The marketing tech stack we actually run across 11 brands in 2026

The marketing stack we run across 11 brands is smaller than most three-person startups’ stacks. We use roughly a dozen tools, total monthly software cost stays well under what a single enterprise “growth platform” charges, and the most valuable layer — content and GEO — runs on a static site generator and a handful of scripts, not a six-figure suite. This is the actual list, what each tool does, and where we deliberately spend nothing.

We’ve written about the AI content engine and AI agents for marketing teams that sit inside this stack. This post is the full inventory around them.

The principle: own the layer that compounds, rent the rest

Before the tool list, the rule that produces it: own the assets that compound, rent the commodities.

Your website, your content, and your owned audience compound — every post and every subscriber is worth more next year. So we own those outright: the sites are static Astro builds we control end to end, hosted on Cloudflare for a few dollars a month, with no platform that can raise rent or deprecate us. Everything that’s a commodity — analytics collection, email sending, rank tracking — we rent, and we switch the instant something cheaper and equal appears.

This inverts how most teams buy. They rent their most important asset (a website locked into a $2,000/month “experience platform”) and then bolt on cheap point tools around it. We do the opposite: spend the engineering effort once on the compounding core, and keep everything peripheral swappable and cheap.

Across 11 brands, total marketing software cost stays in the low-four-figures per month — less than a single seat of many “all-in-one” growth platforms — because the compounding layer (sites + content) runs on a static stack we own, not rented SaaS.

Layer 1: the owned core (sites, content, hosting)

This is where we spend engineering, not subscription dollars.

• Astro for every site — static-first, fast, crawlable, which is the technical floor for both SEO and GEO. AI engines can’t cite a page their crawler can’t render.
• Cloudflare Workers Assets for hosting — pennies-to-a-few-dollars per site per month, with security headers and a contact API handled in one small Worker.
• Markdown content collections — posts are flat files in git, not rows in a CMS database. Diffable, portable, no vendor lock-in.
• Resend for transactional and contact-form email — usage-priced, cheap at our volume.

The whole point of this layer is that it has no per-seat pricing and nothing that can be taken away. The build pipeline (OG-image generation, broken-link checks, llms.txt generation, monthly date refresh) runs as zero-dependency Node scripts in CI. Cloudflare publishes its Workers pricing openly in the Cloudflare Workers docs, which is part of why we can predict cost across 11 brands.

Layer 2: measurement and search (the rented commodities)

This layer is deliberately interchangeable. None of it is precious.

• GA4 + Google Tag Manager for web analytics. Free, and despite its quirks it’s the standard. We extend it with a custom channel group to surface AI-search traffic — the setup in GA4 for GEO.
• Google Search Console + Bing Webmaster Tools for index and query data. Free, non-negotiable.
• One rank-and-keyword tool (Semrush-class) shared across the portfolio rather than per-brand seats — the single biggest line item in this layer, and the one we’d drop first if budget tightened.
• A manual citation-audit spreadsheet for GEO. There’s no SaaS that measures AI citations well yet, so we run 40 lookups by hand per brand per quarter. The cheapest tool in the stack and one of the most predictive.

Everything here is a commodity we rent because building it ourselves would be a waste. The day a cheaper equal appears, we switch — no migration pain, because none of it holds our owned assets hostage.

Layer 3: AI and automation (the cheap force-multiplier)

This is the layer that changed most in 2026 and the one most teams over-buy.

• Claude for drafting, editing, alt-text generation, and running the content engine — API-priced per use, not per seat.
• Claude Code / agent workflows for the repetitive operator work: generating OG cards, auditing links, refreshing dates, drafting localized variants. These are scripts and skills, not a “platform.”
• A scheduling tool for social distribution and a lightweight email/newsletter sender for the owned audience.

We resist the urge to buy a $1,000+/month “AI marketing platform.” Those bundle a thin UI over the same model APIs we call directly for a fraction of the cost. The leverage is in the workflow — the marketing AI agents we wire up — not in a vendor’s wrapper. Per Generative Engine Optimization, the citable content this layer produces is what actually moves AI traffic.

What we deliberately don’t buy

The honest part — the categories where we spend zero, on purpose:

• No all-in-one “growth platform.” They lock your most important asset into rented infrastructure. We’d rather own the core.
• No expensive CDP for a portfolio this size. GA4 plus a simple audience list covers the need until volume genuinely demands more.
• No dedicated GEO SaaS yet. The category is immature; a spreadsheet and manual audits beat paying for tools that mostly re-skin a rank tracker.
• No seat-priced design suite per brand. SVG templates and generated OG cards cover most needs without per-seat fees.

A second category we keep out on purpose: tools that exist mainly to produce dashboards for other people. A portfolio this size doesn’t need a $1,000/month reporting suite that re-draws GA4 and Search Console data into prettier charts — the underlying numbers are free and the prettiness rarely changes a decision. We’d rather spend that budget on the rank-and-keyword tool, which actually surfaces new opportunities, than on a layer of presentation. The test we apply to every line item: does this tool change a decision, or just describe one? If it only describes, it’s a candidate to cut.

The limitation to be honest about: this stack assumes in-house technical ability. If you can’t maintain an Astro build or write a CI script, a managed CMS may be a rational trade — you’d pay rent to avoid hiring an engineer. The math flips based on whether you have operators who can run the owned core. We do, so we own it.

FAQ

How much does this whole stack cost per month? Total marketing software across all 11 brands stays in the low-four-figures per month — less than a single seat of many enterprise “all-in-one” platforms. The biggest line item is the shared rank-and-keyword tool; hosting is a rounding error.

Why static Astro sites instead of a CMS like WordPress or Webflow? Speed, crawlability, and ownership. Static sites clear the technical bar for SEO and GEO by default, cost a few dollars a month to host, and never hold your content hostage behind a subscription. The trade-off is they assume in-house technical ability.

What do you use to measure AI-search / GEO traffic? A custom GA4 channel group for the clickable share, plus a manual quarterly citation audit (40 lookups per brand across ChatGPT, Perplexity, Gemini, and Google AI Overviews) for the zero-click share. No dedicated GEO SaaS — the category isn’t mature enough yet.

Should a small team copy this stack exactly? Copy the principle — own what compounds, rent the rest — not necessarily the exact tools. If you lack engineers to maintain a static build, paying rent for a managed CMS can be the rational call.

---

Want a stack audit that cuts the tools you don’t need and reinforces the layer that compounds? Book a strategy call and we’ll map your current stack against this one.