Modal Organic Growth Opportunities

Create a standardized, SEO-optimized landing page for every major open-source model and variant, providing deployment recipes and hardware requirements.

Example Keywords

• deploy llama 3.1
• serve mistral api
• {model} inference endpoint
• self host {model}
• {model} requirements GPU memory

Rationale

This play extends Modal's existing success in technical implementation content into a programmatic directory, capturing users looking for specific deployment instructions for the latest models.

Topical Authority

Modal already earns meaningful organic traffic from deeply technical AI infrastructure content and GPU/infra explainers, signaling strong authority in implementation-focused queries.

Internal Data Sources

Use Modal docs, code examples from the sitemap, pricing primitives, and internal benchmark snippets to provide differentiated, copy-pasteable code.

Estimated Number of Pages

20,000+ (Covering model families, sizes, and intent variants like deploy/serve/self-host)

Develop programmatic cost-to-run and specification pages for every accelerator Modal users consider, including pairwise comparisons and workload-specific recommendations.

Example Keywords

• {GPU} price per hour
• {GPU} cloud cost
• {GPU} vs {GPU}
• {GPU} VRAM bandwidth
• best GPU for {workload}

Rationale

This targets the long-tail of GPU economics and hardware education that Modal's current blog posts do not yet fully cover, capturing users in the infrastructure selection phase.

Topical Authority

Modal's organic winners already include GPU pricing-oriented blog pages and a GPU glossary, proving Google associates the domain with hardware economics.

Internal Data Sources

Leverage Modal pricing model details, internal recommended instance guidance, and anonymized utilization patterns to provide unique cost-per-workload insights.

Estimated Number of Pages

8,000+ (Covering all non-covered chips, cost/spec pages, and pairwise comparisons)

Build static troubleshooting pages for exact error strings and symptom-based issues, providing causes, reproduction checklists, and Modal-specific fixes.

Example Keywords

• {exact error message}
• fix {error} pytorch
• NCCL timeout fix
• CUDA illegal memory access fix
• device-side assert triggered fix

Rationale

These queries are extremely high intent as they occur when an engineer is blocked; providing a fix alongside a 'run it cleanly on Modal' snippet drives high-quality conversions.

Topical Authority

Modal's GPU glossary already ranks for low-level GPU concepts, providing the necessary topical foundation to expand into technical troubleshooting.

Internal Data Sources

Utilize internal support runbooks, container base image recommendations, and known-good dependency pins to offer authoritative solutions.

Estimated Number of Pages

3,000+ (Covering error strings, frameworks, and OS/container variants)

Generate SEO-facing recipe pages that map to common production workloads, providing reference architectures, code templates, and operational checklists.

Example Keywords

• how to deploy {framework}
• {framework} autoscaling inference
• batch inference {framework}
• run {pipeline} in production

Rationale

This play transforms Modal's deep internal expertise and existing documentation into search-optimized landing pages that match specific production intent.

Topical Authority

The sitemap contains hundreds of docs examples, indicating strong internal expertise that can be leveraged to own implementation-related search queries.

Internal Data Sources

Incorporate docs examples, product capability statements (Batch, Inference, Training), and internal best practices for scaling and cold-start mitigation.

Estimated Number of Pages

5,000+ (Covering recipe archetypes, frameworks, and deployment modes)

Create static benchmark report pages that answer throughput and cost questions for specific model and GPU combinations at various quantization levels.

Example Keywords

• {model} benchmark
• {model} latency
• {model} tokens per second
• cost per 1M tokens {model}

Rationale

This captures users in the decision-stage of infrastructure evaluation, closing the traffic yield gap by providing the specific data points needed for procurement.

Topical Authority

Modal is already rewarded by Google for performance-oriented content; benchmarks are a natural authority expansion that maps directly to the product's value.

Internal Data Sources

Use internal benchmarking harnesses, Modal runtime characteristics (cold starts, concurrency), and pricing formulas to generate accurate cost-per-output data.

Estimated Number of Pages

15,000+ (Covering models, GPUs, quantization levels, and context-length tiers)

Improvements Summary

Rework the NVIDIA GPU pricing articles to answer “how much does X cost?” faster, then support the answer with clear tables, FAQs, and citations. Add a new GPU pricing hub page plus a tighter internal-linking path that routes commercial researchers from pricing posts to /pricing.

Improvements Details

For /blog/nvidia-b200-pricing, expand coverage for "dgx b200 price", "nvidia dgx b200 price", and "gb200 price" with an above-the-fold price-range block, a sticky table of contents, and separate tables for GPU vs DGX system vs cloud hourly pricing. Add page-specific FAQ sections (with FAQPage schema), a “methodology + sources” block, a visible “last updated” date, and a lightweight changelog; repeat similar structure for H100/H200/L40S, including sections like "H100 hourly price" and "H200 hourly price in cloud". Publish /blog/gpu-pricing-guide as a cluster hub that links to all pricing pages, cross-link pricing pages to each other, and build an internal linking ladder from /blog/serverless-gpu-article and relevant docs to the hub and B200 page with rotated anchors.

Improvements Rationale

Most demand concentrates around the B200 page, and DGX-intent long tails (for example, "dgx b200 price") have far lower competing result counts than generic terms like "b200 price", making page-1 movement more realistic with tighter intent matching. Pricing queries have high CPC and require current, citation-backed content, so “last updated,” sources, and quarterly refreshes help relevance and trust while improving snippet capture via tables and short FAQs. A hub-and-spoke structure plus stronger internal links should lift the whole cluster and route qualified visitors to /pricing for higher conversion intent.