I Analysed 298 Job Applications. Here's What the Data Actually Told Me.

After more than seven months of job searching, I reached a point where I genuinely didn't know what I should be doing differently.

Apply more?

Target different roles?

Focus on junior positions?

Change my CV again?

Build another portfolio project?

There were plenty of possible explanations for why my search wasn't producing the results I expected.

The problem was that they were mostly assumptions.

Fortunately, I had something else.

Seven months of data.

So I opened Excel.

How this started

I have been tracking my job applications since I started looking for analytical roles.

Initially, the spreadsheet had a very practical purpose. I wanted to know where I had applied, when I had applied and what happened afterwards.

I wasn't exactly expecting it to become a dataset.

Then, during a conversation about my job search, my career adviser said something that caught my attention:

"You're approaching this like an analyst, so let's actually treat your job search like a dataset."

Probably not the safest thing to say to someone who enjoys analysing things.

I may have taken it slightly too seriously.

I went back through my tracker, checked my emails, cleaned up some of the outcomes and started asking questions.

Not:

"Why haven't I found a job yet?"

My data couldn't answer that.

Instead, I wanted to understand something more useful:

Where does the process appear to be breaking down?

First, I looked at the overall picture

My tracker contained 298 submitted applications.

There were also approximately 30–40 additional applications that I had started but couldn't submit after discovering eligibility requirements such as British citizenship, minimum UK residency periods or security clearance requirements.

I hadn't recorded those as applications because technically they weren't submitted.

So I kept them out of the main analysis rather than quietly adding them to make the numbers look more dramatic.

That left me with 298 applications where I could reasonably analyse the outcome.

And the first result was difficult to ignore.

Of those applications:

• 171 (57.4%) received no response

• 116 (38.9%) were rejected at the CV/application stage

• the remaining applications resulted in a very small number of recruiter contacts, further stages, interviews, cancellations or unclear outcomes

  
  

Together, no response and initial rejection accounted for approximately 96% of submitted applications.

That changed the way I looked at the problem.

The bottleneck wasn't where I thought it might be

It's easy during a long job search to start questioning everything.

Maybe I'm not interviewing well.

Maybe I'm not explaining my experience properly.

Maybe I'm answering questions badly.

But my data couldn't really support any of those conclusions.

I wasn't progressing far enough through the process to test them properly.

Only around 1.3% of applications had resulted in what I considered meaningful employer engagement - a recruiter contact, progression to another stage or an interview.

So the biggest measurable bottleneck wasn't:

interview → offer

It was much earlier:

application → meaningful employer engagement

That distinction matters.

If the problem happens before the interview, improving interview technique alone isn't going to fix it.

Then I started testing my own assumptions

One suggestion I'd considered was that perhaps my search was simply too narrow.

Maybe I had been concentrating too heavily on jobs called "Data Analyst".

The data said otherwise.

My applications were spread across:

• 93 BI / Reporting / Insight / Analytics roles

• 73 Data Analysis roles

• 54 Operations / Commercial / Planning roles

• 36 other analytical or adjacent roles

• 34 Business / Process Analysis roles

• 8 Data Support / Quality / Governance roles

Only around a quarter of my applications were straightforward Data Analysis positions.

So "broaden the search" sounded reasonable in theory, but the tracker showed that I had already done it.

That hypothesis didn't explain the results.

Maybe I was aiming too high?

That was another possibility.

Perhaps I needed to concentrate more heavily on explicitly junior or entry-level positions.

So I checked.

I identified around 43 applications containing terms such as Junior, Graduate, Intern, Entry-Level, Associate or Assistant.

The outcomes were:

• 33 (76.7%) - no response

• 9 (20.9%) - rejected at application stage

• 1 (2.3%) - progressed to another stage, followed by no further contact

That was useful because it challenged another seemingly obvious solution.

Simply applying for more entry-level roles wasn't producing better outcomes either.

The job title started to matter

Next, I compared the broad categories of roles.

This is where the analysis became more interesting.

Straightforward Data Analysis roles had a no-response rate of around 71%, making them one of my weakest categories for employer engagement.

BI / Reporting / Insight / Analytics generated more responses and some of the limited examples of genuine employer interest.

My completed interview came from the Business / Process Analysis category.

Operations / Commercial / Planning also showed a lower no-response rate than traditional Data Analysis roles, and one recruiter had approached me directly about a role in that area.

The numbers are far too small to announce that I have discovered the perfect career strategy.

I haven't.

But they were enough to make me question something.

Perhaps my most realistic route into analytics isn't necessarily through the job title I originally expected.

Then I looked beyond the numbers

I also went back through rejection emails.

Most rejection emails are templates, so I wasn't about to perform sentiment analysis on every "we were impressed with your application".

But one theme appeared often enough to catch my attention.

Experience alignment.

Employers regularly referred to candidates whose experience was more closely aligned with the specific role.

That doesn't prove why every application failed.

But combined with the patterns in my tracker, it gave me another question to consider:

Perhaps the issue isn't simply whether I have analytical skills.

Perhaps it is how those skills compete against candidates who already have direct commercial experience in very similar roles.

That's a different problem from not knowing SQL or Power BI.

And it requires a different response.

The data I didn't analyse

This was probably one of my favourite parts of the exercise.

Because once you start analysing your own spreadsheet, it becomes very tempting to analyse everything.

But not everything I had recorded was good enough to use.

For example, I had a field for remote/hybrid working arrangements.

Unfortunately, I hadn't recorded it consistently.

Sometimes I entered "remote". Sometimes I recorded more detail. Earlier in the process I wasn't expecting anyone - including me - to audit the spreadsheet seven months later.

So I left it alone.

The estimated 30-40 applications blocked by eligibility requirements also remained context rather than a precise KPI because I hadn't recorded them individually.

There may also have been a small number of rejection emails that I couldn't reliably match to an application.

And that's important.

Having data doesn't automatically mean you should use all of it.

Sometimes the most responsible analytical decision is simply:

I don't trust this variable enough to draw a conclusion from it.

So what did 298 applications actually tell me?

They didn't tell me exactly why I hadn't found a role.

They couldn't tell me what individual hiring managers thought.

They couldn't prove whether the market, my CV, my experience or something else was responsible for each individual outcome.

But they did tell me several useful things.

I wasn't applying too little.

I wasn't restricting myself to Data Analyst roles.

Entry-level applications weren't performing better.

The biggest measurable bottleneck occurred before meaningful employer engagement.

And the limited positive signals weren't coming exclusively from traditional Data Analyst positions.

That was enough to change the question.

Instead of:

"How can I apply to even more types of analytical roles?"

I started asking:

"Should I actually narrow the search?"

Perhaps the better strategy is to identify two or three areas where my previous experience and analytical skills create the strongest combination, rather than continuing to spread applications across an increasingly broad range of roles.

That's not a conclusion I expected when I opened the spreadsheet.

Which is exactly why analysing it was useful.

Final thought

This analysis didn't find me a job.

It didn't produce a magic formula for getting through recruitment systems either.

What it did was separate some of my assumptions from what I could actually see in the data.

And perhaps that's my favourite thing about analysis.

Sometimes you start with a question expecting the data to confirm what you already suspect.

Instead, it tells you that you've been asking the wrong question.

In this case, the question may no longer be:

"What else should I add?"

It might be:

"Where does the evidence suggest I should focus?