Four months into running findindiegame.com, I have 90 published YouTube Shorts and essentially no meaningful organic search traffic. That's partly by design, partly a reflection of how new domains work. But watching the analytics closely has led me to make a bet I want to write down publicly before month 6 arrives.
The bet: by month 6 post-launch, YouTube Shorts will drive more monthly visitors to findindiegame.com than organic search.
This is falsifiable. I have channel analytics, Google Search Console, and a per-video daily snapshot in data/yt-analytics-history.jsonl. I'm wrong about things regularly. The goal here is to record what I think so I can tell the difference between being right and getting lucky.
The bet in one sentence
YouTube Shorts game matchup content ("games like Stardew Valley") targets the same queries that game comparison pages target on Google — but on a platform with algorithmic distribution measured in hours, not a domain authority curve measured in years.
A brand-new domain competes against years-old game wikis, review sites, and Steam-adjacent content on Google. On YouTube, a fresh Shorts channel gets served to audiences of the opposing game's creators, to people who searched that title, to the algorithm's candidate pool — within hours of publishing. The discovery asymmetry is large. That's the core of the bet.
Why I think Shorts win on timeline
The primary reason is the lag difference. On Google, a new page on a new domain goes through: crawling (weeks), indexing, domain trust accumulation, and then actually competing — often months before meaningful impressions appear in Search Console. On YouTube Shorts, the distribution starts on day one if the hook works.
There's also a demand-side case. Game comparison queries are natural YouTube searches. "Is Hollow Knight hard?" "Games similar to Stardew Valley?" These aren't desktop-research queries. Players ask them between sessions, on their phones, usually with the original game in mind. That's a YouTube moment. Game discovery runs on video in a way that developer-tool discovery does not.
I built the two-host video pipeline specifically to produce these Shorts at scale. The YouTube slide renderer generates thumbnails programmatically. The pipeline produces a new video spec from game data each day, encodes it, and queues it for upload — without manual editing per video. That production leverage matters: I can publish more than one short per day if needed, and the marginal cost per Short is close to zero — the pipeline runs on built-in templates and my existing subscription, with no per-call API spend.
What 90 videos of data shows
Fleet velocity is currently around 60 views/day across 90 videos. That's ~0.67 views/video/day on average, which sounds low but games content accumulates over months rather than days. The best performers are still growing.
The A/B test I've been running — hook style: number-first versus narrative — was suspended after reaching a 1.82× variant lead. The suspension was triggered by 26% attrition in the last 30 days of uploads, which is a measurement-pipeline problem rather than a content problem. But I have to be honest about what that ratio is worth: it comes from 3 surviving variant videos versus 2 control videos, with deletions censoring the sample. My own strategy doc says "do not act on ratio — continue recording hook_arm," and that's the rule I'm following. The 1.82× is a number to keep watching, not a signal I'm acting on.
From running the first 48 Shorts: product framing beats build-in-public framing 14×. "Stardew Valley players also love Hollow Knight" significantly outperforms "I built a game comparison site." That matters for the bet: the content that performs well on YouTube is exactly the content that serves a player's comparison intent — the same intent the directory pages serve.
The archetype selection system routes new specs to the game pair formats that have shown the best day-1 velocity for that anchor game. I run view-count snapshots daily and use those to tune anchor selection. Right now, Stardew Valley paired with high-review-count AAA games (War Thunder, Dead by Daylight) is consistently outperforming other pairs.
The thumbnail brightness floor analysis identified dark game art as a click-through suppressor and fixing it improved early velocity. The four thumbnail rules I hardcoded after 19 videos of analytics now run as linting checks in the pipeline before upload.
All of this points to a functional, improving pipeline. 90 videos is enough to see real patterns in the data. Month 6 is where I expect those patterns to have compounded into something measurable against search.
The counterargument: search compounds, Shorts don't
Here's the strongest case against my bet, stated fairly.
Search traffic compounds. A page that ranks for "games like Stardew Valley" will keep serving that traffic for years without additional work. A YouTube Short gets a burst of algorithmic distribution and then fades — the long tail of Shorts performance belongs to videos that caught a viral moment, not to structural search demand.
The directory pages also target this query in their HTML and structured data. Programmatic freshness helps with the indexing signal, but domain authority takes time. At month 6, the domain is still less than 6 months old. The comparison might just be measuring which channel has a shorter lag to distribution, not which channel produces more durable value.
A credible counter-bet: by month 12, organic search overtakes YouTube referrals, because three years of crawl history on a competitor's domain outweighs one year of Shorts accumulation. I think that's probably true at month 12. My bet is specifically about month 6 on a new domain.
There's also YouTube-side risk. The platform changes its Shorts algorithm frequently. I've already seen two shifts in four months — once the "seed views" window seemed to expand, once the mechanic for surfacing game-specific content changed. If the distribution model degrades, the referral advantage disappears regardless of content quality.
The measurement approach
I'm tracking this through three lenses.
Direct referral traffic: this is the lens I want but don't cleanly have yet. Right now most video descriptions carry only the bare domain; a handful of earlier uploads do deep-link to the specific game page, but none of them — bare or deep — carry UTM parameters, so Shorts taps can't be separated from other traffic. Adding UTM tagging to the description template has been on my TODO list since May and hasn't shipped. Until it does, this signal is blurry, not clean.
Brand search uplift: when someone watches a matchup Short and searches the site name directly, that shows in direct or brand-organic traffic. Harder to attribute precisely, but YouTube drives brand recall even without a click.
GSC impressions for "games like X" cluster: I check weekly whether any comparison pages are appearing in the Search Console impressions report for the target query cluster. Right now: essentially nothing. If that changes sharply before month 4, the search trajectory changes and I'll update the bet.
I'll run a PDCA check-in at month 4 — not month 6 — because I want a corrective window. If search is already clearly winning at month 4, I'll redirect effort toward link-building rather than expanding Shorts volume. The 92-day detection lag that PDCA closed was caused by waiting too long to look. I won't repeat that here.
What would change my mind
Google indexes faster than expected. If findindiegame.com starts ranking in positions 10–30 for mid-tail "games like X" queries before month 4, the search ramp is faster than I've modeled and I'll revise the bet conditions.
YouTube distribution degrades for this content category. If per-video day-7 velocity falls below 8 views across five consecutive new uploads, algorithmic distribution has degraded to the point where the bet premise breaks.
Attrition is real, not a measurement artifact. The 26% attrition rate I mentioned is a pipeline tracking issue — but if it reflects actual video removals (copyright claims, audio issues), the fleet velocity is worse than the analytics show. I'd re-examine the content production constraints before month 4.
Month 6 organic search referrals exceed YouTube referrals by more than 20%. If that happens, I'll publish a follow-up calling the bet wrong and documenting what I missed. A 20% margin is enough to be clear; within that margin, the result is ambiguous and I'll extend to month 9.
Why write this down
I have no credible external reference point for what traffic channels work best on a new game comparison directory. Nobody is publishing real channel breakdowns for this type of site. The only data I have is my own.
Writing explicit, falsifiable bets is how I avoid confusing motion with progress. Ninety videos feels like a lot of effort. But effort doesn't imply the channel works. Month 6 will tell me whether the effort pointed at the right thing. The follow-up post will be more useful than this one, regardless of which way it goes.
FAQ
What's the current referral split between YouTube and organic search?
Currently it's not a meaningful comparison — organic search is near zero for the comparison page cluster, and I don't yet have clean attribution for YouTube Shorts referrals because most descriptions carry a bare domain link and the few that deep-link to a game page have no UTM tagging either. Site-side traffic overall is near zero too. The bet is about where month 6 lands, not where month 4 sits today.
Does this bet conflict with the article cross-publishing strategy?
No. Articles on Dev.to and Hashnode target developer and indie-maker audiences. Shorts target gamers on YouTube. The audiences barely overlap, and the same three directory sites benefit from both channels. I covered the cross-channel approach in an earlier post.
What's the marginal cost of one more Short?
Essentially zero in API terms — the script comes from built-in fallback templates and a scheduled Claude Code routine covered by my existing subscription, with no per-call API spend — plus around 3 minutes of compute for encoding. At 90 videos and $25/month total infrastructure, the marginal cost per Short is negligible. The constraint is not cost — it's upload cadence and A/B testing reliability.
How do you pick which game pairs to use?
Steam review count as the anchor game size signal (higher review count = larger existing audience). The AAA or mega-indie game on one side of the matchup determines whether the Short gets served to a large candidate pool. I wrote about this in the thumbnail rules post — anchor selection and thumbnail brightness are the two biggest levers on early velocity.
What if neither channel dominates by month 6?
If both are within 20% of each other, I'll extend the measurement window to month 9. Ambiguous data at month 6 means the channels are more symmetric than I expect, which is itself useful information. It suggests running them in parallel rather than picking one to scale.
Part of an ongoing 6-month experiment running three AI-curated directory sites. The technical claims here are real; this article was AI-assisted.
Top comments (0)