Wednesday, 23 May 2018
This is a special report on our business model.
Recently I’ve been reminded of the lesson I learned after my first company failed, which is: don’t work hard to avoid making hard decisions. If Invisible fails it won’t be because I didn’t work hard enough, it will be because I didn’t get the model right. And that is why we’ve just made a series of decisions that mechanically prevent operating losses (prevent us from losing money), and increase our margins (make us more money). I’ll review them in order of significance.
The TLDR is that we’re moving to a zero fixed-costs model (or as close to it as we can get), and using automation to increase our margins.
Decision 1: Don’t Pay Agents Cash For Non-Billable Hours
On Monday we rolled out a severe decision that I’ve been putting off since as early as January. If I had not delayed it would have already saved the company over $100,000.
The decision is simple: only pay agents for work done for clients. See? Simple. Cash for billable hours. No cash for non-billable hours. That’s it.
For non-billable hours, we’re going to pay agents in equity instead of cash. I’ve allocated an option pool of 1M shares. Agents earn points in a spreadsheet proportional to their USD rates, and the sum of these points represents the entire pool.
The problem is that up to this point, we’ve been paying agents not just for billable hours, but for non-billable hours too. That means that all the time we’ve spent on management and training has made the company unprofitable. Of the $68,000 spent on agents in April, $44,000 of it was non-billable. From now on, that $44,000 won’t be paid in cash, but in equity.
Last month we began calculating our operational efficiency. We were disappointed to discover that it was as low as 30%! Any number below 50% is gross margin negative. Any number below 100% means that we’re not realizing our initial gross margins, which, on paper, are 50%. That means that growth is costing us money.
The focus in February, March and April was on effectiveness, not efficiency — as it should have been. We had to prove that we could do this: that we could deliver fast and high quality work to our clients, make them happy and keep them delegating more and more work to us. After all, there’s no point making a service efficient if it isn’t first effective. Efficiency comes later.
Now that we’ve proven that we can retain and upsell clients, the focus is shifting to efficiency and growth. I focused on growth in my April report. But didn’t discuss efficiency at great length.
If we were burning money every month, how did we hope to solve this problem? The plan was to get operational efficiency up to 50% by June and up to 90% by the end of the year through a combination of automation and coordination technologies, plus good ol’ fashion management, training, and operations tactics.
But the faster we grew, the faster we burned — so even if that plan worked flawlessly, it wouldn’t work fast enough to prevent us from running out of money in the short term. “Wat do?!” Raise money — yes, I tried. It turns out that it’s easier to do when they need you rather than when you need them.
So we needed a drastic solution that would be immediately effective in solving this problem. Technology is a long-term fix. Operations is a long-term fix. Financial innovation was the only way.
Decision 2: Dynamically Adjust Pay Rates To Achieve Target Margins
On paper, our initial gross margins before automation are 50% on the Process Line. Clients pay us $10/hr and we pay agents $5/hr, so we make $5/hr in gross profits, before volume discounts.
That’s how it’s supposed to work, in theory. The problem is that, in practice, our average agent pay has crept up to as high as $5.60/hr. We start hiring agents at $1.50/hr and pay them up to $8/hr as they learn new skills and handle more complex work. We target an average of less than $5/hr. But there are situations in which more senior agents are required than junior agents, and the average overshoots the target.
As clients spend more money with us, we slowly give away our margins. For a client that buys $20,000 worth of hours, we currently give away our initial gross margins and charge them only $5/hr. In the future, we’ll raise the threshold even higher. But we’ll always provide volume discounts because we want scale.
So even with 100% operational efficiency (see Decision 1), there may be situations in which we are still losing money, say 60¢ per hour. In those situations, moving forward we’re going to dynamically adjust pay rates to achieve our target margins.
Decision 1 + Decision 2: “The Agent Model”
“The Partner Model” (inv.tech/partners) solves the problem: “Without raising millions of dollars in venture capital up front, how do you hire all the talent that you’re going to need to build this company?”. In the same way, “The Agent Model” solves the problem: “Without raising more capital up front, how do you achieve operational efficiency fast enough to sustain an agent work force?”
Both models rely on aligning incentives with equity. Far from being unfair, this feels right to me. It’s neo-neo-Marxist. The workers now own the factory.
Here’s the speech I gave to Agents on Monday. Amazingly, only one agent has quit so far as a result of this change. We have a loyal workforce.
Alignment is not the only principle invoked in these decisions. The other principle is Flexibility. If your revenues are dynamic, your costs must be dynamic. If your demand is dynamic, your supply must be dynamic. Without that, you die unless you are magically 100% efficient, or you have the capital to buy you the time to become magically 100% efficient.
Decision 3: Pause $10K Starting Bonuses To Partners
The Partner Model (see inv.tech/partners) ties salaries to gross profits, in theory. But in practice, it assumes that our gross profits are always 50%. Even with Decisions 1 and 2, gross profits will be less than 50%, because of volume discounts, and because of Premium Services (The Assistant Line, The Specialist Line and The Strategist Line), which have gross margins higher than 25% but less than 50%.
This means that investor dollars are partially subsidizing our Partner Model. And that’s O.K. It’s still way, way better than the typical startup employee compensation model, under which we’d be burning almost $200K dollars per month (30 employees at $75K/yr), if not more. Instead, we’re talking about a subsidy of no more than $5K last month. No big deal. In fact, it feels totally reasonable, given the risk we’re asking partners to take: which is that we’re going to grow revenues to $400K+/m within the next 6 to 12 months.
But we’ve also been offering $10K starting bonuses to partners, which we pay out in $2K increments over their first five months. This is also a reasonable incentive to offer, and still, The Partner Model is on-balance the most company-favorable incentive structure I’ve ever seen.
That being said, it’s not reasonable to offer $10K starting bonuses when you have less than 6 months of runway. Any new partnership offers we make until then will include a delayed bonus. The contracts will be written in such a way that these bonuses are not liabilities, but conditional rewards.
Effective immediately, we’ve paused paying out bonuses to existing partners until we have more than 3 months of runway, and then we’ll resume, as long as we can maintain at least 3 months of runway.
To grow from 10 partners to 29 partners cost us $190,000 in bonuses, so that alone shortened our runway by 4 or more months. We plan on growing the partnership from 29 partners today to 40 partners by the end of the year. So that’s another $110,000 of bonuses, which we now can delay until we’ve raised or earned enough money to have at least 6 months of runway left over after we pay them.
Decision 4: Increase Margins With Automation
We’ve begun to automate our highest volume processes. On a recent 3000 hour project, we built scripts to reduce the work down to 50 hours. This should increase our margins. But how should we price for it?
Moving forward, we’re going to provide manual speed benchmarks for all of our processes. So you’ll see how long it took us to execute a process the first time we did it.
For every 1 hour that we save by increasing operational efficiency or through automation, we’re going to give the client back 30 minutes as cost savings, and we’re going to keep 30 minutes as a margin increase. This aligns incentives.
So now we’re seeing margins as high as 85% on some processes.
Decision 5: Sell Products
Of course, if we fully automate a process, we should sell it as a software product. We haven’t fully automated any processes yet, so we have no software products.
But we do have data products. We work with a lot of proprietary client data that we don’t host and of course cannot resell. But we also generate a lot of data on behalf of our clients, which we own. For example, we have lists of every Node engineer in NYC and SF. These lists are re-sellable, because they save thousands of hours of lead generation work. The margins on these lists are 100%, because there are no COGS. So we’re going to start selling them as data products.
Have We Removed The Possibility Of Defeat?
“Don’t worry about victory, remove the possibility of defeat!” That’s my paraphrase of a strategic principle in Sun Tzu, and, again, it’s the primary lesson I learned after my last startup failed. This mirrors the primary strategic teaching of Machiavelli, which I paraphrase as “Don’t build your power base on that of another” — in other words, minimize existential dependencies.
Assuming we raise at least $250,000, there are very few core ways we can die, now that we’ve made these decisions. I say “core” ways to exclude Black Swan events (e.g. Key Man Risk — I get run over by a bus; Bad Decision Risk — I decide to do something really, really stupid), and to focus on core game-theory (vs. fringe game-theory).
Again, assuming we raise at least $250,000, the only core risk in the business is scaling risk. On the supply side, we’re confident that Digital Assembly Line 2.0 will take us from 100 clients to 1000 and beyond — we’ve got a deep roadmap and a strong handle on the design problems involved. So as insiders, we’re not worried about that scaling risk — either operationally or technologically. We’ve got strong systems in place already and we’re building the architecture we need to scale them.
On the demand side, if we don’t get to $400,000 in monthly revenue by December, there’s a risk that partners leave because they can’t afford such small salaries. That risk dramatically decreases at any number above $240,000 in monthly revenue — progress buys patience. Above $400,000 the company generates net profits (assuming 40 partners with salary caps averaging $6K/m) and existential risk goes to zero (profits = you can’t die). Above $100,000 in revenue the company generates enough gross profits to cover any reasonable operating expenses.
I don’t see core risk in competition or in any other core business metric. The real question is can we acquire clients for $250 or less? In other words, can we find a deep, repeatable sales channel? If the answer is yes, then by the end of the year we’ll no longer be a startup, we’ll be a scale-up.
The core risk in this core risk (the meta-core risk), is whether we can build an extraordinary sales and marketing team, by hiring 11 more great partners, with the three main roles filled ideally by the end of July. With the right sales and marketing team, we’ll be able to explore and exploit multiple sales channels while building an epic brand with epic marketing materials.
The core risk in this meta-core risk (the meta-meta core risk), is whether we can find at least one sales channel in the next 30 days that will show investors that this will scale. If we can, then it will be easy to close this $750,000 round. And if that channel (and maybe others) continue to scale, then we’ll be able to raise a Series A as early as this fall.