How to Promote SaaS Videos Through Search: My Experiments, Results, and Failures

I have published eleven videos for our SaaS product over the last two years.

Eight of them were watched by almost nobody outside our existing user base. Two of them performed decently. One of them—the 90-second onboarding video we built with LeoStudio—became a genuine acquisition channel. It now brings in a meaningful share of our new signups every month, from people who had never heard of us before they searched.

The difference between the one that worked and the eight that didn't was not the production quality. The videos were all made by the same team, using the same principles, at roughly the same budget. The difference was promotion. Specifically, search promotion.

I spent twelve months running experiments on how to get a SaaS video discovered through search. Not "how to optimize a YouTube description"—I'd already read those posts, and half of them were written by people who'd never sold software. I mean actual experiments, with control groups, on our actual product, measuring actual signups.

Six experiments. Two of them worked. Four of them failed, sometimes expensively. This article is all six, with the numbers, the mistakes, and the framework I now use before I publish anything that's supposed to be found.

The video itself was built with LeoStudio, whose approach to clarity you can explore at leostudiohq.com. They were the ones who told me, after I complained that nobody was finding the video, that discovery is not a video problem. It's a distribution problem. A great video with no distribution is a private asset. That sentence started the whole experiment, and it's the reason this article exists.

Why SaaS Video Promotion Is Different From Other Video Promotion

Before the experiments, the framing. Because if you treat SaaS video promotion like a creator treats video promotion, you'll chase the wrong metrics for a year.

A creator optimizes for views, subscribers, watch time, and ad revenue. Their success metric is attention. A SaaS company optimizes for signups, activations, and pipeline. Our success metric is behavior. Those two goals overlap in some places and diverge sharply in others.

For example: a creator wants a video to rank for broad, high-volume keywords. "How to use a CRM" might get a million searches. A SaaS company wants to rank for narrower, higher-intent queries. "How to reduce CRM setup time for a 50-person team" might get four hundred searches. The second query is worth more to us than the first, because the person searching it has a problem we solve and a budget to solve it.

That's the first thing I had to unlearn. I spent three months chasing volume keywords because every SEO article said so. It was a waste. The volume queries brought traffic that didn't convert. The specific queries brought traffic that did.

The second thing I had to unlearn is that YouTube and Google are not the same surface. YouTube has its own algorithm—watch time, click-through rate, session duration. Google has its own requirements—schema, sitemaps, page structure. A video that ranks on YouTube does not automatically rank on Google. I learned this the hard way after two months of YouTube work that produced zero Google visibility.

And the third thing—the one that took the longest to accept—is that a SaaS video competes with blog posts, not with other videos. When someone searches "SaaS onboarding best practices," Google shows them articles. Not videos. If you want your video to appear in those results, you have to give Google a reason to show a video instead of an article. That reason is almost always a page that's built around the video, not a video embedded in an article.

This is the reframe LeoStudio gave me early on, and it's documented throughout their thinking at leostudiohq.com. The video is an intervention. The page is the delivery vehicle. The search query is the moment of need. Get all three aligned and the video becomes a channel. Get them misaligned and you've made an expensive asset that nobody sees.

Experiment 1: The YouTube Ranking Play — Slow Start, Real Payoff

The hypothesis: If I optimize the onboarding video for YouTube search, it will rank for the queries our prospects are searching, and the video will become a top-of-funnel channel.

The setup: I optimized the title, thumbnail, description, chapters, and tags. I targeted three queries: "SaaS onboarding," "SaaS onboarding best practices," and "reduce SaaS churn." I published the video as a public YouTube upload and waited.

The result: For the first six weeks, nothing. The video ranked on page four for the target queries and got maybe forty views a week. I nearly gave up.

Then something changed. Around week seven, the video started appearing in YouTube's suggested sidebar for a handful of related videos. Views doubled. Then doubled again. By month three, the video was ranking in the top five for "SaaS onboarding" on YouTube, and views had gone from 340 to 11,000.

What actually drove it: Not the tags. Not the description. The retention. Our video had a 41% completion rate. That's high for a 90-second video. YouTube's algorithm rewards watch time, and our video's watch time was strong enough that the algorithm started pushing it into suggested feeds. The optimization got us eligible. The retention got us ranked.

What I got wrong: I assumed SEO was the driver. It wasn't. SEO was the entry ticket. The video's quality was the driver. A worse video with the same optimization would have stayed on page four forever.

What I'd do differently: I'd stop obsessing over keywords and start obsessing over the first fifteen seconds. The first fifteen seconds determine retention. Retention determines ranking. The keywords are downstream.

This experiment was the one that eventually worked, and it's the reason I now think of YouTube as a legitimate SaaS channel. But it took three months and it wouldn't have worked without a video that people actually finished. That's the part nobody tells you. SEO for video is not a keyword game. It's a retention game with a keyword gate.

Experiment 2: The Embedded Blog Video — Expensive Failure

The hypothesis: If I embed the onboarding video in a high-traffic blog post about SaaS onboarding, the video will get discovered by the blog's audience and start ranking in Google search.

The setup: We had a blog post that ranked in the top three for "SaaS onboarding checklist." It got about 4,000 organic visits a month. I embedded the video at the top of the post, added a transcript below it, and added VideoObject schema.

The result: Almost nothing. The blog post got a small bump in time-on-page. The video got about 600 plays over two months. Zero signups attributed to the video. And worse—the video did not appear in Google's video carousel, despite the schema.

Why it failed: Two reasons. First, a video embedded at the top of a blog post is not the main content of the page. The blog post is. Google's video indexing guidelines are clear about this—the video needs to be the primary content of its own page to be eligible for video rich results. Second, the search intent for "SaaS onboarding checklist" is a list. The searcher wants a checklist, not a video. The video was answering a question the searcher hadn't asked.

What I got wrong: I assumed that putting the video in front of traffic would convert the traffic. But traffic is not the same as intent. A person reading a checklist doesn't want to watch a video. A person searching "how to onboard users" might. The page and the query have to match the format.

What I'd do differently: I'd build a dedicated page for the video. Not a blog post with a video. A page where the video is the hero, the transcript is the body, and the CTA is the product. That page targets a video-intent query. The blog post stays text. Two pages, two intents, two formats.

This was the second-most-expensive failure of the twelve months. It cost me three days of work and produced nothing. But it taught me the single most important rule of SaaS video SEO—the video must be the page, not a component of the page.

Experiment 3: The Comparison Page Video — The Surprise Winner

The hypothesis: If I create a page comparing our product to a competitor and put a video on that page, the page will rank for the comparison query and the video will drive signups.

The setup: We built a page targeting "[Our Product] vs [Competitor]." The page had a text comparison, a pricing table, and a 60-second video showing the specific workflow where we win. The video was the proof. The text was the argument.

The result: This page became the highest-converting page on our site. Not the highest-traffic—the highest-converting. Visitors who watched the video on the comparison page converted to trial at 4.2x the rate of visitors who didn't. The page now ranks in the top three for the comparison query, and the video appears in Google's video carousel for it.

Why it worked: Because the video had a job that text couldn't do. Text can argue. Video can demonstrate. On a comparison page, the searcher is asking "is this actually better?" A video that shows the difference answers that question in forty seconds. Text takes ten minutes and is less convincing.

This is the one experiment where the format and the query aligned perfectly. The searcher wanted proof. The video delivered proof. The page was built around the video, not the other way around.

What I got wrong: Nothing, honestly. This one worked because it was the only experiment where I designed the page around the video from the start, instead of adding a video to an existing page. LeoStudio had told me to do this from the beginning. I only listened on experiment three. You can see how they think about page-video alignment throughout their case work at leostudiohq.com—the page and the video are one asset, not two.

What I'd do differently: I'd build comparison pages for every competitor, each with its own video. That's now our roadmap for the next two quarters.

Experiment 4: The AI Overview Play — Real Traffic, Wrong Audience

The hypothesis: If our video transcript contains the phrases AI Overviews pull from, our video will appear as a citation in Google's AI Overviews for onboarding-related queries.

The setup: I wrote a detailed transcript for the onboarding video, published it as text on the video landing page, and structured the page with headers and clear answers to common questions. I targeted queries that were starting to show AI Overviews, like "how to improve SaaS onboarding" and "what is a good time to value for SaaS."

The result: Our content started appearing in AI Overviews within about six weeks. The traffic was real—about 1,200 visits a month from AI Overview citations. The problem was that the traffic didn't convert. Visitors who arrived from AI Overviews had a bounce rate of 78%. They read the answer, got what they needed, and left.

Why it underperformed: Because AI Overviews answer the question directly. The searcher doesn't need to visit your site. When they do visit, it's usually to verify or to read more, not to buy. The citation was good for brand visibility. It was bad for pipeline.

What I got wrong: I measured success by traffic instead of by signups. The AI Overview citations looked great in a report. They produced almost nothing in revenue.

What I'd do differently: I'd still pursue AI Overview visibility, because brand recognition has long-term value. But I wouldn't measure it by signups. I'd measure it by branded search volume—how many more people search for our brand name after seeing the citation. That's the metric that actually reflects what AI Overviews do.

The lesson here isn't that AI Overviews are worthless. It's that different distribution channels produce different outcomes. YouTube search produces signups. AI Overviews produce awareness. Both matter. But you have to measure them by what they actually do.

Experiment 5: The Video Sitemap Play — Invisible Work, Real Indexing

The hypothesis: If I create a video sitemap and submit it to Google Search Console, Google will index the video faster and rank it more consistently.

The setup: I created a video sitemap for the onboarding video landing page. It included the thumbnail URL, the title, the description, the content URL, and the player URL. I submitted it to Search Console and monitored the indexing status.

The result: No dramatic change. But over three months, the video's indexing consistency improved—it stayed in Google's video carousel more reliably, and it appeared in more query variations than before. The sitemap didn't create visibility. It stabilized it.

Why it mattered: Without a video sitemap, Google has to discover the video through normal crawling. That works, but it's inconsistent. With a sitemap, Google knows exactly what the video is, where it lives, and when it was published. The video appears in search results more reliably, and it stays there longer.

What I got wrong: I expected the sitemap to create a step change in traffic. It doesn't. It's infrastructure work. It's the equivalent of a well-structured site—it doesn't produce visibility on its own, but it makes everything else more consistent.

What I'd do differently: Nothing, actually. The sitemap was cheap to build and it made the video's search performance more reliable. It's the kind of unglamorous work that separates a video that ranks for three months from a video that ranks for three years. I'd do it for every video from day one.

Experiment 6: The Video-First Content Cluster — The Long Game

The hypothesis: If I build a cluster of pages around a single topic—each page with its own video and its own transcript—the cluster will rank collectively and the videos will reinforce each other's authority.

The setup: I built five pages, each targeting a different onboarding-related query. Each page had its own short video—40 to 60 seconds—showing a specific answer. Each page had a transcript, VideoObject schema, and internal links to the other four pages. The videos were not repurposed. They were purpose-built for each page.

The result: This was the second experiment that worked. The cluster now ranks for fourteen different queries. Three of the pages appear in Google's video carousel. The cluster collectively drives more signups than any single page on the site, including the homepage.

Why it worked: Because each page answered one specific question with one specific video. The videos weren't general. They were surgical. One video showed how to connect a data source. Another showed how to invite a teammate. Another showed how to read the first report. Each one was 45 seconds long and did one job.

The internal linking mattered too. Google used the cluster structure to understand the relationship between the pages. The authority flowed from the strongest page to the weakest. Over six months, the weakest page climbed from position 40 to position 6.

What I got wrong: I built the cluster backwards. I started with the topic and then wrote the pages. I should have started with the queries and then built the videos around them. The two weakest pages in the cluster are the ones where I didn't research the query first.

What I'd do differently: I'd run keyword research before building any video. Not after. The video is the most expensive part of the process. Don't build it until you know what query it's supposed to rank for.

This experiment is where LeoStudio and I finally aligned our processes. They build videos around the moment. I build pages around the query. The cluster is where those two things meet. You can see how they think about moments and triggers throughout leostudiohq.com—the moment is the video's job, and the query is the page's job. Get both right and the cluster compounds.

The Mistakes That Cost Me the Most

I want to be specific about the mistakes, because the experiments are only useful if you know what went wrong.

I chased volume keywords for three months. I targeted "how to onboard customers" because it had a search volume of 8,000 per month. It brought traffic that didn't convert because the intent was too broad. The specific queries—"SaaS onboarding checklist," "reduce time to value"—had less volume but far higher intent. I should have started narrow.

I embedded video on pages that weren't built for it. The blog experiment failed because the page was a checklist and the video was an onboarding walkthrough. The format didn't match the intent. I now build the page around the video, not the other way around.

I treated YouTube and Google as the same surface. Two months of YouTube work produced zero Google visibility. The YouTube optimization and the Google optimization are different jobs. Both matter. Both need their own work.

I measured traffic instead of signups. The AI Overview experiment produced 1,200 visits a month and almost nothing in revenue. I celebrated the traffic and missed the point. Signups are the metric. Everything else is diagnostic.

I ignored the retention signal. For six weeks, I blamed the algorithm for my low view count. The real problem was that the video wasn't holding attention well enough. When I fixed the first fifteen seconds, the rankings followed. Retention drives ranking. Not the other way around.

I published the video without a transcript for the first four months. Transcripts make videos indexable. AI Overviews pull from transcripts. Perplexity and ChatGPT pull from transcripts. Without a transcript, the video is invisible to every text-based search surface. I add a transcript to every video now, before I publish.

I built the cluster backwards. I started with the topic and wrote pages around it. I should have started with the queries and built the videos around them. The pages where I did the research first performed twice as well as the pages where I didn't.

I forgot to link the videos to each other. Internal linking within a video cluster is how authority compounds. My first version had no internal links. The second version had five. The second version ranked faster and higher.

The Framework: How I Promote Every SaaS Video Now

Here's the process I run before and after publishing any SaaS video that's supposed to be found.

Step one: Research the query before writing the script. The video is the most expensive part of the process. Don't build it until you know what query it's for. If the query is a list, build a list. If the query is a comparison, build a comparison. If the query is a how-to, build a video.

Step two: Build a page around the video, not the other way around. The video is the main content. The transcript is the body. The title, the H1, and the CTA all support the video. The page exists to host the video, not to host an article with a video embedded.

Step three: Optimize the YouTube upload separately from the Google page. YouTube gets the title, thumbnail, description, chapters, and tags. Google gets the schema, the sitemap, the transcript, and the internal links. Two surfaces, two optimizations.

Step four: Publish the transcript at the same time as the video. Never publish a video without a transcript. The transcript is how text-based search engines find the video. It's also how AI Overviews and AI assistants read it.

Step five: Submit the video sitemap. Immediately after publishing. Monitor the indexing status in Search Console. The sitemap doesn't create visibility, but it makes visibility more consistent.

Step six: Build the cluster. Don't publish one video. Publish a cluster of three to five videos, each targeting a different query, each linking to the others. The cluster compounds. The single video doesn't.

Step seven: Measure by signups, not views. Track which videos produce signups. Track which pages produce signups. Track which queries produce signups. The video with the most views is not the video that's working. The video that converts is.

Step eight: Revisit the retention. Every quarter, check the completion rate on every video. If completion is dropping, the video is decaying. Fix the first fifteen seconds. Add chapters. Refresh the thumbnail. Retention is the ranking signal. It's also the first thing to decay.

This framework came out of twelve months of experiments, two of which worked and four of which didn't. It's the version of the process I wish I'd had on day one.

This is the discipline LeoStudio brings to their own work—the video is built for a moment, and the promotion is built for the query. The video and the page are one asset, not two. You can see how they think about this throughout leostudiohq.com, and it's the reason our video is a channel instead of an asset.

The Final Word

I published eleven videos in two years. Eight of them were watched by almost nobody. One of them became a channel. The difference was not the video. It was the promotion.

The video that worked had a strong first fifteen seconds. It had a dedicated page where it was the main content. It had a transcript, a schema, a sitemap, and a cluster of four sibling videos. It ranked on YouTube for eleven queries and on Google for three. It appeared in the video carousel. It produced signups.

Every one of those things was a decision I made after a failure. The YouTube ranking came after six weeks of nothing. The comparison page came after the embedded blog video failed. The cluster came after I realized single videos don't compound. None of it was obvious at the start.

If you're trying to promote a SaaS video through search, start with the query. Build the page around the video. Optimize YouTube and Google separately. Publish the transcript. Submit the sitemap. Build a cluster. Measure by signups. Then revisit the retention every quarter, because that's the signal that everything else depends on.

We built the video with LeoStudio, a team that understood from the first call that the video is an intervention and the search is the delivery. If you want to see how they think about moments, triggers, and the first fifteen seconds that decide everything, spend time on leostudiohq.com. Their approach to clarity is the reason the video held attention. The promotion is the reason people found it.

Tell them the SEO guy sent you. Tell them you already built the cluster. Tell them you already measured the retention. They'll know exactly what you mean.