YouTube algorithm for businesses: how it picks videos
Short answer YouTube doesn't rank videos by one score. It matches each viewer with videos that person is likely to click, keep watching and be glad they watched, using clicks, watch time, likes, dislikes, shares and survey answers. The first day mostly reflects your regular viewers. YouTube keeps testing a video with new viewers over time, so a slow start isn't final.
Want help applying this to your business? Talk it through on a free 30-minute call.
Book a callHow the YouTube algorithm picks videos
It picks viewers, not winners. For each person, YouTube guesses which videos that person will click, watch and feel good about, then shows those. YouTube's VP of Engineering put the job this way:
"the job of our recommendation system is to find that audience." (YouTube Official Blog, 2021)
So there is no fixed formula to game. The same blog post says the system "doesn't follow a set formula, but develops dynamically as your viewing habits change." For a business, that means one thing: make a video that a specific kind of customer is glad they watched, and give YouTube enough clues to find those people.
Which signals does YouTube use?
YouTube names its signals: "clicks, watchtime, survey responses, sharing, likes, and dislikes" (YouTube Official Blog, 2021). Its public "How YouTube Works" page adds subscriptions, viewer feedback, and the reputation and quality of the channel (How YouTube Works: Recommendations).
| Signal | What it tells YouTube | What it can't tell |
|---|---|---|
| Click | The title and thumbnail made someone curious | Whether they stayed |
| Watch time | They kept watching | Whether they were glad they did |
| Survey answers | Viewers rated the video 1 to 5 stars | Nothing alone; it checks the others |
| Likes, dislikes, shares | How viewers felt and whether they passed it on | |
| Subscriptions, feedback | What a person wants more or less of |
Order matters here. YouTube says it learned in 2011 that a click doesn't mean someone watched, so it added watch time. Then it found "not all watchtime is equal," so it now measures valued watch time: time on videos that surveyed viewers rate four or five stars (YouTube Official Blog, 2021). A long video people feel stuck in doesn't win.
Click-through rate (CTR, the share of people who click after seeing your thumbnail) only counts if the video delivers. YouTube Help says half of channels and videos fall between 2% and 10%, and warns that clickbait, high CTR with short viewing, is "less likely to get recommended" (YouTube Help: Impressions and click-through rate).
Search vs recommendations: two ways in
Your video reaches people in two ways, and they work differently.
| Search | Recommendations | |
|---|---|---|
| Where | Results after someone types a query | Home page, Up Next, Watch Next |
| What decides it | Relevance, engagement, quality (YouTube Help) | The signals above, matched to each person |
| Best for | New channels, "how much does X cost" questions | Channels with viewers YouTube already understands |
For search, YouTube says it estimates relevance from "how well the title, tags, description, and video content match your search query" (YouTube Help: How YouTube search works). A local business with few subscribers usually starts here: answer one question customers really type, in their words. How those videos also show in Google is in YouTube vs Google.
Do the first 24 hours decide a video's fate?
No. The first day matters less than most guides claim. YouTube's Creator Liaison explains that a new upload usually gets an early spike from regular viewers, then YouTube keeps testing it with new audiences until that pool runs out. He adds: "YouTube never gives up, though!" A video can find new viewers later through search, current events, a similar video trending or a shout-out (YouTube Official Blog, March 2024).
Timing works the same way. Asked whether the hour you publish matters, YouTube answered:
"The algorithm tries to follow the audience, so if it doesn't matter to the audience, it probably doesn't matter to the algorithm." (YouTube Official Blog, November 2023)
So a launch push helps only if it brings the right people. A business video about furnace repair is built to be found in search for years, not to trend on day one.
What to do in the first 24 hours
Spend the first day on things that bring real customers, not on inflating numbers.
- Send it to people who asked. Email it to customers who raised that question, or link it from the matching service page. They watch longest.
- Check the click, not the views. In YouTube Studio, compare CTR and average view duration. High clicks with short viewing means the title or thumbnail promises something the video doesn't give (YouTube Help). Fix the title, not the video.
- Answer comments. It helps the next customer who reads them. YouTube judges quality by viewers, not effort: "It's how the video made the audience feel." (YouTube Official Blog, July 2024)
- Skip the shortcuts. Asking friends or staff to watch sends YouTube the wrong audience. It may then show the video to more people like them, not like your customers.
Paid views are a separate tool; see YouTube ads for a local business.
Example
A roofer's cost video
A roofer posts "How much does a roof replacement cost in Denver?" Day one: 40 views, mostly past customers. Over six months it keeps getting a few views a day from search, because people keep typing that question. Total: 3,000 views and several quote requests. The first day said little.
A dentist chasing day-one numbers
A dental office asks 50 staff, friends and family to watch a new whitening video on day one. Views jump, but they're not people looking for whitening, so many stop early. YouTube now has weak signals about who the video is for. Sending it to patients who asked about whitening would have taught YouTube more.
FAQ
Does YouTube favor big channels?
Not by size alone. It recommends videos to viewers it expects to be satisfied, and it does consider a channel's reputation and quality (How YouTube Works). A small channel answering a specific question can rank in search.
Is it bad to delete and re-upload a video that started slow?
There is no need. YouTube keeps testing videos over time (YouTube Official Blog, March 2024). Changing the title or thumbnail is the smaller fix.
Does the time I publish matter?
Only as much as it matters to your viewers. YouTube says the algorithm follows the audience (YouTube Official Blog, November 2023).
What is a good click-through rate?
Half of channels and videos sit between 2% and 10%, and new videos vary even more (YouTube Help). Compare against your own past videos, not other channels.
Different for you?
- You want views now, not over months: paid placement works on a different system. See YouTube ads for a local business or /paid-ads/.
- You want the video found in Google too: see Google video carousel.
- The video lives on your own website: see Video SEO for websites.
- You don't know which questions to film: start from what customers search. See /seo/.