SEO Case Study Traffic Growth: How to Read the Numbers

Headlines like "+340% traffic in three months" read well. The trouble is that most case studies leave out the parts that make the number mean something: what the starting point was, what exactly changed, when it went live, and what else was happening on the site at the same time. Going from 50 to 220 visits a month is +340%. It is also still close to nothing.

This is not a client testimonial. It is a guide to reading an SEO case study so you can tell whether there is anything behind the number — and to running your own once you deploy fixes on your site.

What a case study needs before it can be verified

If even one of these is missing, the result cannot be attributed to any specific piece of work:

1. The absolute baseline. Not percentages. Clicks and impressions for the month before the change.
2. The deployment date. An exact day, not "over the spring".
3. A list of changes. How many titles, how many meta descriptions, how many alt texts, how many new articles. Ideally per URL.
4. The metric and its source. Google Search Console (clicks, impressions, CTR, average position) or analytics. Mixing both into one chart makes no sense — they measure different things.
5. The measurement window. Equal length before and after, and the same kind of days. 28 days beats "a month" because it contains the same number of weekends.
6. What ran in parallel. A new campaign, a redesign, seasonality, a migration, a Google update.

If a study only reports a site-wide total, you know very little. Site-wide numbers rise when a fix did nothing, and fall when a fix did plenty.

How I measure impact

For every fix that gets deployed I keep the date and the URL. Then I compare 28 days before deployment against 28 days after for that specific address, not for the whole site. Four numbers from Search Console: impressions, clicks, CTR, average position.

Splitting them matters, because each one says something different:

• Impressions flat, CTR up — the title or description did the work. The page sits in the same spot, more people click it.
• Impressions up, CTR flat — the page started showing for more queries, or higher. That is a position change, not a wording change.
• Impressions and CTR both down — either seasonality, or the fix hurt. Time to look at the individual queries.
• Nothing moved — the most common outcome for a single fix on a single low-volume page. That is not a measurement error.

Two limitations worth stating up front. First, on pages with a few dozen impressions a month, the gap between 6 and 9 clicks is noise, not a trend. Second, average position in Search Console is averaged across every query, so it can drop even when the page moved up on its main keyword — all it takes is the page starting to appear on new, weaker queries. I go into that in more detail in the piece on measuring SEO impact.

A worked example: what better CTR is actually worth

These are not numbers from any real site. This is arithmetic, so you can see the orders of magnitude involved.

An ecommerce category gets 10,000 search impressions over 28 days at a 1.8% CTR. That is 180 clicks. The title is duplicated across four other categories and the description is missing, so Google writes its own snippet from the page content.

If rewriting the title and description moves CTR to 2.6% at the same impression count, you get 260 clicks. That is 80 extra clicks in 28 days from one URL. At a 1.5% conversion rate, roughly one additional order a month from one category.

It gets interesting at volume. If you have 40 such categories and a similar shift lands on half of them, the model produces around 1,600 clicks a month. That is why for ecommerce it makes sense to handle duplicate title tags in bulk rather than one page a year, and why it is worth spending time on ecommerce meta descriptions.

To be clear: I picked 1.8% and 2.6% for this example. Your numbers will be different and you will only find them in your own Search Console.

Which fixes show up fast and which do not

The speed of impact differs by an order of magnitude between finding types.

Days to two weeks

Titles and meta descriptions. They change CTR, not position. Once the page is indexed and Google picks up the new title, the difference in clicks is visible almost immediately. Note that Google sometimes rewrites the title anyway — that is normal, and a case study should admit it.

Weeks

Structured data. When you fix an error that was blocking rich results, the change appears after the page is processed again. More on that in the article on structured data errors.

Months

New content and internal linking. An article published today has almost no data two weeks later. The first usable numbers usually arrive after 6 to 12 weeks, longer on competitive topics. Anyone promising ranking movement within days is lying.

So a case study that measures the impact of new articles after three weeks measures nothing. And the reverse — a study that measures title changes after six months has mixed in so many other influences that nothing can be attributed.

What skews the numbers

Seasonality is the most common reason a fix "worked". Deploy changes on a Christmas-heavy ecommerce site in November and the 28-day before-and-after comparison will show growth even if you did nothing. The fix: compare the same period year over year as well, or watch a control group of pages you did not touch.

Other frequent distortions:

• A Google update. If one ran inside your measurement window, it shuffled the whole site.
• One viral article. It lifts the site-wide total and has nothing to do with your fixes.
• A paid campaign. Invisible in Search Console, visible in analytics. That is exactly why I measure SEO in Search Console.
• An indexing change. If 200 new pages got indexed at once, impressions rise regardless of how good the fixes were.

Running your own case study in eight weeks

A process you can work through on your own site:

1. Pick 10 to 20 URLs with high impressions and below-average CTR. That is where the headroom is — people see these pages and do not click.
2. Write down the baseline. Last 28 days: impressions, clicks, CTR, average position, per URL.
3. Leave every other page alone. They are your control group.
4. Deploy the changes on one day and note the date. Through a WordPress plugin that takes a few minutes; through a CSV export, somewhat longer.
5. Wait 28 days. Do not touch those pages in the meantime.
6. Compare both the edited URLs and the control group. If both rose by the same amount, something else caused the growth.

Choosing which pages to start with is easier with a straightforward SEO audit — it returns specific findings with URLs, so you are not working from guesses.

Three questions to ask when someone shows you their study

These separate a documented result from a chart:

• What were the absolute numbers before the work started?
• Which exact URLs changed, and on what date?
• What else was happening on the site, and how did you subtract it from the result?

If the answers come back in percentages with no dates, the study proves nothing. Hold me to the same standard: for every fix you deploy, I show the 28-days-before against 28-days-after comparison for that URL — including the cases where nothing changed. Those are part of reality, and without them the average would be a lie.

If you are planning content alongside the fixes, it makes sense to pick topics from queries your site already appears for. I describe how in the article on finding blog topics in Google Search Console.

#seoobsah