Two weeks ago, I graduated with a master’s degree, travelled around Ireland, and briefly became a Viking.

This week, I scraped strangers from LinkedIn and argued with Claude Code about why my image filenames refused to load.

Life comes at you fast.

Welcome to Week Five.

NOTHING HAS HAPPENED

If you look at this week in the simplest possible way, I have made absolutely no progress.

Clickrabbit has no new clients.

Nothing has launched, nobody has paid me and there has been no dramatic moment where everything suddenly clicked into place.

Behind the scenes, though, quite a lot has changed.

I now have roughly 2,500 raw leads across six different niches, a completely new outreach angle to test and an unnecessarily advanced system for putting photographs onto a personal website.

If none of that sounds remotely interesting to you, I completely understand.

For some people, workflows, APIs and databases probably sound like hell on earth.

Unfortunately, this is my journey, and this week mostly involved building things nobody can actually see yet.

APIs, databases, Framer components, new code and old code from months ago being dragged back from the dead and forced to serve an entirely different purpose.

All the glamorous stuff.

THE FIRST CAMPAIGN FLOPPED

I briefly mentioned last week that the first Clickrabbit outreach campaign did not perform particularly well.

That was me being polite.

It was shit. You can view the emails in Week Three

The emails were opened, but almost nobody replied.

Looking back, I think the biggest problem was how much trust I was asking for.

I was essentially appearing in someone’s inbox and saying:

Hello. You have never heard of me. Would you like to book a call so I can convince you to give me a lot of money for a system that might improve your business?

A compelling offer.

I was trying to sell the entire pilot programme before proving that I could build anything useful.

There was no case study, no reputation and no reason for them to trust some random bloke with a rabbit-themed company.

I do not blame them.

So, I have changed the approach.

THE NEW TEST

Instead of immediately asking people to book a call, the new emails focus on one specific problem their business might already have.

I then explain how Clickrabbit could solve that problem and offer to send them a working demo.

They do not need to believe every promise I make.

They only need to be about ten per cent interested and reply with “yes”.

I can then send them something tangible instead of expecting them to trust the promises of a complete stranger.

I still need to build the final demo, but the idea is simple. They will be able to click through the system themselves and see exactly how it would work for their business.

The first problem is outdated enquiry forms.

A lot of businesses still use forms that ask for a name, phone number and email address, then dump every enquiry into the same inbox.

The business learns almost nothing about the person enquiring, and the customer gets no immediate value in return.

A better system could understand what the customer needs, measure how serious they are and send the information directly to the person best placed to respond.

The second problem is shared leads.

Some businesses buy lists where several competitors receive the same customer details.

That turns the process into a race to call first, followed by a race to offer the lowest price.

Instead, Clickrabbit could help them generate and qualify their own enquiries.

The third problem is people booking calls and then disappearing.

Someone might book a meeting on Monday while their calendar looks completely empty.

By Friday afternoon, their manager has scheduled a meeting they cannot refuse, their week has gone to hell and your sales call is suddenly the least important thing in their life.

Without reminders and useful follow-up, that lead disappears.

Each email tackles one recognisable problem and offers to show the solution working.

It is a much smaller ask.

Rather than telling people I can build something useful, I can actually show them.

I PANIC-SCRAPED LINKEDIN

To test all of this, I needed people to message.

Conveniently, as I mentioned two weeks ago, I still had a two-month LinkedIn Sales Navigator trial running.

During the first month, I followed the traditional free-trial strategy of signing up and doing basically nothing.

Then I realised the trial expired on 4 August, which is also the day this weekly is being published.

I suddenly began scraping LinkedIn as if the entire website was being deleted.

I now have around 2,500 raw leads across roughly six niches.

By raw leads, I mean I have their names, roles and company information, but the data is not ready to use yet.

I still need to run the leads through Clay to find and verify their contact details, organise the final lists and remove anyone who is clearly a terrible fit.

Once the leads are enriched, I can upload them to Lemlist and build the actual campaigns.

I have also finally been paid, so I am upgrading my Lemlist plan and testing both email and LinkedIn outreach instead of relying entirely on email.

Hopefully, approaching people through more than one channel will make me look slightly less like a suspicious rabbit-themed stranger.

For the exact scraping process, including the code and how to export everything, you can read all about it in Week Three

I explain the whole process there because repeating it would turn this weekly into technical documentation, and I have already tested your patience enough.

WELCOME TO THE NERD CORNER

If this weekly has not been nerdy enough yet, do not worry.

It gets worse.

I have also built a new photography system for Life of Jake.

Previously, adding photographs to the website meant manually resizing them, converting them, uploading them, writing alt text, organising the information and making sure the important part of each picture was not awkwardly chopped off.

Now I upload a finished photograph into the correct Google Drive category folder.

That is it.

From there, the system checks whether the image already exists in the database.

If it is a new image, the system compresses the file for the web, extracts the camera settings, writes accessible alt text and identifies the most important part of the photograph.

That last step tells the website where to focus when the image needs to be cropped, rather than casually removing somebody’s head.

The finished image and all its information are then uploaded to Supabase, the database behind the photography section of the website.

All without me doing anything else.

Could I have continued uploading the photographs manually like a normal person?

Yes.

Did I instead build an automated photography pipeline involving Google Drive, n8n, AI, image processing and a database?

Obviously.

Isn’t it just beautiful 🥹

The goal is to remove enough friction that I actually keep the photography side of the site updated.

There is also a reason I needed every image organised in this particular way, but that is a project for next week.

NEXT WEEK WILL ALSO BE BORING

I am not going to lie to you.

This week has been pretty boring, and if you have read this far, you are a trooper.

It has mostly involved repetitive setup, fixing old code, building new components and trying to make several different systems speak to each other without having a breakdown.

Not every week can involve graduating and briefly becoming a Viking.

This is probably only part one of the boring setup period.

Next week, I still need to:

So, next week may involve even more databases, workflows and staring blankly at loading screens.

Once everything is live, though, I can finally stop building the machinery and start seeing whether any of it actually works.

That’s a lie.

If you can’t tell already, I love building.

My favourite show as a kid was Bob the Builder, after all.

Also, this weekly is being published on my mum’s birthday.