Google Analytics Source / Medium Report & User / Metro as Secondary Dimension

I had a client call yesterday which prompted the need to write this article.   The business is in the early stage of building an online presence.  While business has been strong through word-of-mouth, their goal is to reach new regions and demographics via SEO / PPC and/or Social Media.

Google Analytics was installed in March of 2022, so there is not yet a full year’s worth of data with which to compare Year over Year/ Month to Month traffic trends, so the question became “How do we know if we are making progress?”  My standard answer for this is, “Well, without Year over Year trends to account for seasonality, the only thing we can focus on, is gradual, incremental month to month improvement.   

High level traffic has been steady for the last few months with a few peaks and valleys, but nothing significant to speak of.  This is mostly branded keyword traffic, friends and family with a few non-branded outliers.   The previous agency had been sending over a monthly report with traffic totals, although its not clear where these numbers were coming from and lacked any deeper analysis.

Which brings me to the Geographic reports in Google Analytics.   For my own sites, and client sites, the Source / Medium report is the first report I look at, followed by the GEO report under Audience, or I choose User/Metro as the secondary dimension in the Source Medium report.  

There are countless other reports in GA which provide valuable information on other aspects of the site, however, this report is my Go-To, first report to understand the lay of the land for a new client, or to see what happened on the previous day/week for current clients.  

Under the User reports, you can also choose “City,” however this is confusing when you get traffic Columbus, Columbia, Athens, Paris or any other city name that exists in multiple states, as it’s not clear which state GA is referring to.  There is also a “Metro ID” option which is just a series of numbers and meaningless to most.  The “Metro” option is more descriptive as it will read as “Dallas - Ft Worth TX” or “Washington DC (Hagerstown, MD)”.  

Back to yesterday’s client conversation.   Once we drilled down into this report, we were able to see that most of the client’s traffic was coming from another state, that this business does not service.  In fact, Michigan is about 2,000 miles from the region where this client does business.  

With a few search on Google for the companies name and “Michigan” we were able to determine that there was a similar business with a similar name in Michigan and they were ranking pos 3 or 4 under the Michigan company in position 1, and so most of their Google organic traffic was coming from customers in Michigan looking for the Michigan company. 

Unbeknownst to my client and it’s CEO, these monthly traffic reports they had been receiving were a bit misleading as the traffic volume and fluctuation was driven largely by entirely irrelevant traffic and visitors who would never and could never become clients.   

While marketers are consumed with traffic volumes and fluctuations, the complexion, or make up of the traffic is often more telling than the traffic totals and can in fact, be misleading.  

The GEO reports are also beneficial for understanding traffic quality.  If I log in and see 100 visitors in 2 hours from Russia, I know that’s not real traffic and a bot snuck in around some filters.  If I see 2 visitors from Indiana, 3 from Arizona, 2 from Florida, 3 from New York, 4 from California and 2 from Texas,  I know that’s normal traffic.  It smells like normal traffic.  Businesses that operate nationally typically traffic by state proportionate to the state’s population.   More often than not, reports by state will lead with the business’s home state having the most traffic, followed by California, Texas, New York, Florida, Pennsylvania, Illinois and so forth.   When I see traffic patterns deviate from this, for example, disproportionate traffic from Massachussettes, I know something out of the ordinary happened in this time period.  Either someone mentioned the site in a blog, there was a news article in a town, or there was simply a flurry of word of mouth from a good, or even bad, customer experience.

Emperor Augustus & Your SEO Traffic

Executive Summary

Be careful with Monthly reporting. It can lead you down a path where the rubber doesn’t meet the road.

Below is the more Helpful Content:) It’s about a 10 minute read.

In 2019, the fourth Thursday of November fell on the 28th day of the month, pushing Sunday and Cyber Monday into December, the 1st and 2nd. In 2018, November 22 was the fourth Thursday of November leaving an entire week of post-Black Friday revenues attributed to November. When the busiest day in online spending moves into the next month of December, November traffic and sales can appear to be 15 - 20% worse than the previous year, while December deceivingly appears better, to the same degree, simply because of where the fourth Thursday fell in November. This shift will happen again November 2023 vs November 2024 and 2025, when CyberMonday will fall into December, 2 years in a row, which hasn’t happened since 2013 and 2014, when Thanksgiving also fell on the 28th and 27th, respectively.

10 Years of Thanksgiving

2016 - Nov 24

2017 - Nov 23

2018 - Nov 22

2019 - Nov 28 - Sunday & Cyber Monday Dec 1 & 2.

2020 - Nov 26

2021 - Nov 25

2022 - Nov 24

2023 - Nov 23

2024 - Nov 28 - Sunday & Cyber Monday Dec 1 & 2.

2025 - Nov 27 - Cyber Monday Dec 1

I observed this turn of events while optimizing SEO & PPC, in-house, for a fast-growing retail, E-commerce site. After the retail rush subsided, I kept coming back to this idea of how the days and weeks fall into the monthly calendar. I’ve played the in-house/agency/owner musical chairs for 22 years, and the one constant throughout, has been monthly and weekly traffic and revenue reporting. I felt compelled to dig deeper into this question of large % fluctuations occurring, due simply to the way days and weeks fall into months.

SEO Reporting - Weekly vs Monthly

Agency life revolved around weekly and monthly reporting for clients. Larger Fortune500 clients with high volume would receive weekly reports while clients with less volume would receive monthly reports. Each Monday began the iterative process of pulling data from the past week and comparing it to the week before that, or to the same week the previous year. Sometimes traffic was up and sometimes traffic was down. Each report would conclude with a few paragraphs explaining the “Why’s” of the traffic fluctuations.

At the beginning of my career, the idea of running weekly reports and analysis for clients was daunting. Populating impression, traffic, CTR, conversions and revenues for each channel, campaign or keyword every week, followed by a written analysis, seemed tedious. And it was. The monthly reports seemed tedious too, but there were only 12 of those a year, not 52. Over the years, however, I grew to prefer the 52 weekly reports. They made life easier. They also made the clients’ lives easier.

Over time, it became second nature. I had a system. I had my logins, my templates, my tabs and formulas where I needed them to quickly populate the data with a combination of semi-automated exports and copy and pasting. By mid afternoon Mondays, I knew where everyone stood. In the hey day of PPC growth from 2002 to 2007, more often than not, it was good news. Sometimes bad news, but at least I knew it before everyone else.

When traffic is down, as marketers, we instinctively begin to question the efficacy of a certain campaign, ad messaging, seasonality or perhaps a tracking tag was momentarily broken. When traffic is up, we are of course quick to attribute our victories to our ad copy, keyword research or audience targeting. The analysis would attempt to explain the “real” reasons for the ups and downs, in a few paragraphs.

SEO Data Analysis - Monthly Vs Weekly

When it came to the “Analysis” section of these reports, however, there were no short cuts and no copying and pasting. Each traffic or conversion fluctuation over a certain threshold needed a justification or explanation. Sometimes its a runaway keyword with negative ROI. Often times it’s an administrative issue such as a broken tag and missing data, or an expired credit card shut the campaign down. Perhaps the website was down for a short period for maintenance, or worse, a DNS issue that couldn’t be fixed for 4 days. Maybe its early November and retail season kicks in. Any or all of these things can be factors in traffic fluctuations in weekly or monthly reports.

Over time, the “Analysis” section of the weekly reports was much easier for me, than the monthly reports. For one, the more frequent processing of the data once a week simply cultivated deeper intimacy with the data. After month 4 or 5, I begin to notice when things simply do not feel right. “This Google Ads campaign has always had between 400 and 450 clicks, but this week has 750. What just happened?” Or conversely, “This organic keyword historically averages 150 clicks in a week, and only had 30 this week.” Did it drop in position in the SERPS? Was there a Google Algorithm update? Does the page have an error on it? Did we drop a backlink? Whatever the case may be, it’s much easier to pinpoint the culprit, knowing that it likely happened in the last 7 days since the last weekly report and explain it as such in the “Analysis” section of the report.

Writing the “Analysis” section for monthly reports is much more difficult. For one, it is trickier to identify the culprit in a 31 day window than a 7 day time period. Second, It is more difficult to remember what happened 3 weeks ago, than 1 week ago. Third, if the problem occurred in the first few days of the month, it means that you’ve gone 25 days without knowing that there was a problem, in order to fix the problem and you’ve lost weeks of performance. If you are lucky, the client didn’t figure it out before you did. If you’re best performing campaign was not performing for 3 weeks, don’t be surprised when you’re year-end, Year Over Year report is off by 6% from the previous year. 3 weeks/52 = 5.76% for those keeping track at home. And Excel will round it up to 6%.

Over the years, I began to realize that each month, year over year, had very slight but similar fluctuations every year, regardless of industry vertical or market strength.

The Moon, the Sun & Emperor Augustus

There was another reason monthly reports were more difficult. How many hours had I spent over 20 years tussling with clients over a 4%, or 8% or 16% sudden drop or increase in traffic from one month to the next? These conversations were much more difficult for monthly reports than weekly reports because weekly reports eliminate the uncontrollable variables caused by nothing more than the moon, and yes, believe or not, some Roman politics. We like to compare a given month to a previous month to show trends, which while worthwhile, carries the caveat that months have different numbers of days and encompass varying parts of weekly ebbs and flows, which for B2B, can be significant.

I’ll leave it to Wikipedia to explain the complete history of the Roman calendar, but the current number of days in the months was set in 46 BC, when Julius Caesar updated the original Greek lunar calendar to the 365 1/4 solar calendar we know today. He renamed the month Quintilus, July, (the 5th month from March, the original first month) since this was his birthday month. In addition, Caesar dictated that January, March, May, July, September and November would have 31 days (186) and February would have 29 days, or 30 every 4th year. This left April, June, August, October and December with 30 days (150). After Julius Caesar’s death, Augustus took over and the month of Sextilis (the 6th month from March), was named August, in his honor. It was later rumored that he moved a day away from February and gave it to August, so that his month would have the same number of days as Julius ( ie July with 31 days), but earlier documentation showed August already had 31 days. He also took a day away from September and November and gave those days to October and December, also rumored, to avoid having 3 “long” months in a row.

If your February traffic is down 10 percent from January, it is (at minimum) because February has 10% fewer days in it than January. (28 vs 31). So, as a baseline, March will be 10% higher than February, while April will be 3.2% lower than March(1/31 = 3.2%). July will be 3.2% higher than June. October will be 3.2% higher than September, while December will naturally be 3.2% higher than November, for which you can thank Emperor Augustus. Maybe.

B2B (Business-to-Business)- Weekends vs Weekdays

Monthly Report fluctuations are further complicated in B2B businesses who see higher standard deviations in weekday traffic vs weekend traffic. Conversion levels may see an even higher deviation. A common scenario, is that a B2B site may see 100 visits per day, Monday through Friday, and average 5 visitors a day on Saturdays and Sundays. 10 WEEKEND days a month @ 5 visitors a day is 50 visitors, plus 21 WEEK days @ 100 visitors(2,100), totals 2,150 visitors in the month. All else being equal, the same 31 day month with only 4 weekends, or 8 WEEKEND days, would total 2,340. (8*5)+(23*100)= 2,340. Then, 2,150/2,340= 91.8%.

As you can see, a 31 day month that happens to only have 4 weekends in it, as opposed to 5 weekends, can see an 8.2% increase in traffic, due solely to the way in which weekends fall into the calendar month. August of 2022, for example has 8 weekend days, while August of 2021, had 9 full weekend days and August of 2020 had 10 full weekend days. If nothing else changed for a B2B company with these numbers, there would appear to be a increase in traffic (4.1% from 20 vs 21, then another 4.1% increase from 21 vs 22), or an 8.2% increase from 20 vs 22.

Conclusion

The next time your client, boss or CMO asks why traffic and sales are off by 4%, 8% or even 16%, check your calendar first. You might be able to blame it on the moon, or Emperor Augustus.

How Stephen King taught me SEO

My secret to SEO?  Learn to write.

Demonstrating E-A-T (Expertise, Authoritativeness and Trustworthiness) on websites requires writing well. 

I have no idea if Stephen King knows anything about SEO.  What I do know is that his book “On Writing” inspired me to become a better writer, and in turn, a better SEO.

I read this book when it came out in 2010 and often reflect on his infinite imagination and how he puts it into words.  

With time and practice, you can write an authentic, helpful article in less time than it takes to run it through an AI machine and make the inevitable revisions.

Your inventory will be limited only by your life experiences and imagination.  

Then fill in your Title Tags and H1s. 

Here’s the Amazon link if you need a starting point.  It’s short.

On Writing, by Stephen King