Media Budgets: Where the Real Cost Shows Up

A performance budget is a simple idea: define the maximum cost of your page and hold the line. Tim Kadlec’s framing is the one that tends to stick: set a budget, then don’t allow the page to exceed it. That cap can be expressed as load time, but it’s usually easier to negotiate when it’s broken into number of requests or total page weight. Typical yardsticks include overall weight, HTTP request count, load time on a specific mobile network, and Core Web Vitals like FID, FCP, CLS and LCP.

With a documented budget, teams can have productive discussions about content trade-offs — whether a page holds five images, or three images and one video, and still sits inside the agreed limits. But a budget as a standalone metric is weak. Its real value appears only when it’s tied to business outcomes.

Why Engagement Depends on Optimization

The data connecting page speed to revenue is not thin. WPOStats aggregates hundreds of case studies where delays measured in hundreds of milliseconds, or at most a few seconds, correlate with heavy losses in annual sales or conversions. Slow sites hurt conversion directly, which puts online businesses in a bind: how do you keep an audience engaged without blowing the performance budget?

A major part of that answer is optimized visual media. Research cited in the State of Visual Media report indicates the brain can process visual input in under 13 milliseconds; text comprehension takes considerably longer. That gap helps explain why short-form microvideo — usually 10–20 seconds — tends to earn outsized engagement and conversion gains.

Video’s Appeal and Its Real Costs

Consumers increasingly expect more than static images. Survey data from WebRetailer points to a strong preference for video: 96% of consumers say video helps them make purchasing decisions, and 79% would rather watch a video about a product than read text. The right product video can lift conversions by over 80%.

Yet video delivery carries a heavy freight. Scott Jehl has highlighted that average video weight climbs every year, and faster on mobile than desktop. In some cases that’s justified by high-resolution mobile screens, but it’s also a symptom of a format gap: HTML alone offers no simple way to serve contextually appropriate video sizes the way responsive images do. Many large marketing-page videos are hand-placed without an intelligent media server behind them.

The playback stats bear this out. Conviva’s Q4 2020 State of Streaming report found mobile devices suffered 20% more buffering, a 19% higher video-start failure rate, and 5% longer start times than other device classes. Video delivery also inflates bandwidth spend when you can’t or don’t transcode to formats the browser handles best. Skip a CDN or fail to map users to the nearest edge, and video start is slowed further by suboptimal routing.

Images are no better behaved. The HTTP Archive’s Web Almanac data has long shown images as the leading cause of page bloat. More telling, the byte differential served to mobile versus desktop is very small — meaning a lot of bandwidth goes to devices that simply don’t need it. The lesson from all these numbers is the same: rich, unoptimized content quietly works against the business goals it was meant to serve.

Balancing Rich Media With Performance Goals

Rich media can drive engagement, but it must be weighed against delivery costs and business objectives. One common option is hosting video through platforms like YouTube or Vimeo, which saves bandwidth. The trade-off, however, is losing the ability to create a fully customized branded experience or deliver personalized content — and image hosting and delivery still remain your responsibility.

You don't have to offload your content entirely. There are several ways to revamp your media delivery system to stay within performance budgets:

Know What You're Working With

Start by measuring the weight of your webpages and their media assets. Research expert Tammy Everts advises keeping pages under 1 MB for mobile and 2 MB for everything else, and to identify which resources appear on your most critical pages. For instance, could a short video replace a paragraph of text and associated images, and what impact would that have on business goals? Review Real User Monitoring (RUM) and analytics to spot the pages that lead to higher conversion and engagement rates.

Synthetic tracking of Google Core Web Vitals (CWVs) via tools like LightHouse should be part of your toolkit. You can also measure CWVs in the field using Chrome User Experience Report (CrUX) data. Because CWVs factor into Google's crawling signals, it's worth monitoring and optimizing for Largest Contentful Paint (LCP), First Input Delay (FID), and Cumulative Layout Shift (CLS).

Pick the Right Format and Size

Serve images and videos in a format appropriate to the device and browser — in both size and resolution. An image CDN can help here, or you can create variants like WebM, AVIF, JPEG-XL, or HEIC and choose the right content type based on User-Agent and Accept headers. For one-off conversions, tools like Squoosh.app work well. A related practice is converting animated GIFs to video, which can produce significant savings.

Over 41% of images on mobile devices are improperly sized. Rather than serving fixed widths, crop media to fit the container with tools like Lazysizes. AI-powered cropping tools can detect areas of interest and save significant time. For below-the-fold images, leverage native lazy-loading to avoid unnecessary downloads.

Improve Video Experience

Since nearly 85% of videos are played without sound, adding subtitles serves both accessibility and engagement purposes. Though transcription can be tedious, AI-based transcription services can automate much of the workflow.

Delivering through multiple CDNs can reduce last-mile latency, shorten video start times, and mitigate buffering. A multi-CDN strategy also offers resilience against localized outages at edge nodes, according to research from Citrix. Tools like Cloudinary's Media Optimizer take a unified approach, optimizing format and quality while serving content through multi-CDN edge nodes — prioritizing both visual fidelity and small file sizes.

Progressive Rendering Strategies

Auto-playing preview videos on platforms like YouTube have shown to increase watch time by over 90%. However, autoplay comes with accessibility drawbacks, so it's important to use it judiciously and always provide a pause option. A better approach for page-size budgets is to serve AI-generated video previews and poster images only, loading the full video only when the user clicks play. This eliminates unnecessary downloads and accelerates initial page loads.

Another strategy is to load a preview video first and have the player check the device's connection type via the Network Connection API. If connectivity is adequate, the player can seamlessly swap the preview source with the full video. Since CDNs can detect network conditions more reliably, production-quality code can let the CDN determine network speed and instruct client code to progressively load long-form content.

"A performance budget doesn't guide your decisions about what content should be displayed. Rather, it's about how you choose to display that content. Removing important content altogether to decrease the weight of a page is not a performance strategy."

— Tim Kadlec

This distinction matters: visual media will continue to dominate storytelling on the web, but whether it serves your purposes depends on aligning content choices with both business strategy and performance constraints.