Case Study - Client Analytics
Alterations by Josetta, 2018 to 2025. Duplantier Advertising.
Alterations by Josetta is a single location alterations and custom tailoring business. I have handled the analytics side of the business since 2018. I built and maintain the website, set up the tracking, and create the reports the owner uses to decide when and where to spend money on advertising.
This was not a practice dataset or a made up business problem. Every number in this case study came from a real business that needed to keep customers coming through the door. The decisions made from this data involved real money.
An alterations business is seasonal. The owner could feel those seasonal changes, but she could not put numbers behind them.
Prom, weddings, graduations, school events, and the fall return to work all increase demand. Between those events, there are months when business slows down every year.
The real question was, "When should I spend money, and what should I spend it on?"
That is a timing question before it becomes a media buying question. The traffic history gave us enough information to start answering it.
| Year | Visits | Change | Unique visitors | Bounce rate |
|---|---|---|---|---|
| 2018 | 927 | Baseline | 777 | 48.0 percent |
| 2019 | 1,284 | 38.5 percent increase | 1,100 | 35.4 percent |
| 2020 | 937 | 27.0 percent decrease | 786 | 41.0 percent |
| 2021 | 1,828 | 95.1 percent increase | 1,500 | 38.3 percent |
| 2022 | 2,587 | 41.5 percent increase | 2,200 | 38.1 percent |
| 2023 | 2,575 | 0.5 percent decrease | 2,100 | 41.1 percent |
| 2024 | 2,367 | 8.1 percent decrease | 2,000 | 44.8 percent |
| 2025 | 3,020 | 27.6 percent increase | 2,600 | 51.4 percent |
The traffic history falls into three main phases.
The first was steady growth through 2019.
Then 2020 happened. The annual total does not fully show how hard the business was hit. February had close to 180 visits. March dropped to around 12. April and May had zero visits.
Not low traffic. Zero.
For two months, an in person alterations business basically disappeared online because the business itself could not operate normally. The year ended 27 percent below 2019, and traffic nearly doubled in 2021 as the business recovered.
After that came a plateau. From 2022 through 2024, the website averaged about 2,510 visits per year and showed almost no overall growth.
Then 2025 broke the pattern. Traffic in 2025 was about 20 percent higher than the average of the previous three years. That made 2025 the year I wanted to understand.
Over the years, the website's contact form had quietly built a list of customers and potential customers. That list now contains 399 contacts. Every person on it had already found the business and taken some kind of action.
The monthly traffic data also showed a repeatable pattern. Traffic usually reached its lowest point around February, climbed to a peak in April that was roughly two and a half times higher, and then increased again around late summer.
Those increases were not random. They lined up with prom, wedding season, graduations, sports, school events, and people returning to school and work.
The plan was simple. Use the contact list based on the business calendar instead of only thinking about how much money was available to spend.
Emails were sent before prom, wedding season, sports, and school events. Campaigns were also sent during slower months, when traffic was lowest and the business had more open capacity.
Instead of waiting for customers to remember the shop, the campaigns gave them a reason to come back at the right time.
Email traffic can be difficult to identify in website analytics, especially when campaign links are not properly tagged.
Someone who already knows the business may click an untagged link, type the business name into the browser, or visit the website directly. In many analytics platforms, those visits are grouped under Direct traffic.
In 2024, Google was the largest source with 1,141 visits. Direct traffic was just behind it with 1,135. In 2025, that order reversed.
Direct traffic increased 22 percent to 1,388 visits and became the largest traffic source. Google increased by only 5 percent.
Most of the growth did not come from search. It came from people who already knew the business or reached the site without a recorded referral source.
That is the type of pattern an email campaign can leave behind when the campaign itself is not being tracked separately. It is consistent with the work that was done, but it does not prove that the emails caused the increase.
One year of Direct traffic passing Google could have been noise.
Through the first seven months of 2026, the pattern had not reversed. Direct traffic still led with 872 visits compared with Google's 814. That meant people who already knew the business remained its largest source of website traffic for a second year.
Two years still do not prove causation, but the pattern becomes harder to dismiss when it continues.
There are also two details in the partial 2026 data that need to be presented honestly.
Total visits were about 5 percent lower than during the same period in 2025. That is a real decline, but it is only a partial year comparison, and a large part of the growth in 2025 happened later in the year.
Bounce rate also fell to 48 percent, about 12 percent lower than the previous year. This was the first decline after six years of mostly increasing bounce rates. That could be the beginning of an improvement, but one partial year is not enough to call it a turnaround.
Most conversions happened over the phone or in person. The owner books alterations through calls and walk ins. There is no online checkout, no reliable connection between a website form and a completed sale, and no revenue recorded inside the analytics platform. The platform reports zero dollars in revenue and zero completed checkouts for every year in this report. That is technically correct, because none of the money passed through the website.
The emails were also created and sent through a separate platform. Those historical campaign records are no longer available, so I do not have the original send dates, open rates, or click rates.
Because of those limitations, the increase in Direct traffic is an association. It appeared during the period when the email campaigns were used most heavily, it matches what I would expect to see from those campaigns, and I cannot see another clear explanation in the available data. It is not a causal finding, and I would not present it as one.
A tracked phone number on the website, with different numbers assigned to different campaigns, would have connected marketing activity to actual customer calls. That one change could have turned the main result of this analysis from an inference into something directly measurable. Call tracking is widely available and relatively inexpensive. Not having it is the biggest measurement gap in the eight years of data.
Every email link should have included UTM parameters. That would have separated campaign visits from Direct traffic and allowed us to measure exactly how many website visits came from each email. Instead of making an argument based on a pattern, I could have reported a clear campaign number.
These are things I understand now that I did not fully account for when I first built the tracking in 2018. The honest version of this case study is that the strategy appears to have worked, but the measurement did not keep up with it.
Bing increased from 26 visits to 309 in one year, almost a twelvefold increase, while Google remained relatively flat. The change is too large and too isolated to assume it came from normal growth in Bing usage. It needs to be investigated before anyone uses it to make a marketing decision.
Bounce rate increased almost every year from 2019 through 2025, rising from 35.4 percent to 51.4 percent while total traffic increased. The website was bringing in more people, but those visitors were becoming less engaged. Traffic growth was hiding a possible content or traffic quality problem. The partial 2026 results are the first sign of that trend reversing.
The Wardrobe Seamstress page has a 91.9 percent bounce rate across 563 views. It is the worst performing page on the website by a large margin and should be one of the first pages reviewed or rebuilt.
In 2022, the main summary reported 2,587 visits while the device report showed 2,597 for the same period. In the partial 2026 data, one report showed 1,803 visits and another showed 1,804. The differences are small enough that they do not change the conclusions, but they still matter. When two reports from the same platform do not match, the data cannot be treated as perfectly reliable.
The most useful part of this analysis was not a complicated model.
It was recognizing that the business already owned an asset it was not fully using: a list of nearly 400 people who had already shown interest.
It was also recognizing that the business calendar was more predictable than the advertising budget. Prom, weddings, graduations, and the return to school and work happened at roughly the same time every year.
The slower months were not necessarily the months when the business should stop marketing. They were the months when there was open capacity and a greater need to give previous customers a reason to return. The strategy was based on using the right message at the right time.
The part I would change is the tracking. I helped create a strategy that appears to have worked, but I did not build the measurement system needed to prove exactly how well it worked.