How To Do Things That Don’t Scale

Examples of taking multiple products from zero to one and beyond

Hi there, it’s Adam. I started this newsletter to provide a no-bullshit, guided approach to solving some of the hardest problems for people and companies. That includes Growth, Product, company building and parenting while working. Subscribe and never miss an issue. If you’re a parent or parenting-curious I’ve got a podcast on fatherhood and startups - check out Startup Dad. Questions you’d like to see me answer? Ask them here.

---

In one of my earliest professional roles I worked for Mervyn’s Department Store in their marketing department. We weren’t exactly a startup. We had thousands of employees and a retail footprint across 13 states. We were one of the anchor tenants in a lot of shopping malls. Remember those? It was early in the era of eCommerce and we didn’t have a Website you could buy something from. I believe the term was “brochure-ware.” 

What we did have was a lot of love for our brands. And it happened that the love extended all the way to Oprah. What do you do when you realize that several of your products are going to be featured on Oprah in front of an international audience and you don’t sell your products online? You improvise. You do things that don’t scale.

In that case, we took orders over the phone. Thousands of them. We leveraged the brochure-ware to direct people to our customer support line  and drop shipped them products. It wasn’t pretty, but it worked.  

I’ve long believed in the importance of doing things that don’t scale. The original reference to this is Paul Graham’s essay. Recently I’ve been reflecting on this concept from my own past and some of my experience working with other companies and boards. 

In this article I’ll share some of the most profoundly unscalable, effective and enjoyable tactics I’ve experienced in my career across customer acquisition and retention, company culture, forecasting, product-building and more. 

Customer Acquisition and Retention

Most of my favorite, unscalable tactics fall into the bucket of customer acquisition and retention.

Guerilla (Beaver?) Marketing

Before Lyft was a $5B company it was a tiny startup called Zimride. We focused on long-distance, peer-to-peer ridesharing often connecting cities or college campuses. We sold the platform directly to universities and some companies as a way to generate B2B revenue – they would then establish on-campus carpool networks between students.

The earliest demand-gen activities (predating me) were our founders John and Logan dressing up in frog and beaver costumes on college campuses to generate awareness among students. When I joined we did the same thing at Berkeley’s campus and gave away free pizza in between two food trucks at Sather Gate. The students loved it and signed up in droves; the food trucks… not so much.

After we pivoted Zimride to Lyft and we had to figure out how to jumpstart demand in a new market. First in San Francisco and later in 70+ additional cities during my tenure. There were two, primary unscalable demand generation tactics that worked to cold start new markets: startup raids and launch parties.

Startup Raids sound quite a bit more nefarious than they actually are. We’d essentially show up at a series of early-stage tech companies, balloons in tow, and talk to them about Lyft at lunch while handing out ride credits. In one such example we rented a coffee cart and parked it out in front of Techcrunch Disrupt which was hosted just a few blocks from LyftHQ.

We also hosted a launch party in most major cities. It would kick off our demand generation and we would have drivers there to meet and mingle with passengers. Then, at the end of the party, we’d “turn on” the city and passengers could take a free Lyft home from the venue. 

Non-traditional Marketing Channels

It’s well acknowledged that one of AirBnBs earliest, unscalable growth tactics involved Craigslist. But few know that this was also true for Zimride (and eventually Lyft). So true that eventually I got a cease and desist from Craiglist. Yikes.

Craigslist had (and still has) a robust rideshare community. But it’s a bit of a mess. Most days, it looks like this:

We did two things in the early days of Zimride to take advantage of the mess of Craigslist: an organized calendar as a post and publishing Zimride listings as individual rideshare posts. The calendar was my favorite. Because each day’s listings look like a dog’s breakfast we would post a cleaned up version of them and interject our Zimride listings amongst the deep-links to craigslist posts. Both these tactics worked for a very long time and were able to siphon traffic from Craigslist to Zimride.   

Another favorite Craiglist tactic of mine was the Zimride Elite program. One of the consistent challenges of Zimride (and later Lyft) in the early days was ensuring enough drive supply. One way we addressed this was to “recruit” road warriors with Craigslist jobs postings who were already consistently driving between two cities (like San Francisco and Los Angeles) and guarantee them payment if they posted their ride on Zimride. When we launched Lyft we’d employ a different version of this tactic to find our earliest Lyft drivers—leveraging Craigslist job postings.

In Person Activations / Human-powered Tactics

There’s a ton of pressure to create “scalable” or “automated” marketing and I imagine AI will make this even more acute. But some of the best early-stage operational tactics don’t benefit from automation in the beginning. The opposite is true in fact.

At Lyft the earliest cities were powered almost entirely by human effort and the occasional Google Sheet. Here are some examples:

Get On The Damn Phone

We started Lyft with John Zimmer making phone calls to interested drivers to get them to come to our office for an interview and eventually a training. After John it went to college interns, then employees, and then eventually we recruited the initial “drive team” which came largely from the driver community. Calling drivers served a few purposes—it helped us introduce a new concept to people in the way that a landing page could not, it created a very high bar for our earliest drivers by scrutinizing them in a 1-on-1 interview, it created a relationship and a community that was a breath of fresh air in an otherwise impersonal experience.

Driver Interviews & Training

I modeled our earliest interview process off of Trader Joes—all the people you interact with in that grocery store are always so fun. I had a few calls with a director of HR there and they sent me their employee onboarding materials and interview questions. Scalable? Nope. Effective? Very.

Early on we knew that your experience of riding with a complete stranger; sitting in the front seat, fist-bumping, etc. had to feel normal and fun or it would be awkward and people wouldn’t come back. So we orchestrated in-person training around figuring out the technology and a role-played interaction between me and Nick Greenfield that we called “The Perfect Ride.” In it, Nick (who is from Maryland) would do a New Zealand accent, accuse me of being Murray from Flight of the Conchords, and hilarity would ensue when I told him I needed to go to Oakland (not Auckland).  

Building Referrals on a Google Sheet

One of my favorite tactics that bridged the gap between a manual and automated referral platform and led to product features that would drive 50% of Lyft’s acquisition at peak was the referral system. In the early days of Lyft we hacked together a rudimentary coupon code system that allowed us to offer up the ride credits described above. But in the earliest days of a new market we had a lot of drivers who we had recruited ahead of launch and not yet enough demand. So we turned those drivers into an extension of our demand gen efforts through a brilliant, unscalable tactic.

We created a coupon code that was unique to each new driver and gave drivers cards with their coupon code printed on them. The same cards that we’d use with our own street team efforts. This was before the days of the “double-sided referral incentive” and so each week we would run a query that pulled the number of coupon redemptions by coupon code and map that back to the driver in a Google Sheet. We’d then bonus out the drivers based on the number of passengers they recruited as measured by redemptions and completed rides. This system proved to be wildly successful and we eventually incorporated it into the app to make it truly scalable. But before that this unscalable tactic drove thousands of new passengers to the Lyft platform.

But the unscalable, successful growth tactics weren’t reserved for Lyft alone. At both WyzAnt and Patreon we did plenty of things that didn’t scale in their infancy. 

One was the college scholarship program at WyzAnt which served as a mechanism to boost our inbound linking for SEO. Here’s how it worked:

• We created a small, $10,000 college scholarship that students could apply for out of high school.

• We then distributed that information to various high schools, universities and other scholarship aggregators, many of which ended in high-value .gov or .edu top-level domains (TLDs).

• Those sites linked back to the scholarship essentially lending us more domain authority because of those domains.

At peak, over 50% of WyzAnt’s traffic and customer acquisition came from SEO. While it wasn’t entirely related to this the scholarship provided us with a nice organic boost that we wouldn’t have had otherwise.

At Patreon we conducted manual outreach to high-value creators for the first several years of the company’s existence. Rather than use scaled emailing tools which didn’t work when communicating with creators our creator partnerships team wrote and sent personalized, individual communications. Even later, when we were signing up thousands of creators per day we still managed to identify and conduct direct outreach to those with high potential. 

Unscalable Operations

In a recent tweet Nikita Bier suggested:

He’s very right.

Here are some of my favorite layers of operational duct tape applied over the years:

Forecasting

At Hotwire in 2005 there were very few fancy web analytics or reporting tools. Instead, we managed our forecasting and reporting in spreadsheets. And not Google Sheets either, because that wouldn’t even exist until 2006. No, we used good, ole’ fashioned excel spreadsheets on PCs. Each day we’d pull some data from Microstrategy’s data warehouse, clean it up, and refresh our forecasting + actuals. We did that every day because we had to and there weren’t many automations that existed at the time. 

Now you might think, “Adam, this is wild. I can’t believe you and several other people did that every day.” We did AND it was quite useful. By pulling and integrating the data every day we got quite familiar with the performance of the business. It was a forcing function to pay close attention to the details. It’s why, to this day, I emphasize that people are self-sufficient in gathering their own data. Does it scale? Nope. But when you’re early in a company’s lifecycle staying close to the data is critical.

A random aside: we also used to send a mid-day update out to the executives via email showing the day’s trend. We formatted it for easy viewing on a Blackberry. My how times have changed.

Giving Drivers Cell Phones

At Lyft one of our earliest challenges was the quality of the device that a driver used the Lyft app on. Lower-quality Android devices and certain service providers would struggle to get pings from the routing service. This was problematic—it’d appear we had fewer drivers available, passengers would lose sight of drivers on the map, drivers would miss rides entirely, and the overall customer experience suffered. 

So what do you do in this situation? You buy thousands of older iPhones from Apple for pennies on the dollar and activate them for new drivers during training. Problem solved! Driver gets a functional phone for the Lyft app and everyone has a great experience.

Building Creator Pages

As I mentioned above, we deployed numerous unscalable tactics at Patreon to help drive traction with creators. One was our process of manually building creator pages with interns! This was critical for two reasons: first, we hadn’t yet invested in the product surfaces to make building your page really simple and second, it allowed us to show a creator the possibilities of what a great page could look like. Once a creator saw our example they were ecstatic; it was like putting their name in lights. Building pages manually helped us close a lot of new, big creators and bring them onto the platform which fueled the organic, viral growth engine.

Two Tactics that Don’t Fit Neatly Into The Above

My last two are unscalable favorites but don’t fit neatly into any of the above. Here they are anyway.

Culture Building with the Internal Newsletter

One of the most amazing aspects of Zimride, then Lyft, was the incredible culture we collectively created from the earliest days. One of those elements was a semi-factual, The-Onion-esque newsletter created by Paul Thompson and Grayson Badgley. In it, they shared news and updates happening at the company, some stretching of the facts, and important personal updates for our teammates (hey, we were still small and spending a lot of time together). Here’s an example:

Charging Patrons? Later.

My final unscalable tactic was launching Patreon without a mechanism to charge patrons. We launched with the ability to store credit card information securely but not the ability to actually charge those cards and recoup some of the funds. For the first few months we paid creators out manually via PayPal.

Because we paid creators on the 1st of every month as we grew quickly we basically had ~30 days to figure out how we’d charge the stored cards. We pressed the launch button without knowing how we’d solve that problem. If that doesn’t foster a sense of urgency, I don’t know what will!

Wrapping Up

When Paul Graham says that the most common advice they dispense at YC is to do things that don’t scale he’s talking about the hard work of building a company, its product, and the demand needed to start your growth trajectory.

A lot of people approach me in the earliest stages of traction and ask about scalable tactics, loops, etc. While it’s important to think through what growth loops your product enables (or disables)—a necessary precondition for building a really big company—the reality is there’s no excuse for doing the unscalable work to get the engine going and build your culture. It’s often this work that brings you closer to customers and makes your product even better.

What’s your favorite unscalable tactic?!?