Software runs my life

Category: Internet Page 1 of 16

padlock

Adding a free SSL certificate to Google Compute Engine

Google provides a one click install option for WordPress on their Google Compute Engine instances. This is a very economical and customisable way to spin up a WordPress blog if you’re comfortable with a Debian shell. Unfortunately it only spins up a HTTP rather than secure HTTPS instance, so here’s how to add a free SSL certificate.

Firstly, log into your Cloud console and open an SSH shell for your instance

Google Cloud VM Instance SSH login

Then in the shell, run the following commands as mentioned here:

sudo apt-get install certbot python-certbot-apache

Accept the downloading and installation of some Python packages, then to request and install a new Let’s Encrypt certificate into Apache the :

sudo certbot --apache

It will ask a number of questions such as the server hostname and whether you want to redirect HTTP to HTTPS (you should), then link to the SSL Labs server test to verify your installation.

Finally you can optionally run a dry run renewal, just to make sure it all works:

sudo certbot renew --dry-run

And that’s it!

Tableau and Mable webinar

Recently I was invited to present at an AWS and Tableau conference on how companies use data effectively. There were multiple parts to this collaboration, including this linked article that acts as the overall summary of our business and Tableau involvement.

As it was during COVID, the full conference was recorded in a webinar format. This is my quick introduction video:

Tableau intro final

The full webinar can be viewed here. It captures a lot more detail such as an overview the data and systems architecture used, as well as specifics around our moderation service algorithm.

Finally the data architecture that I discuss, and show on a visual during the full video itself, is reproduced again below:

The 3 stages of a services marketplace

The Silicon Valley alter that is the whiteboard

1. The Hands On phase

Like every great episode of Silicon Valley, it all starts with a whiteboard. Customer (demand) and worker (supply) names are written on the board, with relevant skills and availability scribbled down. Everyone gets on the phones and calls each side, trying to make connections. Every connection is a new possibility, a new thread in the web you’re trying to spin. You’re losing money on every one, but if you’re doing it right they start to become sticky.

This means customers are not just delighted that you introduced them to someone that can solve a problem for them, but that it has the seeds of an ongoing relationship. They enjoy working with the person, they need that problem solved ongoing and (for your benefit at least) they see the value you add as the third wheel.

At worst they appreciate the comfort of having you on the phone to handhold them through the process, at best you are a critical part of the value chain that is hard to replicate without scale. Except you don’t yet have scale so this might cost you big time upfront – software, insurance, data, templates, logistics, call centre etc. Don’t resent the cost, because it’s also your barrier to competition. Watch that burn while you find market fit.

2. The Acquisition Marketing phase

Once the relationships are sticky and your churn is under control, you need to scale quickly to bring your unit cost down and start to pay off that upfront investment (or raise on a growth pitch to kick this can down the road). It’s time to look at the chicken and egg problem of liquidity. What services are not getting traction, and how might each translate into an efficient performance marketing campaign?

You probably have no brand equity, so your performance marketing needs to be very targeted and ideally not competing with established sector players. It’s also tricky to scale, because your segments could be very small and fast moving. 95% of marketplaces are demand constrained, so start by focusing on 1 side only.

Start testing campaign segmentation by customer type first until you get your CPA down to at least breakeven, before trying to solve for secondary variables like geography or availability. You’ll also need to start building a view on Lifetime Value (LTV) by each segment, so eventually you can think about bidding based on Return On Ad Spend (ROAS) or similar metrics. But every customer is different, and so this quickly also becomes a scale challenge that will require marketing data investments.

Don’t purely optimise on cost however, as the goal of this phase is to build number of connections not the number of users. Don’t underestimate power users who create a lot of connections even if they don’t spend a lot, you need to talk to and learn to love these people without becoming too dependant.

This phase is why 50% of startup investor’s money goes to Facebook and Google, you need to learn to love their tools and optimise (aka the sexier “growth hacking”) like crazy. But this needs to be just a phase – if you never graduate then you are at the mercy of the next algorithm change and/or well funded competitor. Don’t become the next GroupOn who spent like crazy here but never graduated.

3. The Product Growth Phase

Also called the Network Effect phase. This is the holy grail, where your product flywheel starts to spin and growth happens organically. There are 3 major things required here.

Finally and foremost – you need to understand what part of those calls in phase 1 relied on the human touch, and what was to overcome objective friction. Pour the gold of your customer service team over the trust issues, and scale the rest in-product with great UX that makes your funnel so intuitive as to be invisible.

Focus relentlessly on removing any friction in making a new connection; this could be through removing double commits, applying data and recommendation engines, teaching your users how their behaviour affects their success, systemically weeding out quality and fraud issues and many more concepts.

Finally you should clearly commit only to verticals where you can be number one. Your next key competitor will play in a narrower not a wider vertical than you, so you need to lead market share not just overall but more importantly in the key customer segments you care about. The network effect is a big moat, but it isn’t the winner takes all endgame it was first thought to be.

Page 1 of 16

Powered by WordPress & Theme by Anders Norén