I’ve been meaning to write about this for almost two years.
Back in 2020, I started a project with two friends that came to be a platform called Reflr. It was our shared first journey into entrepreneurship, our first stab at building something of our own. It was exciting, exhausting, and exhilarating.
We failed.
This is going to be my weird, cathartic goodbye to something that ended in the late summer of 2022. A dive into what went wrong, where I fucked up, but also a celebration of our achievements, of which there were (without sounding too self indulgent) many.
Reflr’s journey really began many years before, on a lunch break in a car park in the West Midlands. My two future cofounders were sharing lunch, talking about how great some kind of peer to peer recruitment process would be.
To summarise a conversation laden with Birmingham accents.
We all know great people right? When you look at jobs, you can always think of someone who’d be great at that.
And that was the idea - a platform to refer, not apply, for jobs.
It benefited from some loose, untested social proof. When this idea was shared with me in March 2020, three short months after meeting one of the characters above, it followed mere hours after a LinkedIn message from a recruiter “promising a finders fee of £400 for any successful candidate”. I smelt money.
At this point, I knew nothing about hiring. I was less than a 100 days into my second role as a software engineer, I didn’t know the costs companies factored into hiring for a single position, the processes they had, the volumes of candidates they dealt with.
I knew nothing, but I wanted in.
So far, so good.
Following some brief socialising along the lines of “how great would it be if you could refer me to a job and be paid a grand for it” to a gathering of nodding heads, we charged head on into making this idea a reality.
Looking back, I genuinely laugh at what we thought we needed to do - we built a mobile app.
We weighed up the need for a web dashboard, crossed it off, and started building things out. First the ability to view some listings, apply to them, basic searching. We knew .NET so that powered our APIs, and we’d danced with some Flutter at work so that was our mobile stack.
It’s really obvious to me now we weren’t building this because we genuinely thought that was our way out into the world. We just really wanted to build a mobile app. So much so, we even designed an applicant tracking system (ATS for those into the hiring lingo) bundled into this app.
And yes, we were planning to launch with that as our only surface for companies to interact with.
We fucking cheered when we had an end-to-end flow working with notifications. I loved it, we all did. But what in the living fuck were we doing.
As an aside, dear reader, if you’re wondering what my thoughts now are on the MVP we should’ve built and shipped for this, I’ll share this briefly.
A static website, with some job listings, and a link to email us with a prompt for a short recommendation and a link to their LinkedIn.
If you’ve ever heard of the phrase “do things that don’t scale” this is a really good demonstration of that. No need for a system designed to fight off spam submissions (more on that later), or handle and process 100s of competing applications in flight for a company - just us emailing their HR team “hey, we’ve had this recommendation, we’ve checked them as legitimate, let us know if you hire them and we’ll bill you”.
Once the fun wore off, it just didn’t click. This didn’t feel how we’d want this to, it was too slow to work on, we didn’t have experience building good mobile UIs. We’d built a bad thing.
We came to our senses, scrapped the app, and focused on a web dashboard. Again, having not spoken to a single customer at this stage, we set out to build an ATS and a frontend search. It looked great, took a while, but genuinely looked and functioned as a mature product, which is what we thought we’d need.
How wrong we were.
You always have to make assumptions, the idea that you can get through the early stages (or arguably even any stage) without making any isn’t true. However, we made a few too many.
We started out thinking that these were some of the problems we’d have to solve first.
Straight away, you can see the assumptions forming. We’re assuming that we’re going to be inundated with referrals. We’re assuming we’re going to need a payment system to handle bad actors.
These were subconsciously locked in and agreed by all of us on day one. The system we built, on top of Stripe Connect (honestly really good for a three-way-marketplace), could handle all of this for us, and we argued for days about how to combat our quality problem.
We introduced max limits per account, we required Stripe identity verification before a referral could be submitted. We did all of these things before we even posted about Reflr online anywhere.
You’ll be surprised, I’m sure, to learn we didn’t need any of these things.
As I’m about to talk about, we did really quite well on our B2B sales. We landed some good companies, got some great roles with some strong referral “bounties” on offer - but we completely failed to get referrals.
We assumed that was the easy part, we were wrong.
I loved sales.
One of my cofounders was really good, far better than me. He hated it, but was so creative. We bought some companies personalised brownies, we’d mock up their jobs for them and say “it’s a click away, why not try”.
I think here we did really well. Despite knowing nothing about the industry, we learnt it quick. We built too early and spoke to customers too late, but when we did we really succeeded.
It’s genuinely funny how unprepared we were. Having worked at a few more startups since, I can’t believe we got away with how unstructured and jumbled our sales calls were.
We leant hard on customer research. We’d trawl LinkedIn, the sites, learn everything about them. We knew the shortcuts for their “quick summary of our product” speech so that we could get them on side. But that’s when it just went crazy.
Demos would be random, there would be no talking points beyond “let’s get you signed up then”. We had a damn good conversion rate on our booked calls - only one prospect didn’t convert after a call (it helps when they don’t have to pay if they don’t get anything).
We set ourselves the goal of having 5 listings on to launch the platform. This would be across at least 3 companies, one of which was our current employer (easy target).
The feeling of hitting that goal, and realising what we’d been working on for so long was ready to go, was thrilling.
We opened our doors to referrals in April 2022. People were really supportive of it. We had good traction on LinkedIn, good number of visitors, but we weren’t seeing what we expected - referrals.
We scratched our heads - maybe we didn’t have the right jobs? maybe people were going to come back in a week when they knew someone definitely looking?
The weeks went by, but it went nowhere. We considered bending the morals a bit and referring people ourselves. We reached out to peers on LinkedIn - surely someone would know someone.
We’d built a defensive solution, guarded against spam, enforced quality - and it wasn’t being used.
This sadness lingered on for months. We tried to maintain momentum and get new roles on - except now we had to answer the questions of “how many referrals have you had?”, “what’s your applicant-to-hire ratio?” With increasingly soul destroying dodging and spin.
We’d burnt ourselves out, delaying this fantastic achievement (and it was) of a product while building up this launch to be something so much more than it could’ve been (in our minds).
We persevered for 5 months, before realising we weren’t the ones to take it further. In retrospect I have a million ideas for ways we could’ve recovered, but frankly I got to the stage where it was the last thing I wanted to do.
I wanted to quit the thing I’d built from the ground up.
In dramatic wording, fit for the obituary of a statesman, we shut down our GCP projects in September 2022. Our company officially hung on, in limbo, until May the following year.
When I thought about writing this, I couldn’t count the feelings of disappointment I had towards how I approached this product. But the more I’ve written, the more I realise I don’t really regret anything.
It’s cliché, but let’s look at it. We had no clue what we were doing. The startup two of us worked for at the time took the best part of two years to get to the public. We were making it up as we went along, we hadn’t seen how to do things the right way.
Frankly, we responded fantastically to what did come up. I remember a potential customer call on a Wednesday where they mentioned “Oh yeah we use Greenhouse, be great to get the applications fed into there”. Thursday I applied for their partnerships programme. Friday I got an email granting us access to a developer environment and over the weekend built a working (not just a prototype) integration for Greenhouse. We hustled, we learnt how to ship quickly.
We really should’ve gotten to market in a week, not years. We should’ve made less assumptions, we shouldn’t have taken excitement of the idea as validation of the product and proof of market fit. But now we know.
We often talk about Reflr, how it made us better engineers, how it gave us a strong appreciation for sales, product, and running a business. We had no cash flow to talk of, but it was a real thing.
Reflr was my first startup, it certainly won’t be my last.