Filmcuts Engagement Proposal
Where Filmcuts Sits Today
The audit (May 12) found three things that need to happen together: unblock the AI crawlers (today 5 of 7 major LLM bots get a 403 from filmcuts.io), make the site readable once they're through (server-side render the homepage, about, pricing, and resources pages; add the schema layer that's missing on every page), and start building authority on a parallel track (Wikidata, third-party reviews, listicle inclusions, founder placements — quarter-long lead times). The first two are engineering. The third is editorial and outreach.
This proposal sketches three engagement shapes: a one-time kit, and two monthly partnerships with a 3-month minimum. They differ in who ships what, how fast it lands, and whether ongoing measurement and outreach are part of the deal. Pick the shape that matches how much of this you want to own internally versus hand over.
The audit told you what to fix. This proposal is about who fixes it, who measures whether it worked, and how fast Filmcuts becomes the agent-commerce-ready stock-footage library before any incumbent does.
The Four Layers, In Sequence
AI search visibility is a different discipline from traditional SEO — different surfaces, different feedback cycles, different failure modes. The work that actually moves the needle lives at four layers, and they have to be sequenced. Skip the foundation and the rest compounds in the wrong direction.
- Layer 1 — Crawler accessibility & structured data. AI bots have to be able to read your site at all (HTML they can parse, not JavaScript shells), and the structured data has to accurately describe what you are. The foundation; nothing downstream works without it. For Filmcuts today, 5 of 7 major LLM bots get a 403 from Cloudflare — this is the first thing to fix.
- Layer 2 — Entity graph. AI systems disambiguate brands using a graph of relationships: founders, category, related entities. Wikidata is the canonical reference most of them check. Without a Wikidata record, AI guesses — and for a young brand, "guesses" often means thin or wrong answers.
- Layer 3 — Citation density. AI cites venues, not pages. Most of the visibility signal in your category lives in third-party venues AI treats as authoritative — industry publications, listicles, reviews, comparison content. Building presence in those venues over time is what compounds. A brand-site rebuild moves recognition by a few percentage points; the rest comes from venues you don't own.
- Layer 4 — Counter-misinformation & measurement. Existing fabrications or wrong attributions in AI memory don't disappear when you publish new content. They have to be deliberately overwritten with high-authority correcting signal. This is the slowest layer; it's why continuous measurement matters — without it, you're guessing for months at a time.
Each layer depends on the one below it being right first. Citation density built on top of an unresolved entity graph reinforces the wrong relationships in AI memory. The rest of this proposal is structured around shipping these layers in order.
Three Ways To Run It
One foundation, three operating models. The choice is how much of the work you want to own internally versus hand over.
A handover package. We ship the critical fixes to production, hand over the SSR migration plan for your engineer, and deliver every on-site asset the audit calls for. No ongoing work. Right if you have engineering capacity and want full control of the rollout.
- Cloudflare bot-unblock coordinated & verified — removes the network-level block that's stopping ChatGPT, Claude, Perplexity, and Gemini from reading filmcuts.io today. Nothing else matters until this is fixed.
- robots.txt + og:url + llms.txt + agents.md — the explicit "you may read this" / "this is who I am" signals every AI bot looks for first. Committed via PR to your repo.
- FAQ page built with FAQPage schema — structured data that lets AI assistants pull your FAQs directly into their answers when someone asks "what film stock does Filmcuts use" or "how does licensing work." New Next.js route deployed.
- curl verification per AI bot user-agent — we actually test each LLM bot against the live site and document before/after, so you have proof the fixes landed.
- SSR migration plan for homepage, /about-us, /pricing-plans, /resources — file-by-file engineering plan for converting these pages from client-rendered JavaScript (which AI bots can't read) to server-rendered HTML (which they can). Your engineer executes; we provide the exact data-fetching changes per file.
- 8 JSON-LD schema templates — the structured-data definitions for every page type: who you are (Organization), founders (Person), how to search (WebSite + SearchAction), videos (VideoObject), packs (Product/Offer), FAQs (FAQPage). Without these, AI sees an undifferentiated page; with them, it sees a structured catalog.
- 12 FAQ Q&As drafted with FAQPage JSON-LD — the actual questions AI assistants are likely to be asked about Filmcuts (Super 8 vs 16mm vs 35mm, licensing tiers, scan resolution), pre-written so AI can cite them verbatim.
- Comparison page draft (vs Artgrid, Filmsupply, Filmpac, Stockfilm, raw.film) — the page that wins the "Filmcuts vs Artgrid" answer when an AI assistant compares stock-footage libraries. Without this page on your site, the AI cites a competitor's framing.
- About-us rewrite surfacing Austin's commercial reel — named founders with verifiable work (Aman, Amway, Park Hyatt) weight heavily for AI citation. Currently this credibility signal is buried; we put it in extractable form.
- Homepage H1 + brand definition block — the canonical two-sentence answer to "what is Filmcuts." This is what AI assistants quote when introducing the brand in their answers.
- Wikidata entity drafts for Filmcuts and Austin Divine — Wikidata is the canonical entity database AI assistants reference. Without a record, AI guesses. Drafts are ready to submit (you submit; we provide the prefilled record).
- Third-party review surface setup checklist (Trustpilot, ProductHunt, G2) — AI cites venues, not pages. These are the venues it treats as authoritative for stock-footage sentiment, with the right places to claim profiles and request reviews.
- AI Visibility Platform scan + dashboard — your before-state across ChatGPT, Claude, Gemini, Perplexity. You see exactly what each LLM says about Filmcuts today, so future improvements are measurable.
- 60-min handover call with your dev — we walk through the SSR plan and answer questions live so your engineer can ship without back-and-forth.
Walk-away clause: if the handover call doesn't land the deliverables you expected, don't pay the second 50%. The risk on delivery sits with us, not you.
Full Implementation Kit delivered in weeks 1–2, then three months of measurement, monthly strategy, PR refresh, and 6 ghostwritten longform articles. Your team ships the SSR migration on its own timeline; we own everything around it. Best fit for Filmcuts's stage and stack.
- Monthly AI Visibility Platform scans — we re-run the scan each month so you see the delta between month 1 and month 3 in the dashboard. Without ongoing measurement, you're guessing about whether anything moved for 12+ weeks.
- 30-min monthly strategy call — 15 minutes reading the scan delta, 15 minutes deciding what moves next. The plan adapts to what's working, not what we predicted in month 1.
- PR & Authority Strategy, refreshed monthly — this is Layer 3 citation work. Updated target map of industry publications, editorial angles per pub, talking points and quote anchors from Austin's reel. Strategy doc you reference; outreach execution stays with you (the founder voice matters for pitches).
- 2 ghostwritten longform articles per month (6 total over 3 months) — this is what AI cites when "best stock footage shot on film" gets asked. Article/HowTo JSON-LD, internal linking to relevant clips and packs, ready to publish. Without content depth, your category answers default to whoever else has written more.
- Wikidata monitoring — Wikidata edits get reverted by editors who don't know the brand. We watch the Filmcuts and Austin Divine submissions for reverts or edits and fix them as they appear.
- End-of-engagement wrap-up document — showing the delta across the three months: what moved, what didn't, where the highest-leverage next moves sit. The before/after of three months of work in one place.
Everything in Option B, plus we ship the heavy engineering and double the content cadence. SSR migration deployed to production, full schema rolled out across the catalog, 12 ghostwritten longform articles. Zero implementation time on your side. Right when you want one team accountable for engineering, content, and measurement end-to-end.
- SSR migration shipped to production (homepage, /about-us, /pricing-plans, /resources) — the heaviest engineering piece in the audit, off your plate. Without SSR, AI bots see empty JavaScript shells where your content should be. With SSR shipped, everything else in this proposal actually compounds.
- Full schema rollout across the catalog — VideoObject for every clip, Product/Offer for every pack, global Organization + Person + WebSite + SearchAction in the layout. AI agents that recommend or transact stock footage need this metadata to know what's available and at what price. Filmcuts becomes machine-readable in a way no incumbent currently is.
- 4 ghostwritten longform articles per month (12 total over 3 months) — double B's cadence. At one article per week, the category-level questions ("shot-on-film vs digital," filmmaker interviews, license tier explainers) start being answered by Filmcuts in AI responses, not by Artgrid or Pond5.
- The work itself + the measurement layer that shows you what moved. We don't promise specific recognition percentages or citation timing — AI training cycles are outside our control. What we promise is that you see what's changing in real time, and that the strategy responds to it.
What's Included Where
| Deliverable | A — Kit | B — DWY | C — DFY |
|---|---|---|---|
| SSR migration plan (file-by-file spec for the 4 main pages) | ● | ● | ● |
| llms.txt + agents.md written | ● | ● | ● |
| JSON-LD schema templates (8 types) | ● | ● | ● |
| FAQ page content (12 Q&As) | ● | ● | ● |
| Comparison page draft (vs 5 competitors) | ● | ● | ● |
| About-us rewrite | ● | ● | ● |
| Wikidata entity drafts (Filmcuts + Austin) | ● | ● | ● |
| Third-party review surface checklist | ● | ● | ● |
| Handover call with engineer | 60 min | 60 min | 60 min |
| AI Visibility Platform (Scan + Dashboard) | 1 scan | 1/month | 1/month |
| Monthly 30-min strategy call | — | ● | ● |
| PR & Authority Strategy (refreshed monthly) | — | ● | ● |
| Ghostwritten longform articles per month | — | 2 | 4 |
| Total articles over 3 months | — | 6 | 12 |
| Wikidata monitoring | — | ● | ● |
| Cloudflare bot-unblock coordinated & verified (clears the 403 blocking 5 of 7 LLMs today) | ● | ● | ● |
| robots.txt + og:url fix committed to production | ● | ● | ● |
| llms.txt + agents.md deployed to production | ● | ● | ● |
| FAQ page built with FAQPage schema in production | ● | ● | ● |
| curl verification per AI bot user-agent (documented before/after) | ● | ● | ● |
| SSR migration shipped to production (we write the code) | — | — | ● |
| Full schema rollout (VideoObject per clip, Product/Offer per pack) | — | — | ● |
| Investment | €2,000 one-time | €3,000 €1,000/mo × 3 | €6,000 €2,000/mo × 3 |
On SSR migration specifically: the file-by-file plan is in every option — homepage, /about-us, /pricing-plans, /resources, with the exact data-fetching changes per file. The difference is who writes the code. In A and B, your team executes the plan (the spec is detailed enough that it's a clear engineering task, not a research project). In C, we write and ship the code. The plan-vs-ship distinction is the same logic that applies to comparison content, About-us rewrite, etc. — we deliver the asset in all tiers; C extends to "and we deploy it for you."
How An Engagement Runs
Option A — Implementation Kit (10 working days)
Content & schema drafted
FAQ Q&As, comparison page draft, About-us rewrite, homepage H1, llms.txt and agents.md written. All 8 JSON-LD schema templates assembled. SSR migration plan written file-by-file.
Critical fixes shipped to production
Cloudflare bot-unblock coordinated & verified (curl-tested per AI bot user-agent). robots.txt + og:url + llms.txt + agents.md shipped via PR. FAQ page built with FAQPage schema, deployed. The 5-of-7 LLM block is cleared.
AI Visibility Platform scan run
Prompt set built around Filmcuts's categories. Named competitor tracking configured (Artgrid, Filmsupply, Filmpac, Stockfilm, raw.film). Dashboard provisioned. First scan executed across ChatGPT, Claude, Gemini, Perplexity — your before-state is locked.
Authority anchors prepared
Wikidata entity drafts for Filmcuts and Austin Divine, ready to submit. Third-party review surface checklist drafted (Trustpilot, ProductHunt, G2). Internal review and packaging.
Handover & delivery
60-min handover call with your dev. Full delivery via shared Notion + ZIP. Dashboard walked through live so you see your before-state alongside the deliverables. Your engineer leaves with the SSR migration plan ready to execute.
Options B & C — Three-month sprint
Kickoff & foundation shipped
Full Implementation Kit delivered exactly as in Option A — content drafted, critical fixes shipped to production, AI Visibility Platform scan run, Wikidata drafts and review surface checklist prepared. In Option C only: SSR migration of the 4 main pages shipped to production, full schema rollout across the catalog (VideoObject per clip, Product/Offer per pack).
First content live + PR Strategy delivered
In B, the first 2 ghostwritten articles published; in C, the first 4 articles. Initial PR & Authority Strategy doc delivered for both — target map, editorial angles per publication, talking points and quote anchors. Wikidata entities monitored. Monthly scan accumulating on the dashboard.
Cadence + monthly strategy
B delivers 2 articles/month, C delivers 4. PR & Authority Strategy refreshed monthly based on what's resonating. 30-min strategy call each month reads the delta against month 1 — not just the previous scan. Monthly scans accumulating, dashboard tracking the trajectory.
Wrap-up
Wrap-up document showing the delta across the engagement — what moved, what didn't, where the highest-leverage next moves sit. The before/after of three months of work in one place.
Quick Answers
Option A: 50% on engagement start, 50% on delivery.
Options B & C: monthly in advance, first invoice on kickoff, then same date each month.
Wire or card via Stripe.
Walk-away clause on Option A: if the handover call doesn't deliver what you expected, don't pay the second 50%. The risk on delivery sits with us, not you.
On B and C: the 3-month engagement closes cleanly at month 3 — no auto-renewal, no minimum extension.
Option A: someone to merge our PRs (small, reviewable patches — repo access optional), Cloudflare account access for the bot-unblock rule, and your engineer for the 60-min handover call.
Option B: same as A in week 1; after that minimal — articles come ready to paste, scans run automatically, strategy calls are 30 min/month. Your engineer ships SSR on your own timeline.
Option C: GitHub repo and Cloudflare access during week 1. Then minimal.
We send the 1-page agreement and Stripe link within 24 hours of you replying. Kickoff the same week.
The Three Windows
Three timing factors worth mentioning — none dramatic, all real.
- The agent commerce window. Artgrid, Filmsupply, Pond5, Storyblocks — none publish the llms.txt / agents.md / MCP layer today. Filmcuts shipping it in the next quarter would be first in the category. By Q4 the larger libraries will have caught up and the first-mover signal is gone.
- Authority lead times. The slowest levers — listicle inclusions, Wikidata propagation, podcast bookings — take 6 to 12 weeks from first outreach to first citation lift. Starting now means measurable results in about 8 weeks. Starting three months from now puts those same results into Q4 — the season when brands are already locking Q1 footage budgets.
- Before-state scans decay. AI citation rankings shift every few weeks. The scan we run in your kit this month is the cleanest before-state to compare against; one run in three months means measuring the lift from a noisier starting point.
All three windows argue the same thing: this quarter is a better starting line than next. Not because of us — because of how AI training cycles, brand-budget calendars, and citation memory actually work.
Next Step
You're building something I genuinely care about seeing work. The shot-on-film angle is more defensible than people realize, but the AI visibility window doesn't stay open forever.
Whatever you pick, I want it to actually fit where Filmcuts is right now. If A, B, or C isn't quite right, we can adjust the scope based on your needs.
Cristoforo