6 Ways to Reduce Your Video Encoding and Delivery Costs by Up to 60%

Most teams trying to reduce their video costs focus on encoding — comparing per-minute rates, negotiating volume discounts, optimizing encoder presets. This makes sense on the surface, because encoding is the most visible cost. But it is also the wrong place to start.

Encoding typically accounts for 10-20% of your total video infrastructure cost. The rest is CDN delivery (40-60%), storage (10-20%), and origin egress (5-15%). If you cut your encoding bill in half but do not touch the other 80%, you have saved 5-10% of your total spend. That is not nothing, but it is not the 50-60% reduction that is actually achievable.

The six strategies in this article target the full cost stack — encoding, storage, and delivery together. Applied in combination, they can reduce total video costs by 50-65%.

#Where your video money actually goes

Before optimizing, you need to understand your cost distribution. Here is a representative breakdown for a mid-scale video platform serving 10,000 hours of content per month at 1080p:

CDN delivery dominates because every viewer who watches a video transfers data from your CDN. A video that is watched 1,000 times generates 1,000x its file size in CDN traffic. Encoding, by contrast, happens once per video regardless of how many times it is viewed.

This means the most effective cost lever is reducing file size. Every byte you shave off your encoded files saves money on storage (once) and on CDN delivery (multiplied by every view). The six strategies below all reduce file size, encoding waste, or both.

#Strategy 1: Use AV1 for modern audiences

The single most impactful change you can make is switching from H.264 to AV1 for viewers on modern devices. AV1 produces files that are 50-56% smaller than H.264 at the same visual quality.

The encoding cost for AV1 is higher — EUR 0.04/min vs EUR 0.01/min for H.264 at 1080p — but encoding happens once. CDN delivery happens with every view. For any content that averages more than a few dozen views, the CDN savings from smaller AV1 files vastly outweigh the encoding premium.

The practical approach is multi-codec adaptive streaming: encode in both H.264 and AV1, and let the player serve AV1 to capable devices (which now include the majority of devices sold since 2021) while falling back to H.264 for older hardware. See our AV1 guide for the full details on device support and implementation.

#Strategy 2: Enable per-title encoding

Standard encoding uses the same bitrate targets for every video, regardless of content complexity. A talking-head interview and a high-motion sports highlight get the same bitrate. This is wasteful: the interview needs far less bitrate to look good, and the sports highlight may need more.

Per-title encoding (also called content-adaptive encoding) analyses each video’s complexity and assigns bitrate targets accordingly. Simple content gets lower bitrates without quality loss; complex content gets higher bitrates to maintain quality.

The typical result is a 20-30% reduction in average file size with no change in perceived quality. Some content types see even larger savings:

Per-title encoding works on top of your codec choice. You get the savings from AV1 and the savings from per-title independently — they compound.

For a detailed explanation of how per-title encoding works and when to use it, see Per-Title Encoding Explained.

#Strategy 3: Right-size your quality tier

Not all content needs the same visual quality. The difference between VMAF 93 (Standard) and VMAF 97 (Premium) is measurable by automated tools but nearly invisible to human viewers on most devices.

Consider the viewing context:

• Mobile (5-6” screen, often outdoors): VMAF 85-90 (Economy) is indistinguishable from higher tiers. Viewers cannot perceive quality differences on a small screen in variable lighting conditions.
• Tablet/laptop (10-15” screen): VMAF 93-95 (Standard) is the sweet spot. Viewers can detect major artifacts but not the subtle differences between Standard and Premium.
• Large TV (55”+, controlled lighting): VMAF 97-99 (Premium) is justified for premium content where viewers are specifically paying for quality.

The cost difference is significant:

Switching from Premium to Standard for the bulk of your content saves roughly 40% on file size and 50% on encoding cost. For user-generated content, social video, or any content primarily consumed on mobile devices, Economy tier is the rational choice.

The key insight: quality tier is not a one-size-fits-all decision. You can use different tiers for different content categories. Premium for your flagship originals, Standard for your general catalog, Economy for user uploads and previews.

For more on VMAF scores and what they mean in practice, see VMAF Explained.

#Strategy 4: Optimize your ABR ladder

An adaptive bitrate (ABR) ladder defines the set of resolution/bitrate combinations available to the player. A typical ladder might look like this:

# Traditional fixed ladder (8 rungs)
2160p @ 15.0 Mbps
2160p @ 10.0 Mbps
1080p @  6.0 Mbps
1080p @  4.5 Mbps
 720p @  3.0 Mbps
 720p @  2.0 Mbps
 480p @  1.5 Mbps
 360p @  0.8 Mbps

Each rung generates a complete rendition — a full set of encoded segments stored on your origin. Eight rungs means eight times the storage cost and eight encoding passes. But not all rungs are created equal.

Diminishing returns. The difference between 720p at 2.0 Mbps and 720p at 3.0 Mbps is barely visible on most devices. The lower rung is only useful for viewers on severely constrained networks — a small percentage of your audience.

Unused top rungs. A 2160p@15 Mbps rung is only served to viewers with both a 4K display and a fast network connection. If that is 2% of your audience, the storage and encoding cost of that rung is disproportionate to its value.

Auto-ABR solves this by analysing your content and selecting the optimal set of rungs based on content complexity, target devices, and network distribution data. A well-optimized ladder typically has 4-6 rungs instead of 8, removing the redundant and rarely-used renditions.

The savings are twofold: fewer renditions to encode (less encoding cost) and fewer renditions to store (less storage cost). A reduction from 8 to 5 rungs saves 37.5% on both encoding and storage.

#Strategy 5: Use CMAF instead of separate HLS and DASH

If you need to support both Apple devices (HLS) and the broader ecosystem (DASH), the traditional approach is to encode your content twice — once in MPEG-TS segments for HLS and once in fragmented MP4 segments for DASH. This doubles your encoding time and storage.

CMAF (Common Media Application Format) eliminates this duplication. With CMAF, you encode once into fragmented MP4 (fMP4) segments and generate both an HLS manifest (.m3u8) and a DASH manifest (.mpd) that reference the same segment files.

The savings are essentially 50% on encoding time and 50% on segment storage. The only trade-off is that CMAF requires HLS with fMP4 support, which has been available since iOS 10 / Safari 10 (2016). Unless you are targeting devices from before 2016, CMAF is strictly better.

For the full technical details on how CMAF works, see CMAF Explained.

#Strategy 6: Do not encode what nobody watches

This is the most overlooked optimization. Most video platforms follow a power-law distribution: 20% of content drives 80% of views. The remaining 80% of content — the long tail — may get viewed rarely or never.

If you encode every uploaded video into a full ABR ladder with multiple codecs immediately upon upload, you are spending encoding budget on content that may never be watched. Three approaches to address this:

On-demand encoding. Only encode content when it is first requested. Store the source file and trigger encoding when a viewer hits play. The first viewer waits for encoding (or sees a lower-quality preview), but you never pay to encode content that nobody watches.

Tiered encoding. Encode a single rendition (720p H.264, Standard quality) immediately for all content. When a video crosses a view threshold (e.g., 50 views), trigger the full multi-codec, multi-resolution encode. This ensures popular content gets the premium treatment while long-tail content stays cheap.

TTL-based re-evaluation. If content has not been viewed in 90 days, delete the encoded renditions and keep only the source file. Re-encode on demand if it is requested again. Storage is not free, and storing encoded renditions of unwatched content is pure waste.

The savings from this strategy depend heavily on your content distribution. For platforms with large catalogs and long tails (educational content, user-generated video, archives), the savings can be 30-50% of total encoding and storage costs.

#Real math: before and after

Let’s put all six strategies together for a concrete example. Assume a platform with:

• 10,000 hours of new content per month
• Average of 100 views per video
• Currently encoding H.264 only, fixed 8-rung ladder, Premium quality, separate HLS + DASH

#Before optimization

#After optimization (all 6 strategies applied)

The exact numbers depend on your specific content mix, audience distribution, and CDN pricing. But the order of magnitude is realistic: teams that apply these strategies systematically see 50-65% total cost reductions, with some achieving over 70%.

The strategies compound. AV1 saves 50% on file size. Per-title saves another 25% on top of that. Right-sized quality saves another 40%. Applied together, the multiplicative effect is far greater than any single optimization.

#Hidden costs checklist

Beyond the six strategies, watch for hidden costs that vary significantly between providers:

These hidden costs can add 15-30% to your nominal per-minute encoding rate. When comparing providers, always calculate the total cost of a complete workflow — from source ingestion through encoding to output delivery — not just the headline per-minute rate.

For a detailed comparison of how major providers price their services, see Video Transcoding Pricing Compared.

#The compound effect

The power of these strategies is in their combination. Each one delivers meaningful savings independently, but applied together, the effects multiply:

1. AV1 reduces file size by 50% for modern devices
2. Per-title encoding reduces the remaining bitrate by another 25%
3. Standard quality tier instead of Premium reduces bitrate by another 40%
4. Optimized ABR ladder eliminates 30-40% of redundant renditions
5. CMAF cuts encoding and segment storage in half
6. Tiered encoding avoids encoding the long tail entirely

Start with the strategies that have the lowest implementation effort and highest impact for your specific situation. For most teams, the order is:

1. CMAF — often a single configuration change, immediate 50% storage reduction
2. Right-size quality tiers — change one parameter, save 40% on bitrate
3. AV1 multi-codec — add one codec, save 50% on CDN for modern viewers
4. Per-title encoding — enable per-job, 20-30% bitrate reduction
5. Optimize ABR ladder — remove redundant rungs, less encoding and storage
6. Tiered encoding — requires workflow changes, but large savings for long-tail catalogs

You do not need to implement all six at once. Each one delivers value independently, and you can measure the impact before moving to the next.

Transcodely supports all six strategies through its API — quality tiers, per-title encoding, auto-ABR, CMAF output, and multi-codec encoding are all available as job configuration options. You can explore the pricing calculator to model the cost impact for your specific workload, or read more about how AV1 fits into the picture.