Data Dive into Joseph Michael Photography

As part of my capstone project for York University School of Continuing Studies Certificate in Advance Data Science and Predictive Analytics, we did a deep dive analysis of a 3D modelling software company designed specifically for woodworkers. The analysis focused on the companies online sales, looked at historical product trends and our team made business insights and suggestions for more rigorous data collection moving forward. As a small business owner myself, I realized I had never actually done any real analysis of my own data collected over the years, and that it might be fun, and hopefully insightful, to take a look. This post is a peak into the limited data I've collected on my own company, Joseph Michael Photography.

This analysis is limited to predominately the invoices I've made since the company began in 2006 and looks at where the revenue generated over the years has come from, what trends the business has followed through its existence and perhaps dispel some myths or assumptions I've been holding about the business itself. I have some vague ideas, but never actually looked at story within the data. I also do a deeper dive into the mystery behind my dwindling wedding revenue, and introduce additional data looking at the relationship between my wedding inquiries, wedding bookings and online advertising. I'm going to visualize the data as much as possible and contextualize where I can what might be going on within the data. This is a business story of one small business, and the challenges and complications of being a freelancer in the ever increasing gig economy. There are too many factors to tell the full story, and definitely not enough data to flush it all out, so consider this a very high level overview. I consider this post a fascinating insight into where the company began, where it peaked, and where it was before COVID-19 decimated its revenue.

Welcome to the data story of Joseph Michael Photography.

A quick note, the first section details work done on the data including data migration and cleaning, which might not be super exciting for some, so if you want to skip to the pretty charts and graphs, I suggest beginning at the section "Data Analysis - Historical".

After many years of wandering the planet, dabbing my toe in the service industry and completing the coursework for my masters in photography, I started my company, Joseph Michael Photography, in 2006, and have been working solely as a photographer ever since. It's been an amazing career and I'm really proud of what I've accomplished, running a profitable business for 16 years!

I classify myself today as a commercial photographer, shooting primarily for businesses and non-profits, but I have spent a lot of time shooting weddings, family portraits, for arts organizations. When I started, I was primarily a journalist if you can believe it.

Throughout those years, I have kept a yearly database of every gig I've ever had, essentially a list of invoices, but I've never done a deeper dive into the details of the business. I of course knew the yearly revenue for tax purposes, and had a general sense of where the money was coming from, but it wasn't until my big data class last year that I had the skills to actually analyze it.

Every year since I began operating the business the 2006, I created two spreadsheets, saved on my computer, one recording income/revenue, one recording expenses.

Expenses

The expenses were typically logged at the end of the year, when I get all my receipts and throw them into a spreadsheet. What I have never done, and what all really good companies would do, is tie expenses to income and invoices, but the ship has long sailed for me to get that historical data, and so for the purposes of this post, there are no expenses factored into this reporting, with the one exception being my online advertising budget, which we will look at in more detail when we analyze my wedding photography history towards the bottom of this post.

The truth is, as a sole proprietor with a home office and no studio, my expenses have been relatively low. I do make capital investments, rent equipment for bigger shoots, and have to hire second photographers and videographers when appropriate, but the expenses were never too high. Over the years I have budgeted that 25% of my revenue goes directly back into the business and I used that percentage when figuring my pricing for gigs (this was my absolute highest threshold for expenses for any particular gig, but I rarely would get that high). For the vast majority of my gigs though, I had little to no additional expenses, and the business income would go into general operating expenses, advertising and eventually bigger capital expenses.

Income/Revenue

As for the income/revenue data, this is what I will mostly be analyzing in this post. I don't give actual numbers and figures, that's a little too personal, but I can tell you the business never made a super big heap of money, but the business was successful, that is, up until COVID-19.

Back in 2006, I started simply, by collecting only a few data details, specifically, the Date (of gig), Invoice Number, Client (name), Amount, GST, (amount) Received, Comments. The comments column was a generic column to remind me what the gig was for, for example, if it was a portrait, a conference, a band, or if I was paid in cash. I would log this information in a table in word processor and would have a new file for every year of business. The files were in no way linked.

In 2010 I decided to included a few more data points, information such as month booked* to distinguish when I booked the gig versus actually photographed the event and I also added the invoice date, more to keep track that I had actually billed the client. I also tried to somewhat codify the comments column, knowing that I would one day like to actually distinguish income from different sources (see image "Original Data - Sample from 2014" below).

By the time the pandemic hit, and I knew I wanted to do a deeper dive into the history of my company. I had a heck of a lot more time on my hands and I decided to uniformly input all the income data I've collected over the years into one spot. Plus, I wanted to keep all this data in one file on the cloud. I started using Google Sheets and migrated all the revenue data I had in all the individual yearly files, expanded the data by adding new columns, cleaned up the data, and then linked all the essential data I wanted to analyze.

Then I threw all that into the amazing data visualization software Tableau, and voila!

In the process of migrating the data, all of the columns I had previously collected were copied and pasted into the new spreadsheet. I then added new columns to better define my historical data for analysis and other columns for better tracking of information I might want to track moving forward. As will be discussed, the most important columns I added were Client Type and Gig Type, but I also added columns like hours shooting per gig (column W) and location (column X). I did not have not have the energy or time to put all this data in for every gig going back to 2006, I may someday, but at least I can use that data moving forward. Among other things, I also included a new column called Date Paid (column E) to identify trends in timeliness of clients paying in relation to invoice date or gig date, but again, I don't have the ability to find all this data on historical invoices.

To get this older version of my data into a new and improved version of the data, I stated by simply copying the data from each year into new tabs (sheets) in a single Google Sheets document.

Invoice Number, Invoice Date, Gig Date along with Amount, HST, Total and Total Received were all copied directly into the new database. The Client column was copied into a new column called Gig (column I) and I populated the new Client column (column F) with a more streamlined version of the client name (described below). I began describing the gig with a new column called Gig Description (column V) which is a more thorough column, capturing all the aspects of the gig, but decided that defining my Client Type (column Y) and Gig Type (column Z) as simply as possible was the most useful way to distinguish the income sources.

For the purposes of this analysis below, I am going to analyze only a handful of the above columns:

Column A - Date Booked (month/year)

Column B - Invoice (number) - Linking identifier used to link all the columns information

Column D - Gig Date (month/date/year)

Column J - Amount (currency, CAD)

Column Y - Client Type (NEW)

Column Z - Gig Type (NEW)

Data Cleaning and New Columns - Client Type and Gig Type

Though I did have to go back and reclassify client type and gig type by hand, the data was in pretty good shape. The one major cleaning step I had to do was to clean-up the Client column (column F). Although I don't get into the analysis of this particular column in this post, you can imagine how important identifying specific clients would be in my overall business understanding. I had been using the client column in the past as a gig identifier, so I would use things like "Emily and Nick Wedding" or "Alpha Company - Factory Equipment" for that column. I realized pretty early, that for my new Client column (column F), I needed to streamline my naming. For example, I lumped all my wedding clients into one client called Wedding, all family portraits or family events into one client called Family. As for the corporate or non-profit clients, I got rid of the gig identifier and named the column just on who I was invoicing for the gig, so both "Alpha Company - Factory Equipment" and "Alpha Company - Executive Portraits" would become Alpha Company. That moved the gig identifier to the Gig column (column I).

As mentioned above, the most important addition to the data was taking the information from the client and comment columns from the old data, and identifying two specific traits, Client Type (column Y) and Gig Type (column Z). As mentioned, I began by making a more thorough description of each invoice, listing the different features of the particular gig in a new column called Gig Description (column V), but realized later that I was going to need two very streamlined and very distinct columns, Client Type (column Y) and Gig Type (column Z).

The Client Type (columns Y) was defined by the major categories of clients that I have, and I distinguished them, most commonly, as company, non-profit, wedding, family, arts*, journalism and individual*. If the gig was canceled, I wrote the word dud.

The other thing I wanted to distinguish is the Gig Type (column Z), more specifically, if it was a portrait, conference, wedding, prints, headshot, portrait, family event or something else entirely. You can see in "New Data - Sample from 2014 - Google Sheets" that rows that are red in Client Type (column Y) for wedding, there are four different kind of Gig Type (column Z), Disc, Prints, Wedding and Engagement. In this way, I would know what income I was making from specifically wedding clients, I would include all these different gigs, but I wanted to know what percent of my wedding income was coming from not actually shooting a wedding, for example, making prints, I could look into that as well.

It's important to note though that for the gig type, I had to choose the dominant distinguisher for that gig. For example, some wedding clients included an engagement shoot or prints or an album in their wedding package. In that case, although the wedding and the engagement shoot might be together in the same invoice number line, and I would classify that as a Gig Type wedding, even if other things were included with it. Only if it was a solo engagement shoot would it show up as engagement.

In summary, Client Type is the type of client who is paying for the gig.

The Gig Type is the actual function that I did to earn the money.

After inputting all of the data from all my individual spreadsheets into one Google Sheet document, adding the new columns and entering the new data, I had invoice related data for all 16 years of the business (I'm not including 2022 in this analysis) in individual tabs within the same document.

Linking the Data

Having the data compartmentalized into years makes sense, but to actually do the analysis of the data, I needed to bring it all together... link it somehow. To actually do this, I need to bring all the data I want to analyze and compare it to the same thing, the identifier of each of these gigs, the invoice number. The invoice number is the one distinguishing factory that identifies all the rows specifically across all the years, and I can compare what went on within that invoice number row with all the other invoice number rows. For other companies, a person might want to link unique client numbers, or unique product codes, but in this case, I am looking at the specific invoices.

To do this, I created a new sheet for each category I wanted to analyze, then populated it with all the invoice numbers in a column, and put that beside all the information from a column I wanted to analyze. For example, looking at Gig Date (column D), I inputed all the rows from all the different years in all the different sheets in the google sheet document, and put them into a new sheet using the function below.

While the function looks complicated, it basically says to copy the data from the rows specified in the D column (Gig Date) from each of the individual yearly sheets in the document, and put them all into a new column called "ALL GIG DATES". For 2013, you can see I asked the function to bring in only rows 2-108 from column D, row 1 would be the header, and I don't want that, and I didn't have any invoices after row 108.

I did the same thing for all the invoice numbers, making a new sheet with a record of every invoice numbers and its corresponding gig date (as you can see below).

The reason this step is so essential, is that I can actually make adjustments in the original document, fix data that was wrong in my 2012 sheet, and it will automatically update the master list I've created here. If I instead just copied and pasted the data into a master list, I would have to change both the original and the master list, and that could get complicated quickly.

Test Drive - New Columns, New Information

To see if the data link works, if things make sense with what I've created, I wanted to start pretty simply, and thought I would see what I could see from the two new columns I had created, Client Type and Gig Type. Not only would it be a test drive of the data, but I could learn a whole lot very quickly about my business, something I had always wanted to do. I had some basic questions, how many family portraits had I actually taken? What percentage of my corporate gigs are for events or for portraits? What have I actually been taking photos of all these years? To have a quick look, I plot the Client Type in one axis, and Gig Type in the other, and did a count of all the gigs.

If I wanted to visualize this data instead, I could put the same information into a colourful bar chart below. You can quickly see that the most common gig I've ever had is shooting a wedding, and I've done that 225 times, followed by company event, 217 times, and journalism (specifically for a paper, not a blog) 136 times. As I haven't shot a journalism gig in over 5 years, this was a bit of a surprise. What this chart doesn't account for, and you can see in the chart above, is that I've shot 159 gigs specifically for portraits across all the Client Types, identified as light blue in the chart below, but not totalled.

Historical View - Number of Gigs

From the summer of 2006, when I started my company, to the end of 2021, I photographed 1225 gigs. I call them gigs for the purpose of representing what I am, a gigging freelance photographer, but that 1225 actually represents invoices that clients have paid and does not include gigs that were canceled, and not delinquent clients that didn't pay (there aren't that many, but it happens). It should be noted however, that the number of actual gigs is considerably higher, something like 20% higher, as there are times I invoiced a client for an entire marketing campaign, and might have had 10-20 different shoots within that one campaign, but for the sake of this post, when I refer to the term gig, I really mean invoices, I just think gig conveys the industry lingo a little better. I gig.

And so, already, I've reduced my business to a clean and simple 1225 gigs. It might now sound that high, but it's a decent amount. The pandemic certainly had something to do with it being smaller than it could have, but I'm not a large company, and trust me, there were times it was about as much as I could handle. That being said, there were a lot of times I wasn't as busy as I wanted too, but there it is.

Truth is, as a photographer, there are at least two parts to the job. One is taking the photos, and the other is editing them. I've kind of figured over the years, for every hour of shooting, I likely have two hours of editing to do, if not more. I've shot a lot of full day weddings over the years and dozens of multiple day conferences, so the hours add up pretty quickly. And again, this isn't really the number of gigs, but invoices. Still, let's just keep calling them gigs.

So how many gigs did I have each year? When have I been busiest? What has COVID-19 done to the business? The best way to answer these questions is to look at a histogram of the number of gigs I shot per year.

Context is important, and to best understand where this company began and where it is now, histograms like this tell a lot. I started the company in the summer of 2006, but I spent the last three months of 2006 and first three months of 2007 in Africa, so that delayed the quick start of the company. I also spent four months living and working in Australia in 2009, and a month in India in 2013 (honeymoon), when I did not have any gigs, and I think that is somewhat reflected in the data.

Clearly, you can see, COVID-19 hit the business really hard. It's not too surprising, almost every photo I take involves lots of people gathering together in a room. While I did shoot 8 gigs before the middle of March in 2020, I lost pretty much everything else I had booked for the rest of the year, including ten (non-wedding) gigs I had booked and were in the calendar. The next year, 2021, wasn't much better. There were a few outdoor family photos and business portrait here and there, but COVID-19 pretty much destroyed the business as I knew it, and I pretty much became a full time dad. It is what it is, I don't have a lot of regrets, I'm just giving you a bit of context.

One other thing I thought I would point out about this chart is that the 2015 data looks hidden as that line is so straight. In 2014 I peaked at 147 gigs, then had 122 in 2015 and 93 in 2016.

So 1225 gigs over 16 years is on average 76.5 gigs a year. If I take out my first two years of business, and the two COVID-19 years, that would be 1133 gigs over 12 years, which is an average of 94.5 gigs a year. That's a busy photographer.

Gigs are one thing, but what about revenue! How has the business income fluctuated over the years?

The chart looks very similar to the gig chart, but maybe there is something to learned by seeing the actual relationship between revenue and gig count per year. Is there a direct comparison between number of gigs per year and revenue?

The remarkable similarity between gigs per year and revenue perhaps isn't that surprising, but there are differences. The major ones I see is the larger drop in revenue in 2013 then number of gigs, and the seemly very efficient 2012, 2015-2018. For example, while I shot more gigs in 2014 than in 2015, revenue in 2015 was considerably higher than 2014. Same with 2018 compared to 2019, where revenue was essentially the same, but I shot a lot fewer gigs in 2018.

Most of the gigs in the early part of my career were journalism gigs, so I was working a lot of gigs that didn't pay particularly well, so I'm not surprised gigs outpace revenue at the beginning. My prices were also a lot lower at the beginning, so the revenue I made per gig was a lot lower.

Photography, like a lot of businesses, is quite seasonal. Winter months can be a little challenging as there are a lot fewer people getting married or looking to have family photos. The off season though is a good time to book gigs for when the weather is much nicer. Factors such as yearly budget surplus and holidays clearly play major a roll in when I actually book a gig compared to when I'm actually shooting. What months, across all the years, did I book the most revenue? What months did I shoot and actually make the money I booked earlier in the year? Below, you will see the difference.

Best way to look at this is to see on the chart on the left as the revenue for each month that I either signed a contract with a wedding couple or was told to hold a date firm from a company, non-profit or arts organization. The chart on the right is the revenue for each month that I actually completed or executed the gig/contract. Chart on the left is money promised, the chart on the right is money paid. Either way, you can see December is pretty quiet.

It's not a surprise the spring is a busy time for booking gigs as yearly budgets are usually being finalized that time of year, couples are finally making their wedding photography decisions and businesses are finally waking up after the sleepy winter months.

When I look at the finer details by specific Client Type (not pictured in a chart here) , I see that July is also a pretty quiet month for actually shooting company gigs, likely a lot of vacations and people enjoying their summer away from work.

What about the ebb and flow of the wedding season specifically? I know the summer is a popular time to get married, but what month is most popular? Let's have a deeper dive into that below.

Variety is the spice of life, and I have been fortunate to have a lot of variety in my career. Looking directly at one of the new columns I created for the new database, Client Type, this is immediately clear. Just a reminder, Client Type is loosely based on what kind of client is paying the bills. I have identified a number of client types in my career, but mainly focus on the top six, arts, company, family, journalism, non-profit and wedding. I have also clarified another client type as "individual" when someone unassociated with an organization would hire me for either a portrait or non-wedding or non-family related event. I've only had 26 gigs like that, so it doesn't happen that often. I also have a category I call other, which includes gift certificates, teaching gigs and publications. In this client type category, there have been only 20 gigs, so both individual and other contribute pretty insignificantly into the overall revenue as well, as we shall see below.

On the left, the total number of gigs I've worked for each Client Type, and on the right, the revenue I've made from each Client Type, where the size of the bubble represents the proportionate amount in revenue in relationship to the other Client Types. So while I've shot considerably more gigs for companies than for weddings, as you can see from the bar chart, the revenue for weddings is significantly larger than for the revenue I've made from companies.

Another way to compare the the number of gigs versus the amount of revenue is by plotting them both on different axis. See the chart below. From now on I'm only going to look at the top 6, Wedding, Company, Journalism, Non-Profit, Family and Arts.

This graph ties the two charts above it together, with the revenue on the up down axis, the higher the dot, the more revenue was made, and the right/left axis shows the number of gigs I had for each Client Type. You can see I've shot a lot more company gigs than wedding gigs, but made about 60% less. You can also see that there is a cluster of Client Types that I've shot between 100 and 150 gigs, with varying degrees of income. The journalism gigs clearly paid the least, and shooting for Arts and for Family clients are about the same.

One note about this chart is that I most often shot a large number of gigs for one invoice with Non-Profit organizations than with any other Client Type. As my invoices, not actual shoots, is represented in the chart above, I'd be willing to bet that if I actually counted the number of shoots I've shot for Non-Profits it would be somewhere around 200-220 gigs, putting them more in line with the line somewhat created by connecting Arts, Family, and then way on the right, Company.

So you can clearly see that I have a variety of revenue sources and shoot. That's a snapshot of the business as a whole, but the best way to see the actual story and history of my business, we should look at the revenue generated by year by Client Type. Here you can really see the changing nature of owning a business, and while there is a lot of things you can control, there are a lot of things you can't.

There is a lot to unpack above. Take your pick, some people love to see the data visually, and you will really enjoy unpacking all there is in the bar chart histogram. For those who mostly like raw numbers, the highlight table below the bar chart shows the percentage of yearly revenue based on the 6 most popular Client Types.

Let's first look at journalism, shown in pink in the bar chart on the top. You can see a pretty significant journalism presence early in my career by 2008, but dwindling slowly each year, until my last paid journalism in 2016. I only had one that year. In 2008 and 2009, journalism accounted for over 17% of my total revenue each year. The gigs never paid very well and they were a lot of hard work. I do miss them kind of, they got me out into the community, and was happy to say I actually have had a journalism gig this year, in 2022, but it was very clear that my business would be better served focussing on other avenues for revenue.

Family and Arts have been pretty consistent over the years and I have a number of great clients in both categories who have called me back for years to shoot for them again and again and for that I am very lucky. COVID-19 though did have a huge impact on the Arts, the one gig I shot for that client type was pre-pandemic in 2020 and I haven't heard from them since. I should point out that all my clients in the Arts are audience driven, and so, large groups were not an option during the lockdowns of COVID-19.

As for my Company and Non-Profit clients, there are many stories to tell. Clients come and go, big projects and small projects, but there are a couple of significant moments that really stick out in the charts. I worked with a corporate agency for years, from 2008-2015. They had a number of great clients and hired me for almost all of their photography needs. That agency essentially folded in 2015 and I never worked with them again. That was a huge hit. You can clearly see it in the data, when my company revenue really took a hit in 2016, and hasn't fully recovered. Ouch.

Fortunately, I picked up an amazing Non-Profit organization in 2016, the very next year, that needed a lot of marketing materials, and I got a lot of work with them, and worked with them a lot, more and more every year, until the pandemic hit in 2020.

Diversity of revenue is always an advantage, for any business, but it's hard not to become a little more dependant one company that keeps calling you and continues to pay well. That client becomes the priority, and you'll turn down other gigs, even stop pursuing other clients, not because you want too, but because you are busy, and money is coming in. When you loose a client like that however, and it happens, things can really change.

The other Client Type we really need to talk about though is my wedding clients. As you can see above, that's all over the map. I have a lot to explore about my wedding data, and fortunately for all of us, we get take a deeper dive into everything weddings below.

Number of Weddings

Wedding Count and Revenue Over Time

So, while I began a little reluctantly photographing weddings, I soon found out I was both good at shooting weddings, and thoroughly enjoyed them. Plus, they were a good revenue stream early in my career. As we shall see in the data below, that stream eventually really dried up, and I hope to look into a some of the factors, internally and externally, that might have played a part.

On the left is the number of weddings photographed per year, peaking at 28 weddings in 2010. On the right, the bar chart shows the revenue per year from weddings alone. Unsurprisingly, the count and revenue makes a strong correlation, mimicking the peaks and valleys almost exactly. I've had a thought as to why there is such a drop in 2013, and beyond the fact my wife and I went to India for about three weeks for our honeymoon, I couldn't think why the number of weddings and revenue was so down that year. I remember we also had a boat load of friends and family getting married in 2013, and I blocked those weekends off. Other than that, I'm kind of stumped.

But 2013 aside, what really began to worry me over the past number of years, even before the pandemic, was how much my wedding revenue was going down. Sigh. There are lot of factors here and I will not be able to solve this through data or discussion alone, but I wanted to have a look and see what we could find.

Wedding revenue peaked in 2012, and I was mighty close to that peak in 2014 and 2015. The drop after that was not planned. Let's have a look at year over year wedding revenue since the business began in 2006.

Data looks so different when you present it differently. Wow. The above chart shows the year over year of wedding revenue starting in 2007, a year after starting the company. The huge growth in 2008, 2010 and 2011, when I finally decide to shoot weddings as a major revenue stream, should not be a total surprise. I'm back from Africa in 2007 and Australia in 2009 and dedicated a lot of energy into booking weddings. While there is a dip in 2013, as explained above, the growth continues in 2014 and 2015. What was concerning, and to this day is quite frustrating to understand, is the consistent and significant drop since 2015. Even going into COVID-19 I had recorded four consecutive years with year over year drops in wedding revenue, including my largest pandemic drop in 2019.

What was going on here? There are a lot of variables, and it's important to look at as many as you can to start eliminating the ones that are likely not a factor, and narrowing in on others that might have had a bigger impact. One factor I realized I might want to look at is pricing.

Why could prices impact my revenue? Well, imagine I raise my prices too much too quickly, and people decide that I'm too expensive a wedding photographer for what I offer and how skilled and personable I am, then I might be booking significantly fewer weddings, and my revenue would go down. Wedding photography is an ever competitive industry and finding the right price is challenging. All business aim for a sweet spot, where they are charge the amount people are willing to pay, not everyone, but enough people, to increase or maximize revenue.

As a gig business, one of the greatest challenges is setting your own price, telling the world how much you are worth, hoping they will pay you for your services. There are a million different business strategies on this, but I can say, I was cheap when I started, and become very reasonably priced as I became more mature and a much better photographer. I valued my skill and time much more once I got confident in my skills, but any market analysis could show you pretty quickly, I'm still far from an expensive wedding photographer. While I roughly tripled my average wedding price from when I began to what I was charging in my peak year, I was a 100x better wedding photographer.

Regardless, price could be a factor, let's look at my average and median wedding price per wedding each year.

On the left you can see my average wedding price by year. It looks as though the graph goes steadily up until it peaks at 2018, then I dramatically reduce my costs. I know this was not the case, so perhaps median price would be a better representation. The chart on the right gives the median price per wedding by year. Median price reduces the influence of outliers in any particular year and when you look at that chart, it looks like I keep raising my prices, with minor exceptions, until 2020, then dramatically reduce my prices. I only shot three weddings in 2020 and only one in 2021, so we can kind of ignore the data from 2020 and 2021, as there isn't a large enough sample size.

Honestly, I don't know how my wedding prices could have had a major impact on the wedding revenue going down so much. The median price from 2009 to 2012 went up 124% over that period and my revenue was climbing steadily. The median price of a wedding from 2012 to 2018 went up only 13%, and revenue stayed pretty consistent, then began dropping significantly. I don't think that 13% had a huge impact. My prices were relatively flat from 2012 to when the pandemic hit, yet my revenue was way down.

Still, can we rule out the wedding prices specifically? Let's look at what's going on in even more detail. To do this, I decided to plot all the wedding prices I've ever shot using a box and whisker plot based on the year I shot the wedding. If you have never seen one before, it can be quite odd looking, because there is a lot going on, but this chart gives a much more detailed picture of what's going on in any particular year as you can see the individual data you're trying to understand.

I'm not going to give a thorough explanation of a box-and-whisker plot, but basically, what you are looking at is all the individual wedding prices I've shot, all 225, plotted out for each particular year. So each one of the dots represents the individual price paid for a particular wedding. You can see the most expensive wedding was in 2012, all the way at the top, and the lowest was in 2008, at the bottom. The actual box and whisker thing that you see in a mathematical calculation determining the interquartile range. "What's that?" you might ask. Good questions. Interworks does a pretty good job defining it:

"You start by determining the median of the data set. That is where the box turns from grey to light grey. Then, the upper and lower quartiles are determined. These are simply the median of the upper half of the data and the median of the lower half of the data. That forms the “box.” The maximum of the data set is the upper range while the minimum of the data set is the lower range. That forms the “whiskers” of the plot."

If you look closely, you can see the chart (above on the right) of "Median Price Per Wedding by Year" matches the median price in the box-and-whisker plot. Outliers are represented by a single dot outside of the box and whiskers and don't factor into the calculation of median. The outliers are considered too far outside of the data to let it be factored in to the maximum and minimum plots. The super high dots are weddings that would have included wedding videos, second shooters or albums to the cost of their wedding, and the lower dots would be reduced cost weddings for friends of friends/family or 1-2 hour weddings that just wanted me to show up for the ceremony and a few family photos.

Another way of looking at the box-and-whisker plot is that if the box is short and squat, there is a larger cluster of data points that are all gathered close together. Looking at 2012, almost all the wedding prices were clustered in the same area, so there were a lot of weddings all about the same price, and the three high priced weddings and one low priced wedding were so outside of this major cluster they they didn't factor into the median calculation. Conversely, if you look at year like 2019, which only had seven weddings to consider, there is no tight cluster, and the box-and-whisker included all plots as a part of the calculation. They were all statistically significant because there was no core cluster. Same could be said for 2020, but with even fewer weddings shot that year, only three.

This plot more clearly demonstrates what was going on with the dip in median and average price of a wedding 2013, something I was trying to figure out earlier in this post. With this box-and-whisker plot, we can see major cluster in 2013 is positioned slightly lower that 2012 and 2014, and I had a number of seriously reduced weddings that year, and only one elevated.

It's quite different looking at 2017 and 2018 as those years I didn't shoot any significantly reduced priced weddings, hence, the significant jump in average price of weddings for those years and minor jump in median price.

From everything I can look at here, while there can be major differences in the price in any one given wedding, the clustering that I had established, slowly rising, but not dramatically, from 2011 to 2018, would not make a significant impact on the revenue going down as much as it did. Especially because, like I mentioned, I knew my prices were very reasonable, especially for a seriously good and seriously experienced photographer :) I would like to officially eliminate pricing as having a huge factor in why my revenue went down.

Outside Factors

There are so many outside factors that I am not recognizing in this analysis, and for that, it's not entirely sound, and I get that. I acknowledge that it is entirely possible that my prices remaining relatively flat from 2011 to 2018 failed to recognize that the wedding industry had actually changed, and all the other photographers had dropped their price by 50% over that time to compete in the increasingly competitive field. If that were true, then my high prices would have been absurd and of course no one would have hired me, and that's why revenue went down. While it is possible, it didn't happen. My prices remain competitive in the industry in Toronto.

Still, there are a lot of other factors that would have an even more significant impact than price. This is a quickly changing industry. When I started, there were still wedding photographers using film. Now, everyone and their brother has access to amazing digital cameras, inexpensive professional cameras and multiple lensed phone cameras, and the impact of this could have, or will inevitably, hit the wedding industry dramatically.

Also, aren't fewer people are getting married today? It seems like it. Plus, those that do, aren't necessarily planning the same weddings people had ten years ago. I get a lot of requests now for casual cottage weddings, weddings at a small restaurant with a few friends. Gone are the days when everyone getting married is spending a boat load of money on a big reception hall wedding.

And while my wedding prices might be reasonable within the industry itself, what I might not be factoring in, because I don't totally have the data, is that there are fewer and fewer people actually willing to pay those prices. There are a lot of outside factors, but I don't really have the time to do all the market research myself, or money to higher an outside consultancy company to do the work for me. It's tough owning a small business, you're just kind of winging it most of the time.

Oh, and let's not forget another major factor outside of the business, my wife and I had a son in 2017. Looking at my life pre-kid and post-kid, I realized I much more thoroughly valued my weekends after the birth of my son. After his birth, I did tell myself that I was going to make more of an emphasis on company and non-profit gigs, and choose to shoot fewer weddings. Shoot fewer weddings, eventually, but the drop in wedding revenue happened way earlier than I was planning or intending.

And so while there is always outside factors that can impact revenue, some we might, and most we don't, have data for, I had a feeling there was something else.

Inquiries and Bookings

One other data point that I was collecting throughout most my peak wedding years, a data point that I haven't talked about until this point, was my inquiries. Specifically, the number of inquiries I was getting each year for weddings.

Before I get into the details of this section, I must point out that I am aware that there is a lot of other ways to track all this information below. It is absolutely true that Google Analytics has a lot of amazing features, there are email forms that automatically populate databases, I could have set-up conversion tracking and found out what my click-through-rate is for specific online advertising. Yes, but also kind of no.

I did it the old school way and started my system in 2010. I only had a partial year of data collection that year, so everything you find below starts in 2011.

Back in 2011, everything was a lot less mature than it is now when it comes to customer tracking. While there are a lot of better and more automated options now, I started mine this way, my system was good, my business was small with not a lot of overhead and I did what worked for me. Regardless, bare with me, there is data here that is useful, I just got it differently than maybe a business would get it now, and while the data I do have is limited, it's still valuable.

I should note, in this section, I treat my bookings a little different. If a bride or groom reached out on August 13, 2013, like they did above, and then we met on August 25th, which we did, and booked a wedding for October 11, 2014, then that was a 2013 inquiry and 2013 booking, even though the wedding is in 2014. That would be the same for if a bride and groom made an inquiry for my wedding services on December 10th, 2013, I met with them on January 10th, 2014, and I booked their wedding for February 10th, 2015. That would also be considered a 2013 booking because that was when I got the actual inquiry. The booking date (the date the money was promised and contract signed) would be the date we signed the contract, which would be 2014.

The reason the booking date in this section needs to be tied to the inquiry is because I'm going to tie the money I spent on online advertising. If I spent money in 2013 to get the inquiry for a bride and groom, I want to tie that to the bookings I made with those brides and grooms, even if I actually sign the contract with them later, like in 2014. It takes a while from actual inquiry to contract signing, but this method of tracking the initial inquiry ties the advertising money spent to acquire them, regardless of when I actually booked them.

So, using this system, what did my inquiries and bookings actually look like from 2011-2019 (pre pandemic). Here is the graph below, plotting both the bookings and the inquiries.

A relatively simple looking line graph tells me a whole lot of my wedding revenue and business over the years. Holy moly. First off, you can see how closely the number of inquiries and the number of bookings are tied together. Get a lot of inquiries in a year, get a lot of bookings. Don't get a lot of inquiries, don't get a lot of bookings. For the record, and it's hard to read in the chart above, but I had only 18 booking in 2012 and 10 in 2018. All the other numbers I think you can read.

So, looking at 2012, did we just solve a mystery? Did we just figured out what happened to the revenue in 2013? I had both considerably fewer inquiries and bookings in 2012, which would have most significantly impacted 2013. Maybe, above everything else I've speculated on before this, simply fewer inquiries in 2012 had the most significant impact on the number of weddings and revenue from weddings in 2013. Hard to argue with that.

But what happened to my inquiries after 2016? After a decent amount of consistency between 2013-2016, inquiries drop off a cliff. Bookings are a little more gradual, but down they go nonetheless. After getting 137 inquiries for weddings in 2016, I'm down to 24 total inquiries in 2019. What happened?

This chart is remarkable because of the implications for my business. I'd even wager to say that inquiries are the most essential part of my business, the more I get, the more I book. This isn't a shocking business statement I'm making, it's true for any business, but looking at the chart, it's fundamental.

Still, I have to wonder, is there something I have altered over the years that might be effecting my pipeline, something that has changed from how I respond to inquiries in order to get a meeting, something I have changed during my wedding meeting that might be preventing me from making as many bookings? I can't think of anything drastic, but the numbers won't lie. Across the nine years from 2011-2019, I had 940 inquiries. What did I do with all those inquiries and what percentage turned into a meeting, and what percentage of those meetings turned into a booking?

Let's have a look at the numbers.

Now that we have eliminated my pipeline, mainly, my response to wedding inquiries to book a meeting and my appearance and presentation skills at meeting as factors that might play into my bookings in any particular year, it returns me to the bigger question, maybe inquiries alone is the most important factor?

Let's have a closer look what kind of correlation there might be between wedding inquiries and wedding booking.

In the chart above, I've plotted the bookings and the inquiries from each year in the x and y axis, but instead of showing the histogram by year, 2011, then 2012, then 2013, this chart shows where each plot is in relation to each other in a linear fashion.

So, for example, in this chart, 2013 is the dot on the very right. That year I had 179 inquiries and 33 bookings. The point to the left of that is 2011, when I had 35 bookings and 157 inquiries. Each year has been plotted, showing the relationship between number of inquiries in a year and number of bookings in a year.

The line going through the chart represents the trend line, and you can see how close each plotted point is to that trend line. This line actually comes up with a calculation to determine how many bookings you would get if you had any number of inquiries (you can see this in the box in the middle of the chart), Bookings = 0.167052 x inquiries + 2.33009. So if you had 250 inquiries in a year (0.167052 x 250 + 2.33009), you could reasonable figure you would book approximately 44 weddings. The 2.33009 is the intersection line of 0 inquiries at the Booking axis, and the 0.167052 is the slope of the line.

How accurate is that calculation? Well, the R-Squared value tells us it's very accurate. The R-Squared value of 0.859034 is statistically significant, meaning the model, the trend line, is strong, and predicts the variables well.

Another math calculation you can do is the Peason's r (not in the box on the graph above), also known as the Pearson correlation coefficient, or simply, the correlation coefficient, is used to calculate the linear correlation between two sets of data, and is calculated by taking the square root of the R-Squared value. The Peason's r value would therefore be 0.9268225. That is a really strong correlation.

P-value (the last value in that box) is another mathematical figure, and it is used to determine the strength of evidence against the null hypothesis. The null hypothesis essentially asks, what is the probability of the observed results arising by chance. The threshold for P-value is usually, 0.05, and anything lower than that rejects the idea that the data arose through chance. In this case, the data has a P-value of 0.0003245, meaning their is a 0.003% chance the data came about by chance. Very unlikely. This supports strongly that this data suggests there is not enough randomness going on (null hypothesis), and we can embrace our alternative hypothesis. Our alternative hypothesis is that the number of inquiries influences the number of bookings.

I am not a mathematician or statistician, you can quote me on that, and some of what I say above might even be slightly wrong, but I do know that the math points to what I kind of already knew as a business owner, inquiries are important.

If inquiries are so important, how is a business going to increase the number of inquiries they get each year?

We have finally come around to discussing my online advertising budget, my website SEO, and difficult business decisions a person has to make when they own a small business. It's not pretty, I'll admit right up front, but let's have a look.

Online Advertising - Weddings

One of the fundamental ways I tried to increase my inquiries, or at least attempted to, was through online advertising. I started advertising online in or around 2009, but the first year I have complete wedding inquiry data and full online advertising spending figures is 2011.

A note about my online advertising budget. I've tried multiple ads over the years, on multiple platforms (mainly Google, but also Bing and Facebook), and no, this post does not look into the success or fail of any particular ad or platform in relation to other ads or platforms. I'm simply curious if there is a relationship between my online advertising spending and the number of inquiries I got in any particular year. As always, there are other factors to consider, but as I was spending a sizeable amount of money on advertising, I want to know if my money was going to good use.

In the above chart the blue line represents the number of wedding inquiries I received each year, and the red line represents the amount of money I spent on online advertising each year.

In 2016 I spent $6,848 CAD on online advertising and received 137 wedding inquiries. In 2015 and 2017 I spent roughly the same amount of money, $5500, but had an entirely different number of wedding inquiries, 135 in 2015 and 61 in 2017. Ouch.

As you can see, I decided to spend less money on online advertising in 2018 and 2019, instead focusing my money on other, what I hoped was a more beneficial investment. We will get to that shortly.

Regardless, from what I can tell looking at the two lines, wedding inquiries and money spent on online advertising, I'm not seeing a whole lot of correlation. I would like to think that if I pump more money into my advertising, I'm going to get more inquiries. Is that's actually what's going on? Let's have a look at the math.

There is almost no correlation between the money I spent on online advertising and the number of wedding inquiries I was getting each year. How do we know this? Again, lots going on here, but the low R-Squared value of 0.0797456 tells us the trend line does not predict the values very well, and the correlation coefficient is 0.2824, which is way too low. Remember, the correlation coefficient between inquiries and booking was 0.9268225. Similarly, the P-value, which is way above 0.05, suggests that we can not reject the idea that the numbers are actually entirely random.

So if there is not a correlation between spending money on online advertising and getting actual inquiries, what was my advertising money actually doing and how much of a factor was it? I would love to know.

Cost per Inquiry and Booking - Weddings

Not only was there no correlation between how much I was spending on advertising and how many wedding inquiries I was actually getting, but even worse, the cost per booking and inquiry I was getting was only going up, and way beyond what I wanted to spend.

When I look at the chart above, the amount I was spending per inquiry and per booking, I'm floored. By 2016 and 2017 I was spending over $350 per booking on online advertising! And while I didn't run these numbers until doing this project, I knew that I was spending a lot of money in 2016 and 2017 and not getting the kind of return I was looking for. By 2017, having spent over $5,500 in online advertising that year, and only getting 61 total wedding inquiries, I knew I wanted to go in a different direction. Weddings definitely bring in a lot of my revenue and is a high ticket item, but to spend over $350 per booking in 2016 and 2017 was simply too much.

Google Ads - Source Data - Weddings

First of all, I am aware that there is a lot, I mean, A LOT of information you can test and track with your online advertising campaigns, things like A/B testing, traffic flow, landing page bounce rate, to know more about your advertising campaigns and to choose which one is working best for you. It's getting more complicated every year.

All I know is that when I began online advertising, the basic online advertising I was doing back in 2011 and 2012 worked pretty well and I wasn't spending a lot of money. Something happened, I'm still not entirely sure what, and then it wasn't working for me anymore. Maybe I clicked a button and changed my settings, maybe Google changed its rates or algorithm, all I know is that from what I was experiencing, the advertising dollars I was spending weren't working the same way they did before.

Here is a look at my google advertising clicks, impressions, CPC (cost per click), and cost (spending) per quarter from Q1 2011 to the end of Q4 2019, the same period we have been looking at above. The chart below does include some ads for corporate events and portraits, but the vast majority of the advertising I did was for weddings. Vast majority.

This chart is taken from my Google Ads account, something I haven't opened for the two year of COVID-19, and unlike Google Analytics, actually gives you the data from your account for over 26 months.

A couple of things to notice. The biggest change I see is kind of in the middle, when the red line (impressions) and the yellow line (cost per click) make some big moves. That change in both those lines takes place between the end of Q4 2014 and Q1 2015. In that one quarter, my cost per click jumps from $1.00 to $2.48 in one quarter, and never really goes back down. As a result, you can see my cost (the green line) starts to jump up as I need to pump more money into the systems due to the fact I'm depleting my advertising dollars quicker and quicker. Why the impressions dropped so much, I don't really know.

I wanted clicks to my website, that was key in my mind. People would could to the website and see my beautiful photos and ask me about my photography (get an inquiry) and I would be happy.

The main thing that seems to be really missing in this graph is that even though I pump more money into the advertising (the green line), I don't see the number of clicks (the blue line) going up. At times, from the middle to the middle right, the blue and green line seem oddly mirrored, which would be something I wouldn't be opposed to, but I know that at that point in my campaign, I was receiving fewer and fewer inquiries per year.

Let's take a closer look at the three years before and the three years after that big jump in the average cost per click.

Over the two three year samples, I'm spending twice as much money and getting significantly fewer clicks. My cost per click has way more than doubled. Something happened that was not in my interest as a business whatsoever. What I know is I wanted inquiries. I wasn't getting as many as I wanted, and kept pumping money into a new system, a new system that was so much more expensive.

And this gets me to my next point about online advertising, just because I got a click, who was actually clicking on my advertising anyway. By 2016, I was really beginning to question if these ads were delivering the right people and being valuable clicks, clicks that would become wedding inquiries.

First of all, let's be honest, I don't click on Google ads. I know where they are on a webpage or a search engine and I avoid them. I know that's true, but the ads seemed to be working before, and I was trusting that they were continuing to work.

But then at some point, around 2015, I started to get a lot of solicitation. A lot. You know what these companies were soliciting me for? Improving my SEO, improving my google ad campaigns, telling me that they could guarantee getting me on the first page of google. Eventually, it felt like the only people who are actually clicking my ads were companies who were looking to solicit my business to improve my google ranking or google advertising campaigns. How do I know this? Well, you've seen the number of actual number of inquiries I was getting for weddings, and it wasn't anything like the actual amount of money I was spending.

Also, I noticed the inquiries I was getting were in greater proportion referrals from previous wedding clients, something I didn't get a lot of when I was beginning my career because I didn't have a large enough historical client base I was working from. Referrals wouldn't be clicking on adds, so all the more reason for me to doubt the inquiries I'm even getting are from the ads themselves.

And so that was why I dramatically dropped my online advertising and eventually left it altogether, lack of evidence it was even working for me. I did need to improve me SEO, I wanted to be higher in google rankings, and maybe if I built a new website around stronger SEO methods, with a professional developer, I wouldn't need to put money into any ads at all. That was the thinking at least.

Websites and SEO

The truth is, and we all know it, the best way to get inquiries is to be ranked high on google. SEO (search engine optimization) is king and you want to be on the first page of google. Anywhere else, you might as well be invisible. It's true. And at one point in my business, I was there! I've been there before, you can probably guess when, back in the earlier days of the business. I'd have a look for you, but finding that kind of information is much harder than it used to be, and getting historical ranking information like that costs quite a bit of money from sites offering the service.

I can't even look at the organic traffic to my site over the years anymore! I remember when I could dive into the deep history of my Google Analytics (website traffic history data), but unfortunately, that information is only available from the last 26 months now. I think that is absurd!

Over the course of my business, I've have had four different versions of my website. The first one was likely my favourite, but it wasn't mobile friendly, so in 2011 I subscribed to the service called 4ormat and used them through 2014. Then I created a wordpress based website using a theme I subscribed to annually created by Minimal. Finally, in 2018, when I was done with online advertising, disappointed in my inquiries going down the tubes so considerably, I decided that I needed to make a professional upgrade to my website. I invested thousands of dollars into my new website, being told that I wouldn't need to put money into advertising anymore as the SEO would go through the roof, and... it didn't do much. I'm pretty much ranked lower than I have ever been. Sigh.

Each of these websites would have had its very own SEO, and what's even more complicated, Google is constantly changing its algorithm as to what is weighted more favourably for a good SEO score. It would be especially helpful if I could go back and see the organic search traffic of my site since its inception, but unfortunately I can't.

I can see websites offering all this great information on historical website rankings, SEO and keyword searches going back in time, and this could be really valuable for me. Most of them have a pretty significant paywall, or want to log in using my google account (I don't know the reputation of a lot of these companies) and so I'm kind of left out wondering. But honestly, even if I get in there to have a look, it's not really my domain, and I can't guarantee I'd even know what the heck I was learning from the information I'd receive.

My business is photography. That's what I do. I'm good at taking pictures, editing them, and I make my clients happy. I am not a SEO expert, google ad guru or website developer. The challenge of being a small business in the digital age is that there are a lot of hats one has to wear anymore. You can't just excel at your particular trade or craft, the world kind of demands that you also be an expert in social media, SEO, online advertising.

And that's part of the challenge too, small business can either spend a whole lot of extra energy and time learning how to do a particular task outside of their usual business activities all themselves, or they can try to turn to an expert in the field and spend a lot of money for someone else's services. Either one can be successful, and either one could fail.

Early in my careers, I attempted to learn what I could and do a lot of the website and advertising on my own, and I was sometimes successful. It was a big investment, investment in time and learning and research. I could have easily not been successful, but I got lucky. But things changed quickly. The Google Analytics I was familiar with earlier in my career looks nothing like the analytics five years later, and I was getting busier as a business, I didn't have all the time I had before to invest into learning all the new features. Now, there are so many new options in Google Analytics and Google Ads, complicated features, and I don't really have the time and energy to sit down and catch up.

Alternatively, you can make a capital investment, hire someone to do the work for you. I did this later in my career, after having the kid. I was busier, a little older, had a little more money and a son I liked spending time with, so I decided I wasn't an expert in SEO anymore, and decided to hire a professional to help me create the best website. Turns out, I didn't make the right decision. And that stings a bit.

My SEO needs a major upgrade, a new website, and maybe I go back to making one on my own. I could save some money doing it, but man, have things changed over the years, I wouldn't even know where to begin. I could return to online advertising, but my business is so down due to the pandemic, I don't have the business capital saved up to throw thousands of dollars into advertising like I used to. I could actually put effort into social media, that would be something, but man, it's not like that isn't a soul stealing endeavour I wouldn't wish on anyone.

Conclusions

Although business had slowed somewhat in the years leading up to the pandemic, I was still busy, had good clients that were calling me back. I was even picking up new clients and had two big gigs booked with new businesses in the spring of 2020. And then everything stopped. Over the pandemic, with a lot of downtime, a lot of daddy time, I decided to go back to school to keep my brain sharp, stimulate the old noggin and even aid my mental health. That's when I started learning about big data.

I had always been interested in data and I'm proud that I have kept the kind of records I had over the years. What was one small bonus of the extended Ontario pandemic lockdown, was the ability, though this post, to actually go through the data, convert it to a useable format, and then finally visualize it, run it through some gymnastics, and learn a lot through the process. To finally dive into the data and learn more about where I began, my peaks and valleys, the ebbs and flows of Joseph Michael Photography, was a real pleasure. Sure, I wish I had kept more data, and would love to have access to more from Google Analytics (limiting that data to the past 26 months is ridiculous!), but from what data I kept, I feel much more knowledgeable as a business owner and feel well equipped for future probing as a newly minted data guy.

Of course, it has left me with more questions than answers, but that's likely going to happen no matter how much data you collect. It also reminded of all the great clients types and gig types I've been fortunate to shoot all these years. It is a nice trip down memory lane. My business has been successful, and I'm proud of it.

If I were to make recommendations to my own business, I would begin and end with getting inquiries up, and I would do that with improved SEO, do more on social media (I loath social media), and leverage old clients and referrals as much as possible. This is generally good practice with any business, I know this, but I'm starting to see how important this is for my own business.

I am also reminded how much I do love data, how passionate I am about digging into the data and finding questions or clues or hints as to what is actually going on. I know there are a lot of business out there, that have maybe be struggling like I have through the pandemic, and would love a little help looking through their data with a little more clarity to see where they have been and where they are going. I know I would love to help.

Notes

• All the graphs and charts on this post were made in Tableau Desktop 2021.4. I was able to use Tableau a lot in my data course at York University, and as I was a student had access to a free year-long trial. Unfortunately, that has been used up, so I used my business email for a 14-day trial for this post.