Best Cities to Get an SEO Job

Data

Combined Statistical Area population – The 2016 Census.gov population estimates for the 75 largest CSAs or MSAs (Metropolitan Statistical Area) in the United States
SEO Title Job Openings – # of full-time job openings in Indeed.com within 50 miles from the nearest location* with “SEO” in the title; data was pulled the first week of the month from August 2016 through January 2017.
SEO Skill Job Openings – # of full-time job openings in Indeed.com within 50 miles from the nearest location* with “SEO” anywhere in the job posting (title or description); data was pulled the first week of the month from August 2016 through January 2017.
SEO Workforce – # of LinkedIn profiles within 50 miles from the nearest location* with “SEO” in their current job title; data was pulled the first week of the month from August 2016 through January 2017.

Formula

I used a method called feature scaling to normalize the population and job data. This allowed me to compare these numbers even though they are on wildly different scales. From there, I created a formula to make something I call the SEO Job Pool Index.

• SEO Job Pool Index – If LinkedIn jobs are ‘A’, Indeed SEO Title Job Openings are ‘B’ and SEO Skill Job Openings are ‘C’, a feature scale normalization was completed from the results of this formula: A/(B/C)

The SEO Job Pool Index was created because while I thought the total CSA population was relevant, it didn’t give me a good idea of who these people were. Some places are going to have a greater share of people qualified for and/or interested in the SEO field. That’s where this formula comes in.

In simpler terms, think of the SEO Job Pool Index as a major indicator of the number of people looking for an SEO job and the Population Index as a minor indicator.

If Population Index is ‘A’, Indeed Title Index is ‘B’, LinkedIn Job Index is ‘C’ SEO Job Pool Index as ‘D’, and with X being the calculation by CSA, here’s the formula I used. Everything in red represents the SEO job Supply and green the SEO job Demand.

X=((B+(C*0.1))/(D+(A*0.25)))*100

Data

Data point: Combined Statistical Area (CSA) July 1, 2015 population estimate
Definition: A CSA is essentially a labor market. Some metros are adjacent and closely overlap when it comes to potential employees. In those instances 2, a CSA may combine some or all of those metro regions for a more complete labor and economic metropolitan area. Nine metros included in this study are not within a CSA 3. Metropolitan Statistical Area (MSA) population estimates were used in those cases. The top 75 CSAs/MSAs were used in this research.
What it represents: The total labor market population
Source: Census.gov 4

Data point: SEO Title Job Openings – May 4^th, June 3^rd and July 4^th of 2016
Definition: A 50-mile radius around a zip code from the most populous city in the CSA was used for these searches. “SEO” was searched “With these words in the title”. Only full-time jobs were counted.
What it represents: The total number of open SEO positions in a CSA
Source: Indeed

Data point: SEO Skill Job Openings – May 4^th, June 3^rd and July 4^th of 2016
Definition: A 50-mile radius around a zip code from the most populous city in the CSA was used for these searches. “SEO” was searched “With all of these words”, which could include the job title or description. Only full-time jobs were counted.
What it represents: The total number of open positions related to SEO in a CSA
Source: Indeed

Data point: SEO Workforce – May 4^th, June 3^rd and July 4^th of 2016
Definition: A 50-mile radius around a zip code from the most populous city in the CSA was used for these searches. “SEO” was searched in the “Title” field. Only current positions were counted.
What it represents: The total number of employed people in a CSA with “SEO” in their job title
Source: LinkedIn

Formula

Before I get into the actual formula, I’ll walk through how I normalized the data. Have you ever tried to weight and compare criteria using datasets with wildly different scales? As you may have guessed, my first stab at this was trying to do just that. The results were ugly and I had no idea why, until I read more about normalization.

Here’s what Wikipedia has to say about normalization:

I specifically used a method called feature scaling to convert all datasets to a 0-100 scale. Otherwise, I wouldn’t be able to evaluate if 8 SEO job openings for a population of 9.5 million is high, low or normal.

To normalize the data through feature scaling, I used the formula below where X_min is the lowest number of all 75 metros and X_max is the highest.
X*100=(X-X_min)/(X_max-X_min)

Here are normalized metrics I used:

• Population Index – CSA population estimates
• Indeed Title Index – SEO Title Job Openings
• LinkedIn Job Index – SEO Workforce
• SEO Job Pool Index – If LinkedIn jobs are ‘A’, Indeed SEO Title Job Openings are ‘B’ and SEO Skill Job Openings are ‘C’, this normalization was completed from the results of this formula: A/(B/C)

I should probably explain the SEO Job Pool Index metric before going on any further. 5. The SEO Job Pool Index was created because while I thought the total CSA population was relevant (if all else is equal, a smaller population means less competition to get an SEO job), it didn’t give me a good idea of who these people were. Some places are going to have a greater share of people qualified for and/or interested in the SEO field.

That’s where this formula comes in. I first divided the two Indeed metrics to get the relationship between the demand for SEO as a job and SEO as a skill. I then applied that relationship to the number of people currently employed with an SEO title as a proxy to estimate the total number of people in the SEO job applicant pool. The only caveat is for cities where the SEO Title Job Openings were at zero or close to it – that either breaks the formula or makes an assertion on an extremely small sample size. In those cases (anything less than five), I just used the 75-city average ‘title’ to ‘skill’ percentage.

Are you confused yet?

In simpler terms, think of the SEO Job Pool Index as a major indicator of the number of people looking for an SEO job and the Population Index as a minor indicator.

If Population Index is ‘A’, Indeed Title Index is ‘B’, LinkedIn Job Index is ‘C’ SEO Job Pool Index as ‘D’, and with X being the calculation by CSA, here’s the formula I used. Everything in red represents the SEO job Supply and green the SEO job Demand.

X=((B+(C*0.1))/(D+(A*0.25)))*100

A couple more notes on how the calculation works: Everything above represents data pulled from a single day. Obviously these metrics are not static and change frequently. Because calculations for lower population areas fluctuated mightily from month-to-month 6, I made two adjustments so the rank was more representative of a typical day.

• I averaged the scores on May 4^th, June 3^rd and July 4^th.
• I applied a volatility penalization to those cities where the minimum score of those three months was less than 25% of that city’s highest score. The final score was multiplied by 0.75 in those instances.

If you’re still tracking with me, you may have some questions on the formula itself. I’ll try to address them here. And just to reiterate, please leave a comment on this post if you have any other questions or critiques.

Why was everything multiplied by 100?
Aesthetics. 185.58 is a more appealing score than 1.8558.

LinkedIn Job Index is counting people who currently fill SEO positions. Why am I using that to help measure unfilled/open SEO positions?
These positions are filled today but we all know turnover exists. And while most will leave one SEO job for another, resulting in an overall wash, some will either change fields or stay in SEO but move out of the CSA. Thus, if all else is equal, a region with more filled SEO positions will have a larger supply of SEO job openings, making it comparably easier to find SEO employment.

How were the 0.1 and 0.25 weights developed?
This wasn’t much more scientific than a blindfold and a dartboard, to be honest. Based on my previous explanations, I already assumed the Population Index should be less influential than the SEO Job Pool Index and the LinkedIn Job Index should be less influential than the Indeed Title Index, but how much so? Honestly, I fiddled with the numbers until the list ‘looked right’, which means unfortunately my own biases played a factor.

Great job. We made it through that section together. Now it’s time for the big reveal. Here are the top 75 metros to get an SEO job in descending order.