Every social platform hands you a dashboard full of numbers, and the biggest one — view count — is the one that tells you the least. A video can collect a hundred thousand views and sell nothing, because a view is counted the instant a feed stops scrolling, not when anyone decides your content was worth their time. The metrics that actually predict whether a short-form video gets pushed, and whether it does anything for the business behind it, are the quieter ones underneath: how long people watched, how many finished, where they dropped off, whether they saved or shared it, and whether they clicked or converted. This guide is the honest map of that dashboard. It groups the metrics into the three questions they answer — did it reach anyone, did it hold anyone, did it move anyone — gives the formula behind each one, spells out the counting traps that make the same clip report wildly different numbers across platforms, explains why the published benchmarks you will find disagree by an order of magnitude and what to compare against instead, names the one metric to optimize for each goal, and closes on the part the dashboards leave entirely to you: acting on what the numbers say, at the cadence that makes acting on them worth it.
Social video metrics are the measurements a platform reports for a video post — how many people it reached, how long they watched, how they interacted, and what they did next. Every one of Instagram, TikTok, YouTube, Facebook, LinkedIn, and X exposes a version of this dashboard, and the trouble starts with how similar the numbers look and how differently they are defined. The same thirty-second clip, posted to all of them, will report six different view totals and six differently-shaped engagement figures before a single real difference in audience is accounted for.
So the useful skill is not memorizing every number. It is knowing which question each number answers and how much to trust it. This guide groups the whole dashboard into three questions — did the video reach anyone, did it hold anyone, did it move anyone — because that grouping is also a priority order. Reach without attention is a feed stopping on your thumbnail and moving on. Attention without action is people enjoying your content and buying nothing. The metrics only become useful when you read them in that sequence.
Most "vanity metric" arguments collapse into one real distinction. Reach metrics (views, impressions, reach) measure exposure — how many times the platform put the video in front of someone. Attention metrics (watch time, average view duration, completion, retention) measure whether those people actually watched. Action metrics (engagement rate, shares, saves, click-through, conversion) measure whether watching changed anything. A healthy video climbs all three; a video that has reach but no attention was served widely and ignored, and a video with attention but no action was enjoyable and pointless.
The reason this matters more on short-form than it used to is that the algorithms themselves reward the middle tier. Platforms learned years ago that a click is not a watch, and they rank on attention — retention and completion — far more than on raw exposure. That means the metric you most want to move for reach is not the view count at the top of the dashboard; it is the retention curve two scrolls down. Get the attention tier right and the reach tier tends to follow, because the algorithm does the distributing for you.
Views, impressions, and reach are the exposure layer. Impressions count every time the video was displayed, including repeats; reach counts unique accounts; views count plays. The trap is the view, because "play" is defined differently on every platform, which makes view count the single most misread number in the creator economy.
Here is how the major platforms count a view, as of late 2026. TikTok fires a view after one second of play (three seconds for videos three minutes and longer). Instagram Reels and Facebook Reels count a view on play, including replays; in-feed video on both counts at three seconds. LinkedIn and X use the stricter advertising standard set by the Media Rating Council — at least two continuous seconds with at least 50% of the player on screen. Snapchat counts a view after one second on organic Stories. YouTube is the one that changed: since August 24, 2026, it counts a public view from the first frame across long-form videos, live streams, and podcasts, matching the near-instant rule it already used for Shorts — the full breakdown is in YouTube's view-counting change.
The practical rule falls straight out of that list: never compare a view count across platforms, because you would be comparing a one-second TikTok play against a two-second-with-viewability X view against an instant YouTube frame. A view count is only meaningful against other view counts on the same platform. The same caution applies to anyone citing a platform's headline view number as proof of reach — on an autoplay feed it is closer to "the feed scrolled past this" than "a person watched this."
This is the tier that predicts distribution. Four numbers live here, and they are related. Average watch time is the mean time viewers spent on the video; YouTube reports it as average view duration alongside average percentage viewed. Completion rate is the share who watched to the end — completed views divided by total views, times 100. The retention curve is the shape of attention over the video's length: a graph of what percent of viewers are still watching at each second, with the near-universal sharp drop in the first three seconds followed by either a slow decay (good) or a cliff (bad). And watch time in the cumulative sense — total minutes across all viewers — is what long-form YouTube ranks on.
Read them together, not separately, because length distorts each one in isolation. A fifteen-second clip and a three-minute clip with identical completion rates are not equally successful — the longer one delivered far more total attention per view. Conversely a long video with low completion can still generate strong watch time. The retention curve is the diagnostic that cuts through this: it shows you not just that people left but where, so a cliff at the eight-second mark tells you the hook over-promised, and a sag in the middle tells you to cut. The task-level playbook for reading and fixing the curve is in improve YouTube audience retention and the strategy version in YouTube audience retention strategy.
Two platform-specific notes worth holding. Instagram weighs total seconds viewed, not just percentage watched, which quietly favors videos that hold attention in absolute time rather than ones that are merely short enough to "complete." And platforms report attention at different granularities — LinkedIn gives you views at 25/50/75% and completion; TikTok gives average watch time and full-watch percentage; YouTube gives the full retention graph. Pull the native number for the platform you are judging; do not translate one platform's attention metric into another's.
Engagement is the interaction layer: likes, comments, shares, and saves. Engagement rate rolls them into one figure — interactions divided by reach or impressions, times 100 — but the single number hides the fact that the actions are not weighted equally by the algorithms. Saves and shares are the strong signals: a save says "I want this again," a share says "someone else should see this," and on Instagram and TikTok both are primary ranking inputs. Comments are second. Likes are nearly worthless to modern ranking — too cheap, too automatic.
That reordering has a direct content consequence. Reference-style content — tutorials, templates, checklists, "save this before it disappears" — earns saves, and save-heavy content gets pushed further than like-heavy content at the same reach. Shareable content tends to be either highly useful or highly relatable. If you are optimizing for distribution rather than applause, you optimize the video for the save and the share, not the like. The denominator caveat matters here too: an engagement rate is only comparable when you know whether it was divided by followers, reach, or impressions, so confirm the basis before trusting a benchmark — the industry-by-industry reality is in social media engagement rates by industry.
The final tier measures whether attention turned into a result. Click-through rate is clicks divided by views, times 100 — the share of viewers who followed a link or CTA, whether that is a YouTube end screen, a TikTok bio link, an Instagram Stories link, or a link in LinkedIn post copy. Conversion rate is the share of viewers (or of clicks) who completed a goal — a sign-up, a trial, a purchase — expressed as a percentage. And video ROI is the money question: (revenue from the video minus its cost) divided by cost, times 100. An $8,000 campaign that drives $26,000 in attributed revenue returns 225%.
The hard part of this tier is attribution, not arithmetic. Native video dashboards do not connect a view to a sale — they stop at the click. To measure conversion and ROI you have to carry the viewer across the gap yourself with UTM-tagged links and a pixel (Meta Pixel, TikTok Pixel, LinkedIn's Conversions API), or the revenue simply never shows up next to the video that caused it. For awareness-led video where a direct sale is not the point, cost-based proxies — cost per thousand impressions, cost per view, cost per engaged viewer — are the honest stand-ins, and they are cheaper to track than a fabricated conversion number.
Search for a "good" completion rate or engagement rate and you will find figures that disagree by an order of magnitude — 45% here, 85% there; 3.8% in one table, 50% in another. They are not all wrong. They are measuring different things under the same word. "Completion" means the share of viewers who reached the end in one source and the average share of the video watched in another. Engagement rate divides by followers in one tool and by reach in the next. Length swings everything — a seven-second clip completes far more often than a sixty-second one — and niche swings it again, with entertainment out-engaging B2B at equal creative quality.
So treat every published benchmark as a rough reference point, never a target, and never cite one without knowing its definition and denominator. The reliable baseline is the one the dashboards give you for free: your own account's median for this metric, on this platform, at this video length, over time. A video that beats your own median is winning in the only comparison the algorithm is actually running — against your other content. If you do need an external number, read the methodology behind it before you trust it, because the headline figure almost always omits the definition that makes it meaningful.
The dashboard is a diagnostic set, but you steer by one metric at a time, chosen by the goal. If the goal is reach and algorithmic distribution, optimize retention and completion — holding attention is the lever the platform rewards, and the hook is where most retention is won or lost. If the goal is community and organic spread on Instagram or TikTok, optimize saves and shares, and shape content toward reference value and relatability. If the goal is traffic, optimize click-through rate and make the CTA single and specific. If the goal is revenue, optimize conversion and instrument it properly, because an un-pixeled funnel reports zero no matter how well the video sold.
The discipline is to pick one north-star per campaign and demote the rest to diagnostics. A video optimized for saves will not have the highest click-through rate, and that is fine — you read CTR to understand why, not to judge the video against a goal it was not built for. The mistake is treating all eleven numbers as a scoreboard to maximize at once; they trade off against each other, and a dashboard with everything green usually means nothing was actually optimized.
Here is the honest limit of everything above: a metric you cannot act on is a scoreboard, not a tool. The retention curve tells you the hook failed at second three; the completion rate tells you the fifteen-second cut outperformed the forty-five; the save rate tells you the checklist format travels and the talking-head does not. None of that is worth anything unless you can turn the finding into the next batch of videos fast enough that the trend is still alive when they land. The platforms give you the diagnosis and leave the treatment — the re-cut, the new hook, the format you now know works, published again across every platform — entirely to you. That production gap is where most measurement dies: the insight is sound and nobody has the throughput to act on it.
Kompozy is built for that half of the loop. It is a full AI content generation and multi-platform publishing engine — not a repurposing tool — so when the metrics say "shorter wins" or "this hook holds," you act on it by producing, not by re-shooting. A single Persona Brief holds the voice, so you can generate several hook variants of the same idea and ship them as Persona Shorts, Clipped Shorts, or listicle and Marketing Shorts and let the retention numbers pick the winner — the deliberate practice of making videos the metrics will reward is covered in AI-generated videos optimized for engagement. Because word-synced captions lift completion on sound-off feeds, they are burned in by default, so the attention tier starts ahead.
The part that makes the comparison trustworthy is distribution. Autopilot fans the same video across the eight social platforms plus blog and email from one queue with a per-post review gate, so you are reading each platform's native attention metric for the same creative — a real apples-to-apples test instead of a different edit per app — and the format that completes best becomes the template for the next week's batch. That turns the dashboard from a report you look at into a loop you run: measure on the platform's own numbers, learn which attention and engagement signals your audience actually gives, and produce the next cohort at a cadence that keeps you inside the window the metric was measured in. Measurement is the easy half; the engine is how a small team runs the half that actually moves the numbers.
Group them by the question they answer. Reach: view count, impressions, reach. Attention: average watch time, average view duration, completion rate, and the retention curve. Engagement: engagement rate, shares, saves. Action: click-through rate, conversion rate, and video ROI. The attention metrics predict whether a short-form video gets pushed; the action metrics tell you whether it helped the business. View count on its own is exposure, not success.
Mostly, on its own. A view is counted almost the instant a video starts on most feed platforms — one second on TikTok, on play for Reels — so a high view count can mean the feed served it widely, not that anyone watched. View count matters as a denominator for rate metrics and as a reach signal, but it says nothing about whether the content resonated. Pair it with watch time and completion before you call a video a success.
Completion rate is completed views divided by total views, times 100 — the share who watched to the end. Engagement rate is interactions (likes, comments, shares, saves) divided by reach or impressions, times 100; confirm the denominator, because tools differ. Click-through rate is clicks divided by views, times 100, and video ROI is (revenue minus cost) divided by cost, times 100.
Because the same metric name hides different definitions and different denominators. "Completion" can mean the share who reached the end or the average share of the video watched; engagement rate can divide by followers, reach, or impressions. Length, platform, and niche swing the numbers further. Published benchmarks are rough reference points at best — the reliable baseline is your own account's median over time.
It depends on the goal. For reach and the algorithm, optimize retention and completion — holding attention is what gets a video pushed. For community and saves-driven distribution on Instagram and TikTok, optimize shares and saves. For traffic, optimize click-through rate. For revenue, optimize conversion and track it with UTM links or a pixel, because native video dashboards do not attribute sales. Pick one north-star metric per goal and read the rest as diagnostics.
Social video metrics are the numbers that tell you whether a short-form video worked. They sort into three questions: did it reach anyone (views, impressions, reach), did it hold anyone (watch time, average view duration, completion, the retention curve), and did it move anyone (engagement rate, shares, saves, click-through, conversion). View count is exposure, not success — the attention metrics predict reach, and each platform counts a view differently, so compare within a platform and against your own baseline, never across apps.
Get started → · ← All guides · Compare Kompozy vs other tools