On 12 November 2026, we examined Serse’s own search performance and public website. This is the starting record for an ongoing improvement study. It is not a customer testimonial, a completed ranking recovery, or a claim that a particular edit has already increased traffic.
We are publishing the baseline because an SEO product should make its own improvement process inspectable. The useful question is which pages need work and whether later evidence supports the decisions.
What we observed
Google Search Console’s domain property for serse.ai, Web search, all countries and devices, reported the following for 11 August–10 November 2026:
| Metric | Recorded value |
|---|---|
| Clicks | 7 |
| Impressions | 632 |
| Click-through rate | 1.1% |
| Average position | 40.5 |
These are historical observations from the Search Console interface. They are not forecasts, keyword search volumes, or customer acquisition numbers. The property includes subdomains. Google’s privacy rules mean visible query rows do not necessarily explain every click in the headline total.
Why the site-wide average was insufficient
The homepage had 7 clicks, 208 impressions, and average position 8.6. Other pages had different profiles:
| Page | Clicks | Impressions | Average position |
|---|---|---|---|
| Free scan | 0 | 162 | 63.0 |
| SEO reporting | 0 | 143 | 51.4 |
| Search Console integration | 0 | 58 | 67.5 |
| Agency software | 0 | 20 | 12.4 |
Page impressions can overlap within a search, so these rows should not be summed to reconstruct the property total. The agency page’s 20 impressions are also too few to establish a stable market position.
The visible query “seo site scan” had 133 impressions and average position 67.9. That suggested a concrete investigation: whether our scan page clearly explained the tool’s job and helped someone use its output. It did not prove that a wording change would improve rankings.
What the public-site review found
The blog had no published articles. The homepage described the product through a broad headline, while its performance demonstrations were explicitly illustrative. The main marketing pages were reachable and exposed indexable metadata. The www address failed certificate validation; its DNS pointed to Vercel, but the subdomain was absent from the project’s domain configuration.
The implementation programme addresses the domain configuration, clarifies the homepage and scan page, deepens reporting and integration explanations, and adds practical guides with sources. It preserves the distinction between demonstration data and measured performance. We will record verified delivery separately from subsequent search results.
What we will measure
The campaign begins on 12 November and ends with a review on 12 December 2026. Weekly checkpoints examine crawlability, index coverage, relevant non-brand clicks, priority-page performance, and the quality of available conversion evidence. The same query-and-page group will be retained so a changing keyword mix does not silently redefine success.
We will compare compatible reporting periods, retain country and device context, and label incomplete data. Signup clicks will not be counted as completed registrations. Financial outcomes will require actual billing evidence, with test activity excluded.
What this study cannot establish yet
There is no post-change ranking outcome in this baseline. A 30-day observation is not a controlled experiment: search demand, competitor activity, Google changes, and other marketing work can influence the result. Low traffic also makes percentage changes unstable.
We will add dated observations when they exist, including inconclusive or disappointing results. Until then, the defensible claim is that we identified specific gaps and established a way to check the work and its outcomes.
For the methods behind the programme, read how to interpret a free scan, Search Console integration checks, and the reporting framework. Google’s position definition is the reference for interpreting the ranking metric.
Why we chose those pages first
The most obvious reaction to an average position of 40.5 would be to target the highest-volume keywords we could find. That would skip the more useful evidence already in front of us. Search Console was showing impressions for existing pages with clear jobs: a scan, a reporting workflow and a connection to Search Console. These pages already formed a path from a search question to the product.
Our first priority was therefore to improve the explanation of those jobs. Someone searching for a site scan needs to know what will be inspected, what the result means and what to do afterwards. Someone evaluating reporting software needs to understand which evidence appears in a report and what it cannot establish. These are content requirements that can be checked immediately, even before traffic changes.
A decision table, not a forecast
| Page | Baseline signal | Question behind the work | Evidence that would change the decision |
|---|---|---|---|
| Free scan | 162 impressions, no clicks | Does the page clearly explain scope and the next step? | Queries prove unrelated to the tool, or users cannot complete the scan |
| Reporting | 143 impressions, no clicks | Does the page explain a useful reporting decision? | Relevant visitors need a different workflow than the page describes |
| Search Console integration | 58 impressions, no clicks | Does setup guidance distinguish connection from usable data? | Collection or property selection prevents the promised workflow |
| Agencies | 20 impressions, no clicks | Does the offer answer an agency's practical evaluation questions? | More evidence changes the apparent query intent or commercial fit |
The table does not convert impressions into a revenue forecast. It records why a page is worth investigating and what could make that priority wrong. That second column of reasoning matters: a plan that has no revision condition is difficult to learn from.
Why the agency page is not the obvious winner
Its recorded position of 12.4 looks more promising than the scan page's 63.0. But the agency page had only 20 impressions. A small number of appearances can be driven by a narrow or unrepresentative query set. We need to inspect relevance and gather more observations before presenting it as a dependable near-page-one opportunity.
Conversely, the scan page's larger impression count does not prove its business value. It provides a more substantial starting signal for a specific question. The task is to inspect those searches and improve the page's usefulness, then observe what follows. A ranking number by itself cannot make that decision.
Keep the baseline intact while improving the method
A public study becomes misleading if the author changes the success definition after seeing the results. We will retain the original property-wide observations and distinguish them from later, narrower reporting groups. A refined measurement method can be useful without being represented as something that existed at the start.
The original baseline covers approximately three months. A later 28-day window must not be compared with that total as if both had equal exposure time. At a checkpoint, we need comparable periods with matching filters. The longer baseline remains useful context; it is not a ready-made denominator for every later percentage.
Treat missing observations as unknown
If a query disappears from a visible export, that alone does not prove it received zero impressions. Privacy filtering and reporting limits affect what can be seen. Similarly, if conversion tracking was unavailable during part of a period, the missing observations should not be filled with zero completed signups.
For the same reason, we will show raw counts alongside rates. With a baseline this small, a handful of additional clicks can create a dramatic percentage. The magnitude may be encouraging, but the percentage does not make the evidence more stable. The study should help a reader understand the uncertainty rather than make a small sample look impressive.
Separate publishing progress from search progress
The first programme changes were published on 12 November: clearer marketing-page explanations, practical supporting guides, a corrected www configuration and improvements to the measurement setup. That is a delivery record. It does not establish that Google has processed the new versions or that the changes improved acquisition.
The article programme also runs through a hosted publishing service, so preparing and publishing a scheduled post does not depend on a laptop staying on. That solves an operational problem. It does not solve the editorial problem by itself. A reliable schedule can distribute weak writing just as reliably as useful writing.
On 13 November, we revised the initial articles after finding that their short treatment and cramped presentation left useful reasoning undeveloped. The second editorial pass adds: expanded worked examples, clearer decision methods, better tables and a more readable article layout. We are recording that correction because publication count would otherwise hide a material quality problem.
What counts as a useful article here
Each substantial guide should answer a narrow reader question, develop an original example or diagnostic method, cite primary evidence where it makes factual claims, and explain how the reader can use the result. It should also address a plausible failure case. A checklist that says “check your data” without showing what to compare is not enough.
Word count is an editorial constraint, not a ranking promise. Google's people-first content guidance asks whether content provides original value and satisfies the reader; it does not prescribe a preferred word count. We are expanding these articles because the first versions omitted useful reasoning, not because a longer page automatically deserves a better position.
What the 30-day review will decide
At the review, we will ask three separate questions. Was the planned work delivered and verified? Did relevant search observations change under comparable reporting conditions? Is there enough verified business evidence to connect that movement to useful outcomes? An honest answer can be yes, unclear and no respectively.
| Checkpoint result | Interpretation | Next decision |
|---|---|---|
| Work delivered, little search data | Delivery is established; impact is not | Keep measuring and review whether the selected questions remain useful |
| Relevant cohort improves, business outcome unclear | Search evidence is promising but incomplete | Inspect the landing journey and outcome instrumentation |
| Broad impressions grow, relevant cohort declines | The headline masks a problem | Revisit affected pages and competing explanations |
| Page remains unavailable or excluded | An eligibility problem persists | Resolve and verify the technical issue before judging content impact |
| No meaningful improvement | The original approach may be insufficient | Reassess intent, usefulness, competition and measurement rather than inventing success |
We will also note events outside the intervention: site changes, campaigns, shifts in demand and relevant search-system developments. Google's traffic-drop investigation guide provides a useful starting set of alternative explanations. These observations cannot remove every confounder, but they can prevent an obviously incomplete story.
A 30-day checkpoint is especially limited for a small site. Google's core-update guidance notes that improvements can take days to months to be reflected. That is a reason to interpret timing carefully, not an excuse to leave unhelpful pages untouched indefinitely.
How to read future updates
A future update should state its observation date, report scope, comparison period and the work already live during that period. It should preserve this starting record and explain any change in measurement. If the evidence is disappointing or inconclusive, that belongs in the update just as much as an improvement.
The useful output of this study is not a polished success story written in advance. It is a traceable series of decisions: what we noticed, why we chose a change, whether it reached the website and what later evidence did to our judgement. Readers should be able to disagree with the decisions using the information provided.
Until a dated follow-up contains new Search Console observations, the numbers in this article remain the historical baseline. No post-change ranking gain, traffic increase or revenue effect is being claimed. The next result must come from evidence, not from the fact that the publishing system is now working.