Design Pickle Hired the Freelancers Its Competitors Resell

Design Pickle employs 470 designers. Not one of them is in the United States, and not one of them is a freelancer.

The agency that taught him what not to do

Before Design Pickle, Perry ran a creative agency for eight years. Its best year did about $3 million on high-ticket work — half-million-dollar branding projects, quarter-million-dollar trade show builds — with never more than eight clients at a time.

We never specialised. We were a generalist agency trying to do too many things for everyone. We couldn’t develop a process, we couldn’t develop a system.

He had even tested the offshore model there, with a 50/50 partner in Buenos Aires and twelve designers. The cost side worked; nothing else did.

We spent so much time redoing work and managing clients that we lost all our profits in the client management side of it. I made about every mistake in the book possible with that agency, which I actually think is like 85% of why Design Pickle has done well — because I got all my mistakes out of the way.

The aha moment was the inverse of the agency: do “really boring design production, straight-level stuff,” get it dialled in and optimised, and have a different kind of business entirely.

Three people and a hacked-together system

Design Pickle launched in January 2015 with one creative, one project manager and Perry. From the first day, the creatives worked for Design Pickle rather than through it.

I just knew how hard it is to always have to find someone, build that relationship, explain what you need, explain your brand. And I never wanted to actually be in between the client and the creative, so our clients just work directly with the creatives.

He picked the price by listening to startup podcasts and deciding he wanted recurring revenue. The infrastructure was less considered: “the running joke is if you knew what email address we used, you didn’t even need a paid account, because you could just circumvent the whole thing.”

The labour model, stated plainly

The 470 production staff sit in the Philippines, Mexico, Indonesia, Argentina, Colombia, with Peru and India being added. Perry does not soften what the arbitrage is, and does not pretend the wage is generous by American standards.

Our teams are in markets where they are on average making three to four dollars an hour. If they want to get a job with us, we’ll start at six to seven dollars an hour. So we’re 3x-ing their take-home, still fractional to what you could hire here.

The competitive argument for it is about supply rather than cost.

You can’t make a living cobbling together jobs off Fiverr or 99designs, and that’s where a lot of those companies get their labour. We give them full-time rates, weekly contracts, benefits, all of that, so we can recruit from the best of the best. Our goal is to be the best employer in every one of those markets.

The first eighteen months of recruiting were exactly what you would guess — sorting Upwork by designers with the most reviews and offering them guaranteed hours. “And then we ran out of people.” There is now a dedicated recruiting team, and the 470 includes HR, training and recruiting functions rather than just designers.

Gross margin runs about 68% in a good month, dropping to the low 60s while adding middle management.

What customers actually buy

The entry plan is about $400 a month, flat rate, unlimited requests — “the all-you-can-eat buffet of Design Pickle.” The plan Perry cares about is the $1,000 one, which puts a designer inside your Slack from nine to five.

I laughed the day we launched the thousand dollar plan. It’s easier to sell, the clients stick around longer, our cost is negligible. It was the running joke — why didn’t we do this sooner?

Asked which feature converts people up, he rejects the framing: they are two different audiences, and what people value is not a feature at all.

I can talk to my designer real time. Here’s dropping a file in Slack, what do you think about this, take a look at this. And their little green dot turns on at 9am and it turns off at 5pm and they’re there. They’re available.

The moat is the routing, not the design

Latka asks whether the competency is design or recruiting. Perry says neither.

We’re not winning any Madison Avenue awards here. What we are is we’ve built a system that optimises creativity. Anyone can go out and make a couple hundred thousand dollars a year copying what we do, but there are legitimately complicated scalable challenges — we’ll do 12,000 individual unique requests this week, serving hundreds of thousands of emails and messages and managing it. That’s the moat. Not just finding some cheaper designers and reselling their time.

Selling annual plans to a service nobody keeps forever

The retention numbers would frighten a conventional SaaS board: about 25% of signups are retained for life, and roughly half of first-year revenue is predictable. Perry is upfront that design is not sticky the way an accounting tool is.

So the sales motion is built around capturing the revenue before the customer is done.

1. Assume a short natural life. A client might otherwise stick around three, four, five months.
2. Sell the discounted annual plan in the first 90 days. “We really aggressively look to sell them on a full discounted 12-month plan to capture that full revenue.”
3. Comp the reps on contract length, not volume. There are no quota-carrying reps; bonuses scale with the length of the first deal, so a monthly signup pays a fraction of what a twelve-month one does.

He has a name for what happens at the end of a good engagement.

We have a whole categorisation of happy churn. I love you, you guys are amazing, my project’s done.

The ladder, and the debt that funded it

Perry gives the whole revenue history without being asked twice: $300,000 in the first year, then $1.2 million, $3.4 million, $4.2 million, and $10 million in 2019. For 2020 he says “we rounded down conservatively, we’ll be doing 14 million” — the figure the GetLatka profile now carries for that year.

All of it bootstrapped except $1.2 million of debt from Lighter Capital in 2018. The economics made it easy: lifetime value well over $4,000 against a cost to acquire of $600–700, so a customer returns their acquisition cost in two to three months.

I always joke it’s like the sketchy college credit card — high interest and super easy to get. But they were easy, they understand it. It was a great decision for my business because of my numbers. If I was selling a $25 or $30 product without a big return immediately, I don’t know if that would have been the route.

He is candid that the reason he took it was that nobody else would lend: “we are not a pure SaaS business, we’re a tech-enabled service, so it was harder for us to find other lending.”

Fiverr came twice

In the year before this conversation, Fiverr approached about buying Design Pickle twice — first at $2–3 million, then at $20 million. Perry said no to both, and is careful to note these were casual conversations rather than term sheets.

I loved the Fiverr team, don’t get me wrong, I think they’re badass. But the deal was going to be public stock, earn-out, performance. It was just complicated, and I don’t think it was the right fit. I knew there was a number for me to make it work and we just weren’t there. I think the timing was off.

What he would rather buy is stranger and more interesting: burnt-out agency owners, for their account managers and their books of business.

I have this sort of weird fantasy of finding burnt-out agency owners and gobbling up their account managers and putting them all on our customer success team, and then buying books of business that we could resell our services directly into to recoup on the M&A transaction.

Asked what he wishes his twenty-year-old self had known, the man who spent eight years failing at an agency before this worked does not talk about business at all.

Stop drinking immediately. It’s not helpful.

Sources Russ Perry’s interview with Nathan Latka, recorded 27 May 2020; revenue and funding rows from the GetLatka Design Pickle profile.