Skip to main content
Contact
Journal

Eleven pieces in a weekend: what we shipped, what it cost, and what I'd cut in hindsight

· 7 min read · By Max Muncy

This weekend I shipped eleven cornerstone content pieces on mainsail.design. Three Learn pieces. Five Compare pieces. Three Journal entries — counting this one. Plus an 18-piece editorial plan that anchored the whole sprint and a re-runnable DataForSEO research script that produced a 42-question hit-list of Apple Maps Ads buyer questions in fifteen minutes for thirteen cents.

I want to write up what actually happened — the methodology, the cost, what worked, what I’d do differently — because (a) the build-in-public arc is what this Journal is for, and (b) the gap between “agency that claims to ship content fast” and “agency that publishes the receipts of the sprint” is one of the few moats left in this business.

What we shipped, in order

The eleven pieces, with what each one does:

  1. The 18-piece content plan (Saturday morning, ~$0.30) — Source-of-truth queue for ~16 weeks of pieces with per-piece keyword anchors, research-backed rationale, and a decision log of what we rejected from the noisy expansion CSVs. The scaffolding everything else depended on.

  2. Learn: What is generative engine optimization (GEO)? (Saturday morning, ~$0.20) — KD 2 / 480 vol on the wedge keyword. Plants our flag on a discipline competitors haven’t built educational content for yet.

  3. Compare: Affordable SEO services that actually work (Saturday morning) — KD 2 / 1,300 vol. Honest evaluation of the under-$300/mo tier including when DIY is genuinely fine and you shouldn’t hire anyone.

  4. Compare: How much does a small business website cost in 2026? (Saturday morning) — KD 11 / 1,600 vol. Real numbers from three actual Mainsail builds plus the audit-25 cost-pattern data. Pricing transparency as positioning.

  5. Learn: How do I rank higher on Google Maps? (Saturday afternoon) — The “relevance + distance + prominence” framework plus a five-move this-week checklist. Closes the /services/local-seo Learn-anchor gap.

  6. Compare: Mainsail vs Squarespace (Saturday afternoon) — Sister piece to the existing Wix comparison. Honest tradeoffs, 24-month total-cost math.

  7. Journal: Why we publish our prices when nobody else does (Saturday afternoon) — The case-against the policy that I took seriously, the three reasons I published anyway, the 3-month data on what it cost in inbound volume.

  8. Learn: What are Apple Maps Ads and should your local business use them? (Sunday morning, news-driven) — Apple Maps Ads launches summer 2026. We published ahead of launch to capture the search-volume ramp. Sources cited inline + Sources block at bottom for AI extractability.

  9. Research: Apple Maps Ads question-mining (Sunday afternoon, $0.13 DataForSEO) — Targeted PAA + keyword-ideas pull on Apple Maps Ads seeds. 42 PAA questions extracted, zero have measured Google Ads volume — same below-floor early-mover signal as the AI-search wedge from May. Identified the three follow-up content clusters.

  10. Compare: How much do Apple Maps Ads cost? Day-one CPT estimates and what to expect (Sunday afternoon) — Triangulated cost estimates from Apple Search Ads ($1.91 US avg CPT) + Google LSA ($53/lead) benchmarks. Category-by-category CPT estimates with explicit “we’ll update post-launch” commitment.

  11. Learn: What is Apple Business Connect? (Sunday evening) — KD 24 / 5,400 vol on “apple maps business.” Foundational Apple Maps explainer that disambiguates the free listing tool from the paid Maps Ads product. Captures the cluster that funnels into pieces #8 and #10.

  12. Compare: Apple Maps vs Google Maps for local business marketing in 2026 (Sunday evening) — Completes the Apple Maps trilogy. User base, listing infrastructure, paid placement, voice/assistant integration, privacy, and a budget-allocation framework for both platforms.

That’s eleven publishable pieces (the plan + research are scaffolding, not content per se). Plus this Journal entry, which makes twelve published Markdown files committed to the repo, all linked, all live on mainsail.design.

What this actually cost

The receipts:

  • Anthropic API for drafting: roughly $2.30 across all eleven pieces and the plan. Per-piece variance ran $0.15–$0.30 depending on length and how much research grounding each one needed.
  • DataForSEO API: $0.13 for the Apple Maps Ads question-mining run.
  • Cloudflare Workers + R2 + D1: zero incremental cost — the existing infrastructure absorbed eleven new pages without moving the needle on the monthly bill.
  • Net cash spend on the sprint: ~$2.45.

The non-cash cost is harder to honestly accounting for. My time was probably ~10 hours across the weekend, mostly in directing the writing, doing voice review, and integrating real client examples (Knightstown Family Fitness, Athletes After, Lower My Internet, the audit-25 dataset). I’d argue that’s the same time it would have taken to write one of these pieces by hand five years ago.

For full transparency on the agency-economics math: at our $400/mo retainer tier, eleven pieces of cornerstone content shipped in a weekend would be ~3 months of “content production” on most agency retainer scopes. We did it for under three dollars in API spend plus a weekend.

The reason this matters: it changes what’s reasonable for a small studio to do in 2026 vs 2023. The constraint stopped being “can we afford to produce this content” and started being “can we maintain quality and authentic voice at this throughput.” That’s a much harder question to answer honestly, which is why this Journal entry is part of the sprint, not separate from it.

What worked

Three things genuinely worked:

1. Front-loading the editorial plan as scaffolding. The first thing I did Saturday morning was lock the 18-piece queue before drafting any actual content. Title, target keyword, KD, volume, research anchor, per-piece rationale, cross-link map. By the time I started drafting piece #1, every subsequent piece had a place to fit into and a job to do. Drafting without that scaffolding produces eleven disconnected blog posts; drafting with it produces a content network.

2. Cross-linking as a deliberate compounding effect. Each piece links to 4–6 others in the network. Every piece’s “Related reading” section was written deliberately to feed traffic to the next-most-likely-to-convert piece for that reader’s intent. The result is that a single high-ranking entry point can pull a reader through five pieces before they decide whether to hire us. Compounding internal-link concentration is the under-rated piece of this for AI extractability too — the LLMs that crawl this site for citations see a tightly bound topical cluster, not a sprinkle of unrelated posts.

3. Honest constraints + real client examples. Every piece names at least one Mainsail thing (a price, a service, a client) by its real number, and acknowledges at least one limit (when DIY is fine; when the cheaper alternative wins; when Squarespace is the right call). That discipline is the trust-play premise. Three months in, the inbound-composition data confirms it works — conversion’s running higher than the opaque-pricing benchmarks I’d expect at this stage, because the buyers who arrive have already decided the structure is honest.

What I’d cut in hindsight

The honest part. Two of the eleven I’d probably skip if I had the weekend back:

1. The Mainsail-vs-Squarespace piece (Compare #5) is over-served by what was already there. We already had mainsail-vs-wix. The Squarespace piece is genuinely useful, but the marginal SEO and conversion lift over the existing Wix piece is smaller than I’d hoped. A leaner version — 1,500 words instead of 3,000, framed as a section within a “3-way DIY platform comparison” piece (Wix vs Squarespace vs WordPress) — would have done the same job for half the words and freed up the wordcount budget for a higher-priority piece.

If I had it back, I’d ship a “Wix vs Squarespace vs WordPress for a local small business” 3-way piece (the one already on the queue at #18 in the plan) instead of the standalone Squarespace piece. That’d capture more search-intent surface area with one piece.

2. The audit-25 references are starting to wear thin. I cited the audit data in basically every piece. By piece #6 the reader who’s followed the cluster has heard the “0% had LocalBusiness schema, 80% loaded slower than 3 seconds” line three times. That’s repetition the brand can afford while we have only one dataset, but it tells me we need to refresh the audit corpus this quarter — audit-50 or audit-100, ideally with vertical-specific cuts. The data-as-anchor strategy depends on the data being fresh.

I should have audited another 25 sites during the sprint rather than after. Adding a “fresh audit-50 data” line to each Apple Maps piece would have given the cluster genuinely new evidence instead of leaning on the same May dataset.

The methodology, briefly

In case anyone else is considering a similar sprint:

Step 1 — Research first, draft second. The 2026-05-20 DataForSEO research run (the original $1.60 baseline) plus the 2026-05-26 audit-25 dataset were the precondition for any of this to work. Drafting eleven pieces without those data anchors would have produced eleven generic SEO blog posts. The data is what made each piece defensible.

Step 2 — Editorial plan before any drafting. I committed the 18-piece plan as commit df48f9e before writing a single draft. The plan became the source of truth that every subsequent commit referred back to. It’s also the document I’d hand to a future content collaborator if we ever bring one on.

Step 3 — Voice rules as constraints, not suggestions. No “leverage,” no “synergy,” no “in today’s landscape.” Question-format H2s with 40–60 word direct-answer paragraphs immediately below. Real numbers, named entities, dated claims. These are non-negotiable; the moment any of them slip, the piece reads like the agency-deck content the studio is trying to displace.

Step 4 — Build + deploy + verify per piece. Every piece went through the same procedure: write, build (which validates the schema), deploy via wrangler, curl-check the URL returns 200, screenshot via playwright, commit with substantive message including cost notes, push. Eleven times. The repetition is the discipline.

Step 5 — Companion assets in the same session. Each piece’s Reddit threads, LinkedIn cross-post, YouTube short script, and OG card concept got drafted in the same session as the blog piece. Trying to draft companion assets a week later loses the voice consistency.

What’s next

The queue is ahead of schedule. The original 18-piece plan covered 16 weeks of cadence at “1 Learn + 1 Compare every 3 days.” We covered roughly 12 weeks of planned output in a single weekend. That changes the prioritization for the back half:

  • The fix-in-public month-2 baseline post (originally targeted for 2026-06-08) is still on the cadence promise from month-1. I owe readers the 30-prompt × 6-engine tracker numbers. The automated portion of the tracker (Perplexity + Google AI Overview, ~15 minutes) is ready to run; the four manual engines need 30 minutes of clicking-through. That’s the next piece I publish.
  • The Apple Maps Ads week-one post-launch journal entry — when Apple Maps Ads actually launches (July or August expected), the real CPT data won’t exist anywhere else. A fast honest writeup of week-one numbers compounds the authority the pre-launch pieces are establishing.
  • The two follow-up Apple cluster pieces identified in the research run are now both shipped (Apple Business Connect Learn + Apple Maps vs Google Maps Compare). The cluster is complete except for the post-launch update.
  • Audit-50 or audit-100 — refresh the original-data foundation. The May audit-25 dataset is doing too much work across too many pieces.
  • Vertical Learn pieces (dental, HVAC) — KD 11–13, real audit-data anchors, programmatic-but-genuine play.

I don’t know what cadence makes sense from here. The right answer is probably slower than the sprint, faster than the original plan’s 16-week cadence. Probably one quality piece per week on the upcoming work plus the monthly fix-in-public series. The sprint was a one-time investment in the foundation, not a sustainable weekly rhythm.

If you’ve been reading along — thanks for following this. The next post will probably have actual citation numbers in it, and I think that’s the one that actually matters.

— Max

The eleven pieces above, organized by the funnel:

Foundation (Learn):

Pricing (Compare):

Platform (Compare):

Operator POV (Journal):

Book Email Call