The VSL Sales Process: A Stage-by-Stage Framework for Turning Viewers Into Buyers

You've probably watched a video sales letter that had you reaching for your credit card before you even realized what was happening. That's not magic. That's a well-executed sales process at work.

If you already understand the basics of marketing and copywriting, you know that converting viewers into buyers doesn't happen by accident. Every compelling VSL follows a deliberate, stage-by-stage structure designed to move people from curious onlookers to confident customers. The problem is, most tutorials out there either oversimplify the process or drown you in theory without showing you how the pieces actually connect.

This guide changes that. We're going to walk through the complete VSL sales process from the opening hook all the way to the final call to action. You'll learn what each stage needs to accomplish, why the order matters, and how to craft transitions that keep viewers locked in rather than clicking away.

Whether you're building your first VSL or fine-tuning one that isn't converting the way you want, this framework gives you a practical roadmap to follow. Let's get into it.

Why Your Sales Process Lives Inside Your Video

Most paid traffic operators draw their "sales process" as a flowchart: ad creative, landing page, VSL, checkout. Clean. Logical. Wrong framing.

That architecture treats the VSL as one step among equals, when it's actually doing all of the work. Your ad handles awareness. Your landing page handles interest. Your order form captures action from a viewer who's already been persuaded. But the persuasion itself? That happens entirely inside the video. The funnel is delivery infrastructure. The VSL is the sales engine.

This distinction has real dollar implications. Pages with embedded video convert at significantly higher rates than text-only equivalents. An Unbounce analysis of 44,000 landing pages found VSL-led pages averaged a 12.7% conversion rate versus 4.8% for text-only pages in equivalent categories, a 164% relative lift. That gap doesn't come from the landing page design. It comes from what happens inside the video.

Here's the problem with where most operators focus their optimization energy: ad creative gets the attention because it has shorter feedback loops and platform-native testing tools. But the ad just gets the click. According to the State of VSL Marketing in 2026, the operators hitting 4-5% conversion on cold traffic are differentiated by VSL quality and data infrastructure, not offer differentiation.

Video is also no longer a differentiator by itself. It's baseline. As the data shows, the gap between strategically architected VSLs and generic template-driven ones has never been wider. Everyone is running video. The edge goes to whoever operates it most intelligently.

The practical implication is direct: optimizing your sales process means diagnosing what happens second by second inside your VSL. Not shuffling page layouts. Not tweaking button colors. Average watch time tells you almost nothing because it smooths over the exact moments where your script loses buyers. If viewers are dropping at 4:12 because your story stalls, or bailing at the price reveal because the value stack hasn't landed, standard funnel analytics won't surface that. You need second-by-second visibility to find it.

The 5 Stages of the VSL Sales Process

Every high-converting VSL runs through the same five stages regardless of what you're selling: Hook, Problem, Solution, Offer, and CTA. Before you can optimize any single stage, you need to see them as distinct, measurable segments of your sales process, each with its own engagement signature and its own failure modes.

Stage 1: The Hook (0 to 30 Seconds)

This is the highest-leverage window in your entire funnel. Viewers decide within the first 30 seconds whether they're staying or scrolling, which makes hook retention your single most important metric, full stop. A strong hook doesn't need to be flashy; it needs to be immediately relevant to the specific pain or desire your cold audience is already carrying. Problem-first hooks consistently outperform claim-first hooks on cold paid traffic, which makes sense when you think about who's actually watching your ad. They don't know you yet, but they know their problem. Speak to that first. Learn how to write a high-converting VSL from scratch and you'll see that hook construction is where most writers should spend the majority of their revision time.

Stage 2: The Problem (30 Seconds to ~25% of Runtime)

Once you've earned the next few minutes of attention, your job is to agitate the pain your viewer already feels, not introduce a new one. This section works when it mirrors the viewer's internal monologue so closely that they feel like you wrote the script specifically for them. If your engagement curve drops sharply during this window, the diagnosis is usually one of two things: your problem framing doesn't match the lived experience of the cold audience you're paying to reach, or you're moving too fast without letting the emotional weight land. Slow down, get specific, and use language your buyers actually use, not polished marketing copy.

Stage 3: The Solution (Roughly 25% to 60% of Runtime)

This is where you introduce your mechanism and build the credibility that earns the price reveal later. A healthy engagement curve in this section looks like a gradual, controlled decline, not a cliff. If you see a sharp drop here, you likely have a mechanism comprehension problem; your viewer doesn't understand how your solution actually works or doesn't believe it applies to their situation. Rewinds in this section are a strong positive signal. They mean viewers are trying to understand something more deeply, which is exactly the cognitive engagement that precedes a buying decision.

Stage 4: The Offer (Roughly 60% to 80% of Runtime)

Price reveals and offer stacking produce the sharpest natural drop-offs in any VSL, and that's expected. What you're tracking here is the exact percentage who leave at the moment you name your price. That number tells you whether your price anchor was set correctly during the solution section and whether your value stack landed before cost entered the conversation. If drop-off spikes the second the price appears, the fix is usually upstream, not in the offer itself.

Stage 5: The CTA (Final 20% of Runtime)

Viewers who reach your CTA convert at dramatically higher rates than your blended funnel average suggests. The problem is that most operators never see this number clearly because they're looking at page-level conversion rates, not watch-depth-segmented purchase data. Connecting watch depth to actual purchase events is the only way to surface your true CTA conversion rate. That's a completely different metric than your overall funnel conversion rate, and the gap between the two tells you exactly how much money you're leaving in the middle of the video.

Format: Keep It Inside the 8 to 20 Minute Window

For cold social traffic in 2026, the 8 to 20 minute VSL format has emerged as the clear performance sweet spot. Creating video sales letters that actually convert in 2026 means respecting the attention budget of a cold viewer who has never heard of you. The old 60-minute marathon format is actively losing conversion ground on cold audiences; that length now belongs to presold or highly sophisticated buyers who are already deep in your funnel. Time your five stages proportionally inside that 20-minute ceiling, and you'll have a script architecture that matches how cold traffic actually behaves, not how you wish it behaved.

Why Standard Analytics Lie to You About Your Sales Process

Let's talk about the metrics most marketers trust without questioning, and why that trust is costing you real money.

Average watch time is the most dangerous number in your analytics dashboard. It sounds useful. It feels like signal. It is not. That single averaged figure tells you how long viewers watched across your entire audience, which hides the exact seconds where your script is hemorrhaging buyers at scale. The problem is not that the metric is inaccurate. The problem is that it is accurate in the most useless way possible.

Here is a concrete example. A VSL reporting 40% average watch time could have two completely different failure modes underneath that number. Scenario one: 60% of viewers bail in the first 30 seconds because your hook does not grab them. Scenario two: viewers watch all the way through, but 60% leave at the price reveal because your offer does not land. Both scenarios produce the same average watch time. Both require completely different fixes. A broken hook means you rewrite the opening and test new lead angles. A broken offer means you restructure your pricing, bonuses, or guarantee. If you optimize the hook when the offer is the problem, you waste weeks and budget chasing the wrong variable entirely. Second-by-second engagement data, specifically engagement heatmaps that show drop-off, rewind, and exit by timestamp, is the only way to tell these scenarios apart.

The conversion tracking side of this is just as broken. Ad blockers and iOS privacy restrictions are not edge cases affecting a small slice of your audience. Research puts ad blocker usage between 27% and 40% of internet users globally, with US desktop users running between 27% and 33%. iOS App Tracking Transparency has reduced opt-in rates to roughly 35% globally, meaning the majority of your iPhone traffic is already partially invisible to your pixel before a single ad runs. The practical result: your pixel-reported ROAS is built on an incomplete picture of what your funnel actually produced. Your ad platform might report 60 conversions while your checkout records 100. That 40-conversion gap is not a glitch. It is structural signal loss.

The downstream damage compounds fast. You look at your ad account and see an ad with a weak reported ROAS. You pause it. Meanwhile, that ad was actually converting at a strong rate, but the pixel missed those conversions because buyers were on iOS or running an ad blocker. You scale a different ad that appears to be your top performer, not realizing it only looks stronger because its audience happened to include fewer ad blocker users. Teams that fix their attribution often cut customer acquisition costs by 20% to 40% simply by seeing what was already working.

Server-side pixel forwarding removes the browser from the data path entirely. Instead of relying on a browser script to fire a conversion event (a script an ad blocker can intercept or iOS can restrict), conversion data goes directly from the server to Meta and Google via their APIs. Ad blockers cannot touch it. iOS restrictions cannot intercept it. Your attribution reflects what actually happened in your funnel, not what a browser was permitted to report. Server-side tracking recovers between 15% and 40% of conversions that browser pixels miss, with near-complete recovery on the iOS and Safari blind spots specifically.

Here is the bottom line. Second-by-second engagement data tells you exactly where your script is losing buyers and why. Complete server-side conversion tracking tells you which traffic sources and ads are actually driving revenue. Without both together, you are running split tests, killing ads, and rewriting scripts based on a partial view of reality. Real sales process optimization requires complete data at both the viewer level and the conversion level, not just a number that averages everything into something comfortable to look at.

How to Read Heatmap Data at Each Script Stage

Think of your engagement curve as a visual fingerprint of your script's persuasion quality. A healthy VSL produces a controlled, gradual slope downward from left to right. Some drop-off at every stage is completely normal; viewers exit for a hundred reasons outside your control. What you're looking for are two specific failure patterns. A flat line usually means autoplay is doing the work, not genuine viewer intent. Turn autoplay off and watch that number drop to its real baseline. A cliff, on the other hand, means a specific moment in your script is triggering mass abandonment. That's actually good news: it means the problem is locatable and fixable.

Hook Stage: 0 to 30 Seconds

This is the highest-leverage zone in your entire funnel. If you're losing more than 30 to 35% of viewers before the 30-second mark, stop. Don't touch your offer, don't renegotiate your price, don't swap your ad creative. Your hook is broken. The problem lives in the first 15 to 30 seconds of playback, and everything downstream is irrelevant until that gets fixed. Cold traffic viewers are making a stay-or-leave decision within the first few seconds of your video. A drop-off spike at second 22 won't show up in your bounce rate; it disappears into an aggregate. Second-by-second data is the only way to see it.

Problem Stage: 30 Seconds to ~25% of Runtime

If your hook holds and you still see a sharp drop in this window, the issue is usually a disconnect between what your ad promised and what your VSL actually opens with. Your ad attracted a specific audience segment with a specific pain point, and your problem framing isn't matching it closely enough. The opening problem statement feels too broad, too generic, or too off-target for the person who clicked. This is a message-to-market alignment problem, not a video production problem. The fix is to audit the exact language in your winning ad and mirror that framing in the first minute of your VSL.

Rewind Clusters: The Signal You Don't Want to Ignore

Rewind behavior is one of the most underrated data points in VSL optimization. When viewers are replaying a segment repeatedly, that moment is landing. They want to hear it again. Something in that language resonated, triggered an objection they needed to process, or made a claim compelling enough to revisit. Your job is to take that exact language and pull it forward. Use it earlier in the script to accelerate engagement. Test it in your ad creative. Rewind clusters tell you where your script has real persuasive weight.

Offer Stage: 60 to 80% of Runtime

Every VSL takes a hit at the price reveal. That's expected. The question is whether the drop is normal attrition or a catastrophic collapse. If you're losing more than 50% of your remaining viewers at the exact second you name the number, the value stack you built upstream wasn't sufficient. The fix isn't a lower price. The fix is in your solution section and credibility sections, which should have been building perceived value aggressively before the price ever appeared on screen. The price reveal is just a mirror; it reflects the persuasion quality of everything that came before it.

CTA Stage: Post-80% of Runtime

High watch depth combined with low conversion rates is a specific and solvable problem. If viewers are making it through 80%+ of your video and still not converting, the issue is almost certainly what happens after the video, not inside it. Weak CTA language, friction at checkout, a mismatch between what the VSL promises and what the order page delivers, or an unclear next step can all kill conversion rates that your video actually earned. The conversion optimization diagnostic framework at VSLStats outlines exactly this failure pattern, where improving conversion rate by even 0.3 percentage points delivers more revenue impact than a 13% increase in raw traffic volume.

None of these diagnostics are possible without second-by-second engagement data. VSLStats surfaces this layer of visibility specifically for direct-response video, not general-purpose viewing analytics built for content publishers. The heatmap doesn't just show you what happened; it tells you where in your script to apply the fix.

Mobile-First Sales Process Realities You Can't Ignore

More than 70% of VSL views now happen on mobile phones. If your VSL was scripted, tested, and refined on a desktop browser, you are actively optimizing for fewer than 30% of the people actually watching it. That is not a minor gap in your process. That is a structural mismatch between where your audience is and where your attention is going.

Mobile viewers behave completely differently from desktop viewers. They are watching in fragmented attention windows, often between tasks or during commute time. Their screen is smaller, their connection is less stable, and at any given moment they are competing with push notifications, DMs, and the same thumb-scroll habit your short-form ad just interrupted. More importantly, a large portion of them are watching with sound off. This is not an edge case. It is the default mobile viewing condition across social platforms, and your entire VSL persuasion sequence was probably engineered to be heard, not read.

That last point is where most VSL funnels bleed revenue without knowing it. Your hook relies on voice tone. Your problem agitation relies on emotional audio delivery. Your offer reveal relies on pacing and inflection. When sound is off, none of that lands. The viewer sees a talking head or a slide, registers nothing that grabs them in the first 5 to 8 seconds, and scrolls away without your script ever making a case.

AI-generated captions are not a finishing touch. They are a conversion variable. When your script surfaces as on-screen text in real time, a muted mobile viewer can follow every stage of your sales argument: the hook, the problem, the offer, the close. Removing that audio dependency is what turns a mobile view from a wasted impression into an active prospect. In 2026, running a VSL without captions on mobile traffic is roughly equivalent to running ads without copy.

Player load speed and rendering also matter before your script gets a single chance. A VSL that buffers, renders at the wrong aspect ratio, or takes more than two seconds to load on a mobile connection creates friction that drops viewers before the first word. Video sales letter funnel architecture covers how the player experience itself functions as part of your conversion system, not just a container for your content.

The practical move: pull your heatmap data and segment it by device type. Desktop and mobile engagement curves are almost never identical. You will likely find your desktop viewers dropping at one stage and your mobile viewers dropping somewhere else entirely, which means you need different diagnostic questions and different script fixes for each segment. One blended average watch time metric will hide both problems at once.

Testing the VSL Sales Process Without Guessing

Page-level split testing feels productive. You change a headline, run traffic, wait for a winner. But here's the problem: your landing page headline is not your sales process. The VSL is. When you test headline A against headline B and measure conversions, every variable inside that 12-minute video collapses into a single number. You have no idea whether the lift came from the headline, the hook, the offer frame, or just a traffic quality fluctuation. That is not a test. That is a coin flip with extra steps.

The correct testing unit is a single script stage in isolation. Hook A versus hook B. Offer frame A versus offer frame B. CTA delivery A versus CTA delivery B. One variable, one stage, one test. This is the only methodology that produces actionable intelligence about what is actually wrong with your script. Anything broader than that and you are back to guessing.

Start With the Hook, Every Time

The hook window is 0:00 to 0:30, and it has more leverage than any other stage of your VSL. If your hook bleeds 40% of viewers before your problem statement even lands, it does not matter how good your offer is. There is no audience left to receive it. Every optimization downstream of a broken hook is wasted effort. Check your engagement heatmap at the 30-second mark before you touch anything else. That single data point tells you whether hook testing should be your first priority.

After the hook holds, move to the offer frame, specifically the section before your price reveal. The order in which you stack value and the language you use to anchor price are variables with disproportionate conversion impact. Small script changes at this stage routinely produce double-digit conversion lifts without changing traffic, creative, or the core offer itself.

The Economics of Rapid Testing Have Changed

In 2026, the financial barrier that previously made re-shooting VSL variants impractical has largely disappeared. AI tools have compressed the production and iteration cycle from weeks to days, and median video production costs have dropped roughly 40%, from approximately $4,200 to $2,500 per finished minute. You can now run three hook variants in the time it used to take to finalize one script. The economic case for high-velocity script-level testing has never been stronger.

But speed only matters if your infrastructure can support it. Serving two video variants to split traffic and attributing conversions separately to each requires a player built for that purpose. This is a fundamentally different capability than running a standard A/B test on a landing page. You need separate conversion attribution per variant, not just a different version of a page element. Without that, you still cannot isolate which video drove the result.

Authenticity Is a Testable Variable Now

One more thing worth testing: your urgency framing. Fake scarcity and manipulative countdown timers are actively hurting conversion rates in 2026 as audiences have grown more sophisticated at recognizing them. This is not a philosophical argument; it is a measurable conversion variable. Test genuine urgency framing against manufactured scarcity and let your engagement data resolve it. Your heatmap will show you exactly where trust breaks down in the script. Stop defending tactics that are costing you conversions and start treating authenticity as something you can measure and optimize.

VSLStats has A/B split testing built directly into the player, so you can serve variants, split traffic, and tie conversion attribution back to each video without stitching together separate tools. If you are ready to stop guessing and start testing at the script level, try any plan for $1 at /pricing.

Connecting Watch Depth to Revenue: The Attribution Layer Most Funnels Are Missing

Here's the attribution reality most funnel operators never confront: your video host knows who watched and for how long, and your payment processor knows who bought and for how much. But those two systems have never talked to each other. So you're left with a funnel-level conversion rate that tells you something is working, but nothing about why, where in the video, or which viewers are actually driving revenue.

Watch depth and purchase probability move together in every VSL funnel. Viewers who make it to your CTA convert at materially higher rates than viewers who bailed during the problem stage. That's not surprising. What is surprising is how few operators can actually prove this for their specific funnel, because without a system that ties purchase events to watch depth at the individual session level, you're inferring rather than measuring.

What Video-Level Revenue Attribution Actually Tells You

Revenue attribution at the video level means connecting three data points for every buyer: which video they watched, which viewer session it was, and exactly how far they were into the script when they converted. That's not just confirmation that your VSL "works." It's a precise answer to how much of it a buyer needs to watch before they pull out a credit card.

The practical output of that data is where it gets actionable. If 80% of your buyers watched past the 65% mark before converting, and 70% of your viewers are dropping off before the 50% mark, you have a clear picture. You're losing buyers at scale before they ever reach the section of the script that produces purchases. That's not a traffic problem. That's a script problem with a known address in your video.

Attribution Beyond the First Sale

Lifecycle marketing is where revenue attribution becomes genuinely powerful. Your onboarding sequence probably contains a VSL. Your upsell flow almost certainly does. Reactivation campaigns, referral programs, post-purchase education — these all involve video, and each one has its own sales process with its own conversion dynamics. When you can attribute revenue independently to each of those videos, you stop guessing which part of your lifecycle sequence is earning its keep.

As third-party tracking gets harder and more expensive, operators with tight internal attribution loops have a structural edge. Every iOS update, every ad blocker, every cookie deprecation raises the cost of knowing what's working at the acquisition level. But if you know which video, at what watch depth, produced which revenue outcome, that signal is yours. It compounds over time. It gets more valuable as acquisition costs rise.

VSLStats' revenue attribution links every dollar back to a specific video and viewer session, giving you script-level intelligence instead of funnel-level averages that hide where the real leverage is. That's the difference between knowing your funnel converts at 2% and knowing that buyers who reach the 70% mark convert at 11%. One number lets you report. The other lets you scale with confidence.

Ready to see your own watch-depth revenue data? Try any plan for $1 at VSLStats pricing.

Short-Form Video and the Hook-to-VSL Pipeline

Short-form video is no longer a nice-to-have at the top of your funnel. It is the primary mechanism feeding your VSL pipeline in 2026. Short-form has ranked as the number one ROI content format for three consecutive years, and 57% of marketing budgets now include a dedicated short-form line item. That is not a trend. That is a structural shift in how direct-response traffic gets warmed before it ever hits your sales letter.

The reason this matters mechanically: videos under 60 seconds generate 2.5x more engagement per impression than any other content type. Every viewer who watches your short-form ad to completion and clicks through has already been exposed to a specific hook frame, a specific pain point, a specific emotional trigger. They arrive at your VSL pre-qualified by exactly the angle your ad used. That pre-qualification is an asset, but only if you use it correctly.

Hook Consistency Is Architecture, Not Style

Here is where most operators lose the advantage they just paid to build. Your short-form ad hook and your VSL opening hook need to be architecturally consistent. If your ad opens on "why your sales calls keep stalling at the price objection" and your VSL opens on a general credibility intro, you have broken the frame the viewer bought into. That disconnect shows up immediately in your 30-second retention data. The viewer arrived expecting a continuation of the conversation your ad started. When the VSL opens on a different frame, a meaningful share of that audience drops before your offer ever enters the picture.

The fix is straightforward but requires deliberate mapping. Take your top-performing short-form hooks, measured by watch-through rate and click-through rate, and treat them as the literal script setup for your VSL opening 30 seconds. The ad hook that wins clicks should be the same pain-point frame your VSL expands on in its first half-minute.

Building the Feedback Loop

This pipeline architecture has a direct, measurable payoff in your VSL engagement data. Audiences arriving from a tightly matched ad hook will show stronger 30-second retention than audiences sent from mismatched creative because the narrative continuity holds. They keep watching because the VSL is delivering what the ad promised.

As AI compresses production cycles and cuts video creation costs significantly, the volume of hook variants you can test will increase. The operators who compound fastest will not be the ones testing the most hooks in isolation. They will be the ones routing ad-hook retention data back into VSL opening decisions, treating both channels as one connected system rather than two separate creative workstreams.

Managing the VSL Sales Process Across Multiple Clients

The diagnostic framework you use for your own VSL funnel translates directly to client work. Hook retention, stage-by-stage drop-off, offer engagement, CTA conversion: those four checkpoints apply to every account you manage. What changes is the baseline. A 65% retention rate at 60 seconds might be strong for a cold-traffic supplement offer and mediocre for a warm-list coaching funnel. The methodology is portable. The standards are not. When you onboard a new client, your first job is establishing their specific benchmarks before you start calling anything a problem.

The operational challenge is infrastructure, not analysis. Most agencies hit a wall when they're managing five or six VSL funnels simultaneously. Each client ends up in a different setup, with different pixel configurations, different reporting exports, and no common framework for comparing what's actually happening across accounts. That fragmentation isn't just annoying; it destroys the cross-account pattern recognition that makes an agency worth more than a freelancer who delivers monthly PDFs.

Here's what siloing actually costs you: if you can't see that price-reveal drop-off spikes consistently across your health and wellness clients above a certain price point, you're treating each funnel as an isolated problem instead of recognizing a pattern you could solve once and apply everywhere. That cross-account intelligence is your real competitive edge as an agency. A fragmented stack makes it invisible.

How you report to clients matters as much as what you find. Raw engagement data means nothing to a client who's thinking about their ad spend and conversion rate. "Your 60% drop-off at 4:30 is concerning" lands flat. "Your price reveal is losing 6 in 10 remaining viewers before they see the full value stack, and here's the specific script change we recommend" is a different conversation entirely. That framing makes you a strategic partner, not a metrics vendor.

Server-side pixel accuracy is non-negotiable at the agency level. A 30% tracking blind spot on one account is a problem. That same blind spot compounded across six accounts running Meta and Google traffic means you are making portfolio-level budget decisions on structurally incomplete data. Every client's pixel configuration needs to be set up independently and verified before you scale a dollar.

VSLStats supports agency white-label sub-accounts so you can standardize analytics infrastructure across every client from a single dashboard, present the intelligence under your own brand, and onboard new clients without rebuilding your stack from scratch.

Build a Sales Process You Can Actually Measure

Every tactic in this guide comes down to one question: can you measure what's actually happening inside your video, at the moment it happens?

Start with hook retention at 30 seconds. If you're losing more than 30 to 35% of viewers before your problem statement even begins, stop. Don't touch your offer slide. Don't rewrite your CTA. Every downstream test runs on a self-selected audience that's already smaller than it should be, making your data unreliable before you've even started.

Before you fix your script, fix your tracking. Browser-side pixels can leave a significant blind spot in your conversion data depending on your audience's ad blocker usage and iOS privacy settings. When that happens, you're not optimizing on reality; you're optimizing on a partial signal. Server-side pixel forwarding is the infrastructure fix that makes every other data point trustworthy. Without it, you're building decisions on incomplete evidence.

Segment your heatmap data by device. Mobile and desktop viewers drop off at different moments, respond to different script elements, and convert at different watch depths. Running a single unsegmented heatmap and drawing universal script conclusions is one of the most common diagnostic errors in VSL optimization. Treat them as two separate audiences.

Test in sequence: hook first, then offer frame, then CTA mechanics. Changing multiple elements simultaneously makes it impossible to know what actually moved the needle.

And connect watch depth to revenue. If you can't answer how far a buyer typically watches before converting, you don't have a measurable sales process yet.

VSLStats gives you every layer of this in one player: second-by-second heatmaps, server-side pixel forwarding, AI captions for muted mobile viewers, A/B split testing, and revenue attribution tied to specific videos and watch depths. Try any plan for $1 at /pricing.