365 days of SEO, written down as it happens

This is a year-long log of the search work on one small app’s website. One entry per working day, published on the day it happened, starting 30/08/26 and running for 365 days. It is not a case study. A case study is written at the end, by the winner. This is a notebook we leave open so we can’t rewrite the middle later.

DayBloq’s own site is the first patient. If we can’t get a stranger to understand Year in Pixels or a digital journal from a search box, we have no business telling anyone else how this works.

What to expect

• The actual change, with the actual file. Not “improved our schema” — which node, what was wrong with it, what it says now.
• Numbers from Search Console or no numbers at all. No traffic screenshots with the axis cropped off. No “up 300%” from a base of two.
• The things that did nothing. Most days in SEO are like this and every public write-up quietly leaves them out, which is why the public write-ups are useless for planning your own week.
• Decisions we later reversed. With the reasoning that was wrong the first time, left where you can read it.
• Short entries. Some days are one deleted meta tag. That day gets a short page, not a padded one.

What you will not find here: a template you can buy, a tool we are secretly selling, a growth number without a source, or a claim about how Google’s ranking systems work that we cannot point at documentation for. When we are guessing, the entry will say we are guessing.

What we are trying to learn

The point of doing this in public for a year is that we genuinely do not know the answers to these, and a year is long enough to find out.

• How far can a thin domain get on one tight topic? We have one narrow cluster and no backlinks worth the name. Roughly 95% of pages on the web have zero backlinks. Most advice is written for sites that already escaped that. What actually happens if you don’t?
• Does app store ranking help web ranking, or are they separate races? The Play Store listing already does well on its main search phrase. The website does not. Same phrase, same product, two very different scoreboards.
• What does it take to get quoted by an AI answer rather than ranked? Being cited in an AI Overview correlates hard with already being in the top three organic results — so is there anything you can do that is specifically about the citation, or is it all downstream of ordinary ranking?
• How much does being mentioned somewhere else matter? The research says off-site brand mentions beat backlinks for AI visibility by roughly three to one. That is a very different job from link building, and we have never done it.
• Where is the honest ceiling? Not every page can rank and not every keyword is worth a page. We would like to know, with evidence, which of our clusters were never winnable — and how early that was knowable.

Drafts can come from an agent. Nothing goes live until a person reads it and is willing to put their name next to the URL. That loop is the only part of “agentic” that isn’t a costume.

The log

5 of 365 days. Newest first.

1. Day 5 — 04/09/26

   We built the thing a page said we didn’t have

   Day five of 365. A free printable maker landed on a site that tells people in six places there is no web version. What had to change, and what deliberately did not.

2. Day 4 — 02/09/26

   The page that says no

   Day four of 365. An audit flagged four high-severity problems: an empty entity graph, an icon posing as a screenshot, thin pages, no honest iOS answer.

3. Day 3 — 01/09/26

   We changed whose voice we were borrowing

   Day three of 365. The essays sounded like a different internet — punchy, named inner animals, openings that torched advice. That is not who reads a journal.

4. Day 2 — 31/08/26

   We threw away half the keywords

   Day two of 365. Three Keyword Planner exports, a filter that deleted 1,424 rows, and a publishing cadence we nearly wrote into the plan as if it were physics.

5. Day 1 — 30/08/26

   The site had two addresses

   Day one of 365. Google kept two houses for one site, the meta tag listed 419 keywords, and the product was all hallway. Plumbing first.

The days that haven’t happened yet do not have placeholder pages waiting for them. Each one appears when the work does.