. low margins, sales dominated org if you're lucky can keep you afloat for a bit. long work days etc. avoid.
the real money is always in enterprise. liability is a huge issue. ain't no huge corp going to throw down millions on some random scheduling software run by a couple of dudes.
in the rare event that it does happened, raise as much money as possible and hire as quickly as possible. reduces risk and your work load.
they just didn't have the capital to compete. the market is there though clearly they just ran out of dough.
I'm currently working on a product with a team and we did a bit of exploration for an idea for SMB's and had that same issue. We were talking to someone who on the surface would be a perfect fit for the product but then we asked how they handled organizing a job. They reached down into their bag, pulled out a file folder, opened it up and spread a few sheets of paper in front of us and went "will this page is this thing, this one is the other thing, and that one is another thing". An entire jobs worth of documentation and info right there laid out in front of us all at once in 3 seconds flat.
That was the moment we realized the reason our target industry was still using paper and spreadsheets. Because they're not just "good enough". In most cases, they're the best possible option. Maybe with enough iterating and testing we could've come up with a UX that could've put all that information in front of 4 people in one shot that quickly. But the odds are against it and even if we did it would've taken a long time to get there.
A lot of startup founders view "legacy" as a four letter word. But sometimes it's simply the best course. Not everything needs a high tech solution. And accepting that can be really tough sometimes.
Also spreadsheets (e.g. Excel) in the hands of a smart user are very powerful. They are incredibly flexible and will handle most data management tasks for a small/mid size organization. And with Google Sheets (which has gotten a lot better in the past couple of years) or other cloud storage, the fear of losing everything with a failed hardrive or lost laptop is gone too.
I run a small nonprofit. I get emails at least weekly from some little tech company that has an app for some aspect of my operations. I usually ignore them, but when I do answer it's always some variation of "I can do everything I need for free with Google tools." And to myself I wonder why these companies are even targeting small organizations that have almost no money for technology.
You are forgetting enterprise space is a very capita/resource intensive field with very high CAC where lot of companies end up spending $2 to make $1. Part of the reason why you can't be a garage startup and expect to close a six digit contract after 1.5 years of trying to close it.
I wouldn't touch enterprise until you've hit some maturity with lower tier markets which are easier to sell to.
Looks like he is retaining at least some of the investors for his next attempt. [0] I'm sure giving them some of their money back made that discussion a lot easier.
Because the investors gave money to StaffJoy, not the idea one founder had when it because apparent SJ wasn't working out. And it sounds like either the other cofounder(s) either aren't interested in the new idea or aren't welcome to the new team as he says there will be a new founding team. That makes it very complicated especially if it's the same corporate structure moving forward with a different idea, you don't want un-involved prior cofounders sitting on equity.
So they're giving the money back, wiping out the corporate structure and starting fresh with one of the cofounders and what sounds like some but not all of the investors.
Really sorry to see this; your candor is refreshing, and I've been interested in the optimization work you've been doing for some time. Best of luck with what comes next.
I think you're right and this isn't the first time for me, there's been dozens of times a startup is announcing they are closing here on HN and it's the first time I have heard of them. Not a criticism of Staffjoy of course, it's just a general truth.
This is superficial, but what is the successrate of startups that end in -joy? Homejoy and Staffjoy closed shop, are there others that closed or survived?
"The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man." --George Bernard Shaw
Many workers didn't have access to email, or they were wary of providing it to an employer. Without smart phones, they lacked access to apps. So basically, they had no way to register for an account and see their schedule on other applications.
During user studies, we saw that restaurants would still print out their schedules, and managers would take a photo and text it to employees.
This created problems when the workers treated this channel as two-way. For example, workers running late would text the manager, and the manager would not notice the message promptly while working.
One of the points mentioned often by @sama is that founders give up too quickly. What advice did you get from your YC Partner on this conundrum of being cockroaches versus shutting down?
From the outside, it seems like the founder saw a bigger opportunity and decided to move on.
So, the primary motivation isn't shutting down this startup. The primary motivation is moving on to the next startup which they believe is a bigger opportunity.
Not too awesome for current employees: hope they are getting taken care of.
So they launched right after the holidays in the slowest time possible and are shutting it down only after a month? I think we don't have the full story here. It looks like a panic either on the investor's side or startup side, or both.
We did outbound for multiple months, and were feature-complete on the product in November. Our first users on the V2 came onboard in early December.
Based on our user research, our primary markets (service and retail industries) freeze all of their software during the holidays. During the lull in January, they evaluate new purchases.
Launch is basically a one-time bump in inbounds that doesn't scale, so we timed it when we thought it would have greatest effect. However, we had been doing more scalable outbound - e.g. emails, canvassing neighborhoods with flyers, etc - for months.
Conversions to signups were remarkably low, and conversions from signups to paid were also low. We basically realized that we were not solving a core burning problem for these businesses. That's more what contributed to the shutdown.
I was a partner in a startup a few years ago doing roughly the same thing. After working on an early prototype, the project fizzled out due to involvement in other projects, but a competitor launched with the same idea soon after that called shiftplanning.com. They did literally the exact same thing except they got to market before us and they (from what I can tell), made it big. The market for this type of software is massive but obviously you need to find some key business up front. Maybe you had the wrong investors? At the time we had an investor that wanted to put money in AND have his large restaurant chain start using it; a strategic partner... Something to think about...
What startup founders don't realise sometimes is that marketing is more important than having the best product.
I've had the luck of being on both development and marketing ends and high quality marketing can definitely sell a product that's slightly inferior to the competition 10:1
If you need big names attached to your product early the best way is to give it out for free to those first strategic customers. Why would anyone turn down something useful and free.
If you can't tell why another startup beat you out when you had a better product it's garaunteed to be marketing related.
Aw man, I run a live English tutoring operation overseas, with real human tutors working shifts in my app. This could've been really handy (since Upwork didn't really have a nice way of tracking time on mobile, and we just didn't have the dev bandwidth to build something like your product ourselves).
Did you take a step back and evaluate marketing strategies? I ask because my last company marketed a product to small businesses and it took us about 4 months and significant churn before we saw any positive returns in our marketing efforts. If you have that much money left it seems premature to throw in the towel without redoing your marketing strategies a few times.
In my experience spamming inboxes and flyers were a fool's errand compared to the marketing power of tageted online ads and customer referrals(offering free month for referring a new customer or similar). Non targeted "spamvertising" makes your company look cheap and desperate before people even try your product.
I expect to find a flyer illegally placed under my windshield wiper or jammed in my front door for a maintinence dude or a smoke shop, but not for any business software I would seriously consider using.
Do you have metrics on how many emails were actually delivered? Not how many the mail server reports, but which email services were actually forwarding your mail to places besides the spam folder? Where did you get your email lists? AFAIK, renting email lists is largely considered a joke in today's marketing circles.
Did you try offering a free trial period to start out? Target users at company domains over those @ Gmail or Hotmail? It's easy to code something that offers free trials only to people without a personal email address. How about sending mail directly to the target companies address to their hr manager? What about old fashioned cold calls? Sometimes they work better than you would think. Did you reach out to companies that expressed interest but didn't convert to paying users?
Did you try visiting local businesses in person to pitch your ideas and get feedback? Some business owners will be fairly helpful saying "yeah this sounds great but when I go to your site it doesn't look like we need this at all". Or "can I sign up my employees without assigning an email address to each one?" " I just want you to make the page printer friendly so I can post the schedule on the door like we've been doing for 20 years"
Was there A/B testing of all marketing materials including emails, flyers, page text, ad text, value proposition (free first month, 10% discount for year subscription, free use for less than 5 people)? Did you try getting individual franchisees to sign up then go to their main corporate office and talk about how much it's helping some of their restaurants?
Did you go to local chamber of commerce and small business leader meetups? Not start-up meetups, regular SMB meetups. This is generally how small business meet each other and is a perfect place for local vendors to get a foot in the door to the local B2B market. Even with the internet a ton of deals are made the old fashioned way between local partners. Dinner and a handshake is key to getting a lot of local businesses.
Why did it take until V2 to know you would need to use text messages? I used to work in the service industry for years and it's immediately obvious to me that smartphones cannot be a requirement. Did you think everyone working min wage as a greeter at Walmart when they're 60 would even know how to use a smartphone? This seems like a pretty glaring lack of market research.
I disagree that launch is a one time bump and that timing is of any importance. I find it rediculous that you ran advertising of any sort before the product was officially launched. Anyone checking out your site would do it the same day they saw the ad and discard you immediately if the product wasn't available. I can't think of a single start-up that would be materially affected by the time of year they decided to go live. Maybe if you're mowing lawns I guess?
It sounds like your market research amounted to asking people and googling things on the internet when this should have been done by getting hard numbers, aka trying to sell the product rather than predict something without data. I can tell you from working with numerous small business clients that low cost items (less than around 400 a month) are purchased at any time of year by lower level managers with little oversight. In the grand scheme of things the monetary investment is low enough that the companies are unlikely to care about budget whatsoever. We worked with small businesses of all kinds and this is common across any industries I've seen.
Not to completely trash you guys, but announcement says little about the actual cause of failure besides "yeah marketing didn't work good". It also mentions considering pivoting with different co-founders!! As if the other founders were part of the problem. That's a pretty scandalous thing to say about the other guys that poured blood and sweat into this with you for over a year.
That, combined with giving up with a mil in the bank and the, IMO, bad marketing blunders makes me think this shutdown is more an issue with co founders not getting along than anything else.
You could have easily scaled back operations to make that million last several years to give you some time to work out your conversion rates. Giving up when you've already built the project because you are having trouble selling it seems like a poor use of the rest of your investors money. Selling an existing product is far easier than building one that doesn't exist.
The holidays is a great time to launch an employee scheduling application... that's usually the busiest time of the year for retail operations and businesses that depend heavily on employee shifts.
It's even harder to be open and honest with yourself, your employees, your customers and everyone else when shutting down. Perhaps somewhat weird, but I find it refreshing to see posts like these. I think it's the sign of a good and healthy entrepreneur/team (albeit hard to write).
I'm sure you and your team learned a lot, and I love the gesture to open source your software.
Best of luck on the next step!