Dalinar is a Stormlight Archive reading-guide and quiz site, and it is ours: Diego owns it and runs it. That matters for how to read this page. It is not a client result, it is an experiment on a site where we can publish every number, including the ones that went the wrong way. If you want to know whether daily content plus interactive quizzes moves anything measurable, this is the cleanest evidence we have.
What we changed
Two things, both on a fixed rhythm.
- A new article almost every day. Daily publishing started on September 10, 2026. Each piece answers one specific reader question first, then goes deeper, with structured data that matches the visible text. Over the second window that meant 22 new posts on a site that had published none in the first.
- A new game or quiz roughly every week. The site now has 13 interactive pages, from character quizzes to a trivia gauntlet and a character bracket. Each one ends with an email gate at the moment the reader most wants their result.
The idea behind pairing them is simple. The articles are how people arrive, and the interactive pages are what turns a visit into a contact. If you want to see what the second half looks like in practice, the flagship is the Which Stormlight Archive Character Are You? quiz.
The numbers, four weeks against four weeks
Window A is August 10 to September 6, 2026 (ISO weeks 33 to 36), before the daily cadence. Window B is September 7 to October 4 (weeks 37 to 40), the latest four complete weeks. Window B includes three days before the first daily post, so the comparison is slightly conservative. Sources: Google Search Console and GA4 for dalinar.com, pulled on October 7, 2026, and the site's own lead table.
| Metric | Before | After | Change |
|---|---|---|---|
| Search impressions | 18,728 | 36,914 | +97% |
| Search clicks | 98 | 158 | +61% |
| Average position | 10.1 | 8.7 | 1.4 places higher |
| Click-through rate | 0.52% | 0.43% | Down |
| Pages earning impressions | 38 | 73 | +35 pages |
| Active users (GA4) | 630 | 752 | +19% |
| Sessions | 676 | 801 | +18% |
| Engaged sessions | 202 | 192 | -5% |
| Email leads, all sources | 26 | 39 | +50% |
These are four-week totals. GA4 weekly active users in window B ran 119, 188, 198 and 251, against 132, 144, 217 and 142 in window A.
Where the engagement actually went
The one line in that table that looks like a failure, engaged sessions down from 202 to 192, is worth taking apart, because the split tells you something.
| Channel | Sessions before | Sessions after | Engaged before / after |
|---|---|---|---|
| Organic Search | 192 | 226 | 102 / 139 |
| Direct | 468 | 559 | 97 / 47 |
Organic Search is the channel the content is built for. Its sessions rose 18% and its engaged sessions rose 36%, which puts organic engagement at about 61% of sessions, up from 53%. Direct traffic grew too, but now makes up roughly 70% of all sessions while engaging about 8% of them. That pattern looks more like automated or untraceable traffic than readers, though that is our inference and not something we verified. The practical consequence is that the headline engagement figure understates what the content did, and we would rather show you the raw number and the split than quote only the flattering channel.
The quiz as the engine
One page does a disproportionate share of the work. The character quiz earned 38 of the site's 158 search clicks in window B, about 24%, and it sits at an average position of 5.4. Search clicks to it were flat against window A (38 and 38), but its click-through rate rose from 14.9% to 18.1% and its position improved from 6.0 to 5.4. In GA4 its users went from 43 to 50 and its engaged sessions from 22 to 40, an engagement rate that rose from 41% to 69%.
Then the part that matters commercially. Since June 4 the site has captured 88 email leads across six sources, and 64 of them, 73%, came from that one quiz. In window B the quiz source produced 30 leads against 21 in window A, and a second quiz, the Radiant Order one, went from 1 lead to 6. A quiz gets this done because it asks a question people actually type, answers it in under a minute, and puts the email request after the reader is invested in the result rather than before.
A quiz is a landing page that people choose to complete.
What did not work
Four honest notes, because the percentages above are the easy half of the story.
- Click-through rate went down. Impressions grew faster than clicks, 97% against 61%. New pages enter at positions around 9 to 10, where visibility is cheap and clicks are rare.
- The biggest pages leak the most. In window B the guide to Kaladin earned 7,367 impressions and 8 clicks, the guide to the ten Radiant orders earned 6,910 impressions and 13 clicks, and the Dalinar character guide earned 5,896 impressions and 15 clicks. That is a click-through rate of 0.1% to 0.25% on pages whose average position is between 7.4 and 9.2. Fixing titles and snippets on those three is the next job, and it is a bigger lever than writing more articles.
- Outbound intent is barely measured. The site monetizes through book links, and GA4 recorded only three click events in a 28-day check at the end of September. Either almost nobody clicks them or the clicks are not being tracked well, and we have not yet determined which.
- AI assistant traffic is negligible. GA4 attributed 3 sessions to the AI Assistant channel in window A and 1 in window B. We are not claiming assistant citations from this work.
What this says about GEO and content
For a business, what carries over is the structure, and the volume matters less. Search visibility came from a steady supply of pages that each answer one real question, and capture came from interactive pages that pair a real question with a result. Both are the same discipline we apply to GEO: pages built to be extracted and cited, with the structured data and the visible text saying the same thing. What we have shown here is the search and lead half. Whether assistants quote these pages is a separate measurement, and for this site we have not run it.
Dalinar is also small, with a few hundred users a month, in a niche with very clear intent. Treat the percentages as direction. Your category will have different volumes, and four weeks is a short window.
Want the same system pointed at your site?
The AI Visibility Snapshot is a free first-party baseline: where your brand stands today across ChatGPT, Perplexity, Claude, Gemini and Google AI, with the findings written up whether or not you hire us.
To see the result for yourself, visit Dalinar and try the Which Stormlight Archive Character Are You? quiz.
Common questions
Is Dalinar a client of Cipion?
No. Dalinar is a Stormlight Archive reading-guide and quiz site that Diego Cipion owns and runs. We use it as a live lab, which means the numbers are ours to publish and we can show the bad ones as well as the good ones.
What did the daily content and quizzes change in four weeks?
Comparing August 10 to September 6 against September 7 to October 4, 2026, Search Console impressions went from 18,728 to 36,914, clicks from 98 to 158 and average position from 10.1 to 8.7. Email leads went from 26 to 39. Click-through rate fell from 0.52% to 0.43%, and engaged sessions overall slipped from 202 to 192.
Why use quizzes and games instead of only articles?
A quiz answers a question people literally type, such as which character they are, and it asks for an email at the moment of highest interest. On Dalinar one quiz produced 64 of the 88 leads captured since June, from a page that earns about a quarter of the site's search clicks.
Does this prove the quizzes get cited by ChatGPT or Perplexity?
No. AI assistant referrals to Dalinar remain too small to call a trend: 3 sessions in the first window and 1 in the second. What we can show is search visibility and lead capture. Citation share inside assistants is a separate measurement we have not run on this site.
Can the same approach work for a business that is not a fan site?
The mechanics transfer: a steady publishing cadence tied to real questions, plus an interactive page that qualifies and captures the visitor. The volumes will not. Dalinar is a small site in a niche with clear intent, so treat the percentages as direction and not as a forecast for your category.