The Good, the Bad, and the Cloud

Episode 14: 01/03/2020

It is spring 2020. We aim to have El Toco live by late summer.

All that’s left is to create a pretty website and load everything onto the cloud. There are no more concepts to explore, just details.

?

The devil, unfortunately, is alive and kicking in those details.

Toyota famously investigate production problems by repeatedly asking why. Each time somebody gives you an explanation for the Bad Thing that happened, you respond with "why?".

The idea is that, eventually, you'll get to the bottom of what really went wrong. It isn't clear how effective this in practice. Toyota do seem to have a habit of recalling cars. I have experimented on other people and it is remarkably efficient at winding them up, in a sort of passive aggressive way. It is especially not recommended the next time your partner accidentally overcooks dinner.

Unbeknown to us, El Toco is four, long, years away from launch. Many of the delays that lie ahead involve battles with third parties. If you keep asking why long enough, you always get back to the explanation that there are just too many details.

?

The cloud is a veritable viper's pit of details. You need to know exactly what hardware to use and exactly how to set up your software so that somebody can't just waltz in with the default password and destroy it all for jokes. It's basically impossible for a small startup to do this all alone, unless one of the founders already has experience in the area. Neither me nor our brave graduate employee have experience in the area, so we had to find somebody to help. That turned out to be surprisingly difficult, as we will see today.

?

?

Butch Cloudy and the Simple Storage Service Kid

(If that title means nothing to you, don't worry. Readers familiar with Amazon Web Services are rolling in the aisles.)

From 2016 to 2019, El Toco enjoyed a quiet life entirely on desktop computers. In order to become a proper search engine, we have to move all of its component bits to servers in the cloud. Our last action in 2019 was to find a company who would help with this. Or so we thought.

?

After an initial meeting in the part-kennel, part-IKEA showroom of a central London WeWork office, the cloud consulting company promised to deliver a shopping list of what we needed, in a few days' time.

?

A week went by. The shopping list did not materialise. I contacted them to find out why and the founder of the company replied in a terse email explaining that we would need to pay them first. The invoice was for about ￡5k, the largest El Toco had faced up to now by a significant margin. The informal tone of his email caused me to raise an eyebrow, but this was the first time we had dealt with anybody actually working in the tech sector, renowned for its casual approach to things like manners and clothing. So I paid up and got on with other matters.

?

Christmas passed, in hindsight the last normal one everybody will experience for several years. In the new year of 2020, I met our graduate employee in a deserted coffee shop to plan the remaining tasks until we went live. Little did we know that, although we had a long future working together ahead of us, this was to be the last time we ever physically met.

In that meeting, it quickly emerged that our launch depends on getting our programs working on Amazon's cloud. There are so many things we either can't finish or don't really know how to start until the cloud bit is sorted out. What sort of things? You may wonder. Things like whether our applications will be running in Windows, or Linux, an operating system I haven't used since university.

?

January 2020 drifted by. Still without receiving our Amazon shopping list.

Thus began a new learning experience for me as a founder: dealing with a rogue contractor. Of course I sent chasers, all through January, with increasing urgency. If we expect to go live in summer, we've already burned a sixth of that time, waiting for a document. Sometimes, the cloud consultants replied to the chasers. Other times, they just ignored them.

?

The shopping list document was eventually produced on Valentine’s Day. It had taken two months by that point, so I was expecting a pretty good document. Actually, that's a lie. You can pretty much guarantee that any work-related document you've waited two months for is going to be a letdown. But I was interested to see what could have justified the delay.

Nothing, apparently. It was a template, consisting mainly of verbiage, where they had changed the company name from “Our Next Victim Ltd” to “El Toco Ltd” and tried to disguise it by pasting in some pretty Amazon diagrams.

I didn't bother reading it. A one minute flick through was enough to see it was the business equivalent of that mouthy kid at the back of the class who is quick to backchat the teachers but copies and pastes from Wikipedia when the time comes to do their homework. No actual work had been done. Especially, nothing had been revealed about what we would need to do on Amazon’s cloud in order to launch.

?

The document asked for another roughly ￡10k in order to set up all our systems. There was no way in Huddersfield they were going to get even more money out of us, if it took them two months to fail to write a shopping list. But we had already paid these contractors, using our hard-fought grant money. Yikes! Do we try and get it back?

They were saved from me pursuing a refund by the second lie they told us during the original WeWork meeting.

They had explained, handily not in writing, that if we ran our Amazon bills through them, there would be a discount applied, because they were bulk buyers.

?

Every part of me was itching to get the money back, because they had already done significant damage to our launch timeline. We haven't discussed it in the last few episodes but the time pressure, which I kept harping on about during the Medellín years, has not gone away. It will never go away until we launch. At any given moment, a competing search engine could launch and steal our thunder. Also, now we are on the real launch countdown, each month where we are not live burns our remaining cash.

But pursuing a refund is very stressful, even if it doesn't end in litigation. So I decided to let them keep the money, run our Amazon bills through their account, and hope that the promised discount makes this worthwhile in the long run.

?

I contented myself by asking them to make various rounds of pointless corrections to the document in question. This was some consolation, especially since I never actually read it. This did not, however, get us any closer to launch.

Interlude from the future

The money we prepaid the cowboy contractor will cover our Amazon bills until 2023, when it eventually runs out. Like all good pyramid schemes, the contractor's original company will have been bought multiple times in the intervening years, meaning that the people running our account in 2023 will have no contact with the original founders we met in the WeWork back in 2019.

By this point, their business has grown vastly. It now has a real office, and we are one of the smallest clients. Our new account managers have no idea that we were originally promised, because nobody wrote it down. This makes it an easy job to get them to explain in passing that there is, actually, zero discount available for running our billing through them. Their company receives some sort of kickback from Amazon, which does not get passed on to their clients.

In summary, they took our money up front, in exchange for setting up our systems on Amazon and giving us a discount on everything we bought going forward. They then didn't set up our systems on Amazon, nor was there a discount. But they received some sort of money back from Amazon.

You can see how, if we fell for it, other people will have too. Have we stumbled on a money printing machine which gets people to pre-pay for Amazon services and then collects the kickbacks?

I remain unsure, but they definitely lied to us.

?

We are all familiar with the stories of dodgy builders. The learning here is that dodgy contractors also exist in the tech sector. Everything they told us was a load of rubbish, designed to take advantage of the fact that we were in a hurry and new to running servers on the cloud. The buzz around cloud computing seems to have spawned lots of consultants who offer vastly marked up “expertise” of this nature. Buyers beware.

?

Back in early 2020, this left us with no path to launch. After wasting two months, we still needed to find somebody to help with setting up the cloud servers. The search was on to find a legitimate company.

?

?

?

Enter the good guys

?

You would expect that Amazon, owning the cloud, has a service allowing you to pay them to set their products up for you. Somewhat paradoxically, they don’t. Anybody using Amazon must either hire a third party contractor, or do the work themselves.

?

I am sure there is a simple business reason why they have arranged things like this, but it eludes me. As a token gesture, Amazon publishes a list of approved contractors on its website, so I went down this list, contacting the UK-based providers who seemed legit.

?

This was the second time while setting up El Toco I really had to work the phones, the first being the grant investigations during 2019. The conversations were surprisingly interesting. Unbeknown to most of the population, the British government is one of, if not the, biggest client of Amazon Web Services in the world. All those websites like that of HMRC are, under the scenes, running on Amazon’s services. Some of the people who set this up realised that they had accumulated a great deal of knowledge about scaling cloud services, so created their own company to share this knowledge.

?

?

All the businesses I spoke to were legitimate, but seemed to rely a lot on “whites of the eyes” tactics when it came to pricing. In most conversations I was asked where we were in the fundraising cycle as they tried to work out how much money they could squeeze out of us. I genuinely didn’t have a webcam at this moment in time, nor did El Toco have a LinkedIn profile, so the companies were forced to guess how much to charge to set up our servers.

?

And guess they did. Now, we know that everybody was quoting for the same service, because they all provided a system architecture diagram exactly like the one we eventually got out of the cowboy consultant after three months of badgering. The new guys all did theirs for free, within a few days of being asked.

?

?

The prices for this work varied however. A lot. The top quote was over ￡100k. Presumably I came across to them as having more money than sense when we did the initial call. About ￡100k more money than sense. After them came the majority of the providers, who wanted ￡30k-￡50k for the same piece of work. Finally, there were the realists, who came in at less than ￡10k.

?

The El Toco we will launch with is not an all singing, all dancing search engine which can take on Google in terms of scale and scope. There are not millions of users, there are not petabytes of data (a million gigabytes).

What we are doing is producing a bare bones, minimum service, to demonstrate that El Toco is commercially feasible. In startup jargon, this is referred to as the minimum viable product. It is the cheapest, most modest thing you can give to the market which people buy enough for it to break even.

?

The best quote was from a company called Mysterious Code Ltd. . They were friendly on the phone and also had the commercial nous to get that we are not throwing huge sums of money at the cloud for a service with no users.

Many of the cheaper cloud consultants achieve this by having the client-facing people in the UK but their actual workforce overseas. India, unsurprisingly, is one of the main places, and Amazon itself has a several large support centres there. Mysterious Code’s are in Eastern Europe, which seems even better because it is a closer time zone.

In fact, it won’t really matter what time zone they are in. Over the coming year, Mysterious Code will spend the vast majority of the time waiting, for us. We’ll get to that next week.

?

The silver lining

?

Being messed around by the dodgy cloud contractor has added about two months to our launch schedule.

This is painful, but it has resulted in one piece of good luck. While we’ve been picking a new contractor over the last few weeks, many of them are finding themselves at an unexpected loose end. Noises about a virus coming out of China are causing clients to get jittery about splashing out on shiny new cloud projects. Suddenly, the contractors are looking at their own cash flows, wondering if they can cover their overheads.

?

For El Toco itself, this fuss about a new Spanish flu is irrelevant. It's a big enough ask to just be launched at all. We certainly don't have the luxury of choosing when that launch takes place. But this scare about a new global flu pandemic has ended up making the cloud companies more interested in our business than they otherwise would be.

It’s also given me an idea. The other thing needed before launch is to finish revamping our website. If people are pulling back on their spending, maybe we’ll be able to find an awesome web designer. Somebody we could not ordinarily afford.

?

For the time being, there is no business reason to be in London now that fundraising has been achieved. So while we get ready for launch, I’ve moved in with my girlfriend in Lille, France. It's rather unclear how long I can stay here because of Brexit. But there are so many British expats in France that somebody is bound to think of something.

And as we all know, those noises about that virus coming out of China are heralding something far more dramatic.