Learning how to measure video performance starts with a hard admission: most of the numbers on your dashboard right now have nothing to do with revenue. Views, impressions, and likes tell you a video existed. They do not tell you whether it worked.
Nine metrics actually predict conversions, and they run in a specific order: leading indicators first, lagging outcomes last. Every one of them connects to a dollar figure somewhere downstream.
This article ranks all nine, tells you which ones to stop reporting, and shows you how to wire the ones that matter into your CRM so a good watch-depth number turns into a sales follow-up instead of a slide in next week's deck.
This isn't a diagnostic on why your video might be underperforming. If playback issues like buffering, slow starts, or poor resolution are the suspected culprit, that's a separate investigation, covered in our breakdown of how playback quality affects activation and conversion.
This article assumes your video plays cleanly and answers a narrower question: once it's playing, which numbers do you actually track, how do you calculate each one, and how do you get them into a system your sales team will act on.
Gumlet is the all-in-one video hosting platform used throughout the examples below, covering hosting, analytics, and CRM event delivery in one system.
If you are a marketer, demand-gen lead, or founder who already has videos and analytics but cannot tell which numbers deserve a seat at the revenue table, this is that shortlist.
Key Takeaways
- Most "video metrics" are vanity. Conversions are forecast by a small set: play rate, watch depth, and CTA click-through, not views, impressions, or likes.
- Play rate tells you if the video is in the right place. Engagement and watch depth tell you if the message landed. CTA click-through rate and conversion rate tell you if it paid off.
- Less than half of marketers connect their video platform to their CRM or email tool, according to Wistia's 2026 State of Video Report. That gap is why most watch behavior never becomes pipeline.
- Set a watch-depth threshold, such as 75% of a demo, as your video-qualified-lead line, then sync it to the CRM.
- Segment every metric by video and by placement. A homepage video and a product-page video with identical view counts can convert at very different rates.
- A view your CRM can attribute to a contact is pipeline. A view it cannot attribute is just content.
Vanity Metrics vs. Metrics That Predict Conversions
A metric is only worth reporting if a change in it changes revenue. Most video dashboards are full of numbers that do not clear that bar.
Draw the line in three groups. Awareness metrics, including views, impressions, and reach, tell you how many people showed up. Engagement metrics, including play rate, watch depth, and retention, tell you if they cared. Conversion metrics, including CTA click-through rate, conversion rate, video-qualified leads, and influenced pipeline, tell you if it worked.
Only the last two groups predict revenue. Awareness metrics set context. They do not forecast outcomes.
Raw view counts deserve a specific warning here. Every platform counts a "view" differently: YouTube, Wistia, GA4, and Gumlet each apply their own threshold for what qualifies as a play.
Comparing raw view counts across platforms without accounting for that difference produces a number that looks precise and means nothing. Compare engagement rates instead, since those are percentage-based and normalize for the definitional gap between platforms.
The table below sorts every metric that shows up on a typical video dashboard by whether it actually predicts conversion, or whether it is context dressed up as insight.
| Metric | What It Measures | Predicts Conversion? | Benchmark / How to Read |
|---|---|---|---|
| Views | Times video started | Vanity alone | Meaningless without play rate and depth |
| Impressions / reach | People who saw the thumbnail | Awareness only | Top-of-funnel context |
| Play rate | Visitors who pressed play ÷ page visitors | Leading | Highest on homepages, video galleries, and product pages |
| Engagement rate / AVD | Percentage of the video watched | Leading | Videos under 1 minute average roughly 52% engagement |
| Retention / drop-off | Where viewers leave | Diagnostic | Find the cliff, fix the hook |
| Watch depth (e.g., 75%) | Hit the intent threshold | Strong leading | Your video-qualified-lead line |
| CTA click-through rate | Clicked the in-video CTA | Direct | The bridge to conversion |
| Conversion rate | Viewer to lead or customer | Outcome | Segment by video and placement |
| Video-qualified leads / influenced pipeline | Watch behavior tied to a contact | Closed-loop | Requires CRM connection |
If a metric can go up while revenue stays flat, it is a vanity metric. Report it last, or not at all.
Before You Track Anything: A 4-Question Setup Checklist
Skip this and every metric below becomes a number with nowhere to go. Answer these four questions first, in order, before you open a single dashboard.
1. What funnel stage is this video actually in?
Awareness videos (homepage, ads, social) get judged on play rate and reach. Consideration videos (product pages, comparison content) get judged on watch depth. Conversion videos (demos, pricing explainers) get judged on CTA click-through and video-qualified leads.
Tag every video with one stage before you assign it a KPI. A single video trying to serve all three stages will produce metrics that contradict each other.
2. Does your video platform expose per-viewer, per-milestone data, or only aggregate stats?
Aggregate engagement rate tells you what the average viewer did. It cannot tell you that one specific contact hit 75% of your demo yesterday.
If your platform only reports averages, you can read this whole article and still have nothing to send to sales. Confirm milestone-level, per-viewer event data is available before you set a threshold in question 3.
3. What watch-depth threshold counts as a video-qualified lead for this specific video type?
Not one number for every video. A 90-second product tour and a 12-minute technical demo don't share a threshold. Set it per video type (see the table below), and write it down somewhere your team will actually reference it.
4. Is your video platform connected to your CRM, and does it fire events on a schedule you can act on?
Real-time and daily-batch are usable. Weekly exports into a spreadsheet are not, because by the time you see the data the lead has gone cold. Confirm the connection exists and confirm its update frequency before you build a single report around it.
| Video type | Typical funnel stage | Starting watch-depth threshold |
|---|---|---|
| Homepage / brand explainer | Awareness | Not applicable; measure play rate and reach instead |
| Product or feature page video | Consideration | 50% |
| Demo or product walkthrough | Conversion | 75% |
| Pricing or ROI explainer | Conversion | 60% |
| Customer testimonial / case study | Consideration | 50% |
Once you can answer all four questions, the nine metrics below stop being a list to read and start being a system to run.
The 9 Video Metrics That Actually Predict Conversions
Each metric below is ranked from leading indicator to lagging outcome. Every one of them ties back to the same question: does a change here move revenue, and if so, how directly.
Most of these you can measure directly inside Gumlet's analytics dashboard, and the CTAs below point to where.
1. Play Rate: Do They Even Start?
Play rate is the share of page visitors who press play. It is calculated as plays divided by page loads, and it should be read per page, since a video's play rate depends heavily on where it sits on that page.
Play rate matters because it is the first leaky point in the entire funnel. Doubling play rate on a product page multiplies every downstream conversion that follows, since nothing downstream can convert if nobody presses play in the first place.
Wistia's 2026 State of Video Report found that videos on homepages, video galleries, and product pages get the highest play rates of any page type, since visitors there are already exploring and a video is a natural next step.
A video no one presses play on has a conversion rate of zero. Fix placement before you fix the video.
2. Engagement Rate and Average View Duration: Did the Message Land?
Engagement rate is the percentage of the video an average viewer watches. The formula is total time played divided by the product of total plays and video length.
A 2-minute video played 1,000 times for a total of 10 hours of watch time has a 30% engagement rate.
Engagement rate predicts conversion because viewers who never reach your value proposition or your video call-to-action (CTA) cannot act on either one. Depth is a prerequisite for conversion, not a nice-to-have.
Wistia's 2026 State of Video report puts videos under one minute at roughly 52% average engagement. Longer videos run lower on a percentage basis but accumulate more total watch time, so shorter is not automatically better. It depends on whether your message needs 40 seconds or 4 minutes to land.
Watch depth is the tax every conversion pays. Viewers cannot buy a pitch they left before hearing.
3. Audience Retention Curve and Drop-Off: Where Do You Lose Them?
The retention curve is the second-by-second graph of who is still watching. The cliff on that graph shows exactly where attention dies, and what was on screen at that moment usually explains why.
Read the retention or heatmap view inside your video platform and look for the steepest single drop, then check what happens on screen right before it. That is almost always a fixable hook or pacing problem, not a content problem.
A drop-off before your CTA is a conversion you will never recover. Moving the ‘ask’ earlier, above the cliff, can lift results without shooting new footage.
Your drop-off curve is a map of every conversion you are leaking, and exactly where it happens.
4. Watch Depth at a Threshold: The Intent Line
Watch depth at a threshold is the share of viewers who cross a chosen milestone, such as 75% of a demo. This is your leading buying signal, the number that tells you someone is seriously considering you before they have told you so directly.
Set milestone events at 25%, 50%, 75%, and 100% of the video, then count how many viewers hit your chosen threshold. Someone who watches 75% of a 5-minute demo has shown more intent than someone who clicked an ad. It is the strongest pre-conversion signal available before a form fill.
Pick the watch-depth number that means "this person is seriously considering us." That is your video-qualified lead.
5. Re-Watches of Key Segments: The Evaluation Signal
Re-watches are viewers replaying a specific part of a video, most often pricing, specs, or security. This is a strong and consistently under-used buying signal, since most teams never look past aggregate engagement to see which seconds got watched twice.
The engagement heatmap inside most video platforms surfaces re-watched segments directly. Repeat viewing of a decision-critical segment often precedes a purchase question, and it tells your sales team exactly what the lead is stuck on before the first call.
When someone rewinds your pricing slide twice, that is not a metric. That is a sales trigger.
6. CTA Click-Through Rate: Did They Act Inside the Video?
CTA click-through rate is the share of viewers who click an in-video call to action, button, or form. It is calculated as CTA clicks divided by plays, and most video platforms log these actions natively.
This metric predicts conversion because it is the most direct in-player bridge from watching to converting. It is the last step before the form, closer to the outcome than any metric above it on this list.
Placement and timing matter more than most teams assume. According to Digital Applied's 2026 video marketing statistics report, interactive video elements deliver a 3.6x higher completion rate than traditional passive video formats. This massive performance surge occurs because viewers who watch all the way through are already deeply engaged and far more likely to take action.
CTA click-through is where attention becomes intent you can count.
7. Conversion Rate by Video and Placement: The Outcome
Conversion rate is the share of viewers who became a lead or customer, measured per video and per page. It should never be reported in aggregate, since two videos with identical view counts can convert at very different rates depending on where they sit and who watches them.
Calculate conversions attributed to the video divided by viewers, then segment by placement. Compare each video against its own page's baseline, meaning the same page's conversion rate with the video removed, rather than against an industry average that has nothing to do with your funnel.
One blended conversion number hides your best video and protects your worst. Segment or stay blind to which one is which.
8. Video-Qualified Leads Synced to the CRM: The Closed Loop
A video-qualified lead is a lead defined by watch depth plus an action, such as a CTA click or form fill, fired to the CRM contact record so sales actually sees it. This is the metric that turns anonymous watch behavior into a scored pipeline attached to a real person.
The mechanism is milestone events plus a threshold, synced to a CRM like HubSpot, Salesforce, or Marketo, which then triggers a follow-up sequence automatically. This is the same integration gap covered earlier: less than half of marketers connect video to their CRM at all, so most teams never reach this step, let alone benefit from it.
This widespread gap prevents viewer engagement data from fueling sales pipelines, causing many high-intent customer interactions to stall before reaching a sales representative.
Gumlet, the all-in-one video hosting platform, fires per-viewer watch-depth, CTA, and form events to CRM and analytics platforms via webhooks and native integrations, closing that loop directly.
Be clear-eyed about where the boundary sits: the video platform can tell you who watched what and how far. Revenue attribution itself lives in the CRM, not in the video tool.
A view your CRM never hears about cannot become a deal. The video-qualified lead is where the video finally shows up in pipeline.
To know more about how you can generate leads from your product demo videos, check out Gumlet’s video lead generation page to know more.
9. Influenced Pipeline and Revenue Attribution: The Business Metric
Influenced pipeline is the pipeline and revenue that video touched, the number a CFO actually cares about. It is the lagging confirmation that every leading metric above it in this list was real.
Calculate it through multi-touch attribution inside your CRM or marketing automation stack, with video engagement tracked as one touch among several. Report it as "influenced," not "caused," since attribution in a multi-touch environment is directional by design, not exact.
Every metric above is a promise. Influenced pipeline is the receipt.
Vanity Metrics to Stop Over-Reporting
A handful of numbers keep showing up in video dashboards despite predicting almost nothing about revenue.
Raw view count tops the list, since every platform counts a view differently and the number is not comparable across tools. Impressions and reach in isolation come next: they measure awareness, not intent, and treating them as a success metric confuses exposure with interest.
Likes and shares as a primary success metric make the same mistake in a different direction, since a share proves distribution, not conversion. Total watch time with no per-viewer context hides whether that time came from 10 highly engaged viewers or a 1,000 who each watched 3 seconds.
Any "average" metric that masks segment differences, such as a blended average across every page a video appears on, hides exactly the variation that matters most for decisions.
None of these five are useless. They are top-of-funnel context. The mistake is treating them as proof that a video worked.
Views are how you start the conversation, not how you know it worked.
How to Actually Measure This in Your First 30 Days: Turning This Into a System
The four-question checklist above gets you oriented. This is what to actually do, week-by-week, so it doesn't stay theoretical.
Week 1: Audit and Tag
List every video currently live on a page that matters (homepage, product pages, demo requests, pricing). Tag each one with its funnel stage using the checklist above. Do not skip videos you assume are "fine." Most teams find at least one high-traffic video with no assigned KPI at all.
Week 2: Confirm Your Data Pipeline
Check whether your video platform can report per-viewer, per-milestone events, not just aggregates. If it can, confirm it's actually wired to your CRM (HubSpot, Salesforce, Marketo, or whatever you run) and check the last event that landed on a contact record. If nothing has landed in the last 7 days, the connection isn't working even if it looks configured.
Week 3: Set Thresholds and Build One Report
Assign a watch-depth threshold per video type using the starting points in the table above, adjusted for your own baseline once you have a few weeks of data. Build a single report segmented by video and by placement. Do not build a blended, all-videos-averaged dashboard, it will hide the one video that's actually working.
Week 4: Review, Cut, and Scale
Look at what the report shows. Any video with a play rate near zero gets a placement fix before anything else. Any video with strong watch depth but no CRM-attributed leads gets a CTA or integration fix. Anything with neither problem gets more budget.
After 30 days, this stops being a project and becomes a five-minute weekly check.
Teams can track play rate, watch depth, drop-off, and CTA actions on Gumlet's free plan, then fire those events straight to their CRM from the Growth tier up, before adding a full attribution suite on top.
Common Mistakes That Make Video Metrics Lie
A handful of habits quietly corrupt video reporting even when the underlying data is accurate.
Reporting views as success is the most common, since a view proves exposure, not impact. Never connecting video to the CRM is the single biggest gap industry-wide, and it means watch behavior stays invisible to the people who could act on it.
Judging performance from blended averages instead of segmenting by video and placement hides which specific asset is actually working.
Comparing raw view counts across platforms treats numbers that are not comparable as if they were. Tracking total watch time without per-viewer depth answers "how much time" without answering "how many people actually got the message."
Setting no watch-depth threshold means engagement never triggers a concrete action, so an encouraging number just sits in a report. Optimizing a metric with no CTA attached to it wastes effort improving a number that has no path to revenue.
Most "our videos don't drive revenue" problems are really "our video data never reached the CRM" problems.
Frequently Asked Questions
1. How do you measure video performance?
Pick one KPI per funnel stage, then track it consistently: play rate or reach for awareness, watch depth for consideration, and CTA click-through rate or conversion rate for revenue-stage video.
The step most teams skip is connecting that data to the CRM, since a watch-depth number that never reaches a contact record cannot trigger a follow-up. Measuring video performance well means tying every metric back to what happens after the view, not just tracking the view itself.
2. What are vanity metrics in video?
Vanity metrics are numbers like views, impressions, and likes in isolation. They show that a video existed and reached people, but they do not show whether anyone acted on it. These metrics matter as top-of-funnel context, not as proof of success. The distinction that matters is whether a change in the metric changes revenue. Views alone rarely do.
3. Which video metric best predicts conversions?
Watch depth at a defined threshold, combined with CTA click-through rate, is the strongest predictor available before a form fill happens. Watch depth shows genuine intent, since someone who reaches 75% of a demo has invested real attention.
CTA click-through shows that intent converting into an action inside the player itself. Together, they form the closest thing to a pre-conversion signal that video analytics can offer. If you only track two numbers from this entire list, track those two.
4. What is a good play rate or engagement rate?
Both vary significantly by page type and video length, so there is no single universal benchmark. Play rate tends to run highest on homepages, product pages, and video galleries, since visitors there are already in an exploring mindset.
Engagement rate tends to run higher on shorter videos, with videos under one minute averaging around 52% based on Wistia's 2026 data, while longer videos see lower percentage engagement but more total watch time. Judge your own numbers against your own page's baseline rather than an industry-wide figure.
5. What is a video-qualified lead (VQL)?
A video-qualified lead is a contact who crosses a defined watch-depth threshold and takes an action, such as clicking a CTA or filling a form, with that behavior synced to their CRM record. It turns anonymous watch data into a scored signal sales can act on.
The threshold itself varies by video length and funnel stage, so most teams set it per video type rather than using one number across every asset.
6. How do I tie video to revenue?
Connect your video platform to your CRM so watch events land on the contact record, then use multi-touch attribution inside the CRM or marketing automation stack to credit video as one touch among several.
Report the result as "influenced pipeline," not as a direct cause, since attribution in a multi-touch funnel is inherently directional. The connection step is the one most teams skip, and it is the reason watch behavior so rarely shows up in revenue reporting.
7. What tools measure video performance?
Native platform analytics like YouTube Studio and GA4 cover basic tracking for free.
Purpose-built video platforms differ mainly in whether they push data to a CRM: Wistia focuses on marketing-side lead capture, Vidyard focuses on sales-side video with HubSpot and Salesforce sync, and Gumlet fires per-viewer watch-depth, CTA, and form events directly to HubSpot, Salesforce, and Marketo via webhooks.
Choose based on whether your video sits on a public marketing page or behind an authenticated product surface, since that determines which integration model actually fits.
8. How is this different from measuring video playback quality?
Playback quality (startup time, buffering, resolution) determines whether a video can be watched at all. It's a delivery problem, fixed at the infrastructure level. The metrics in this article assume delivery is already solid and instead measure what happens once someone is watching: whether they press play, how far they get, whether they act, and whether that action reaches your CRM.
If your conversion numbers are weak and you haven't ruled out buffering or slow load times as the cause, start there before working through the checklist above.
The Bottom Line
Nine metrics predict conversion, and they run in order from leading indicator to lagging outcome: play rate, engagement rate, retention, watch depth, re-watches, CTA click-through, conversion rate, video-qualified leads, and influenced pipeline.
Everything else on a typical dashboard, views, impressions, likes, and blended averages, is context. It is not proof.
The single highest-leverage fix available to most teams right now is connecting the video platform to the CRM. Less than half of marketers have done this, which means the majority of watch-depth data being collected today never reaches a contact record and never triggers a follow-up.
Set a watch-depth threshold, wire it to your CRM, and report the influenced pipeline instead of raw engagement.
Start by measuring video performance on the handful of pages where it already matters most: pick one video, set a watch-depth line, and see what shows up on the contact record once you connect it.
If you want to try this on Gumlet's free plan first, the event streaming is included from the Growth tier up, and you can see exactly which viewers are crossing your threshold before deciding whether to build out a fuller attribution stack around it.
