Internal Reference · Working Draft
The content engine is built as one linear pipeline where each step's output is the next step's required input. TTS runs a shorter, materially different pipeline. This maps which standard steps TTS runs, skips, or replaces; where a skipped step's output is still consumed downstream; and what to cut or substitute.
TTS's engine delivery master points at the universal pipeline, and its config assumes the standard interview-driven, keyword-mapped, 12-month-calendar model. The actual TTS flow is a different, shorter pipeline: no SEMrush keyword chain, a fixed mini-site instead of keyword pillars, and the onboarding call (plus crawl and geo research) as the genesis instead of monthly expert interviews. Nothing in the engine yet lets a tenant declare "I run these steps, not those," so any step that hard-requires a skipped step's output will stop.
| Standard engine step | TTS actual | Disposition |
|---|---|---|
| 1.1 Intake form (discover services) | Order form (identity) + quick intake (license/insurance, service areas, services from the fixed TTS master list, URLs, GBP claimed y/n) | REPLACE (fixed service list, not discovered) |
| 1.2 Onboarding call, transcript 1 | 45-min recorded call: story, edge, the one thing to be known for | RUN (the single genesis) |
| Market research (prompts, ingest) | Crawl (reviews, photos, logo, headshot) + geo research (market by trade+metro, avatar by city) | REPLACE (geo-research library + crawl) |
| 1.5 Competitor gap (SEMrush) | none | SKIP |
| 1.5b Brand Anchor, Cornerstone | Built from the call (intent + edge) + derived customer moment (reviews + recording) + geo research | RUN (renamed Cornerstone) |
| 1.6 Keyword research (SEMrush) | none | SKIP |
| 1.7 Audience-keyword validation | none (audience from geo avatar + reviews) | SKIP |
| 1.8 Keyword strategy (priority matrix, pillars, ITP) | none | SKIP |
| Website pillars + ITP (keyword-driven architecture) | Fixed mini-site page set + SAB per-city model | REPLACE |
| Design prototypes + call 2 (design choice) | Templated mini-site (3-4 pages, one template for every member) | REPLACE |
| Phase 2 content plan (keyword-mapped 12-month interview calendar) | Lighter calendar: 4 blogs/month, topics from geo + services + case studies | REPLACE |
| Phase 3 brand docs, voice skill, site buildout | Cornerstone + voice; mini-site built from crawl + recording (W-2) | RUN / REPLACE (mini-site, not full site) |
| Phase 4 expert interview (genesis), blogs, socials | Onboarding call + case studies + geo research as genesis; blog-then-social loop | REPLACE (no monthly interview) |
Mini-site page set (from W-2 / the portal): business description, specialties, service areas, work examples, per-city page status, FAQ, blog. Social cadence: ~40/month per contractor (4 blogs, 12 Facebook, 8 GBP, 8 Instagram, 8 Threads). Connected platforms: Facebook, Instagram, Threads, GBP. No LinkedIn.
Each row is a standard output TTS will not produce, the downstream steps that currently require it, and the resolution.
| Skipped/replaced output | Downstream consumers | Resolution for TTS |
|---|---|---|
| Competitor gap report (1.5) | keyword-strategy (1.8); positioning | CUT. TTS positioning comes from the canon/P-1b docs, not SEMrush. |
| Master keyword list (1.6) | 1.7, 1.8, content plan, blog writing, pillars | SUBSTITUTE. Light local keyword/topic signal from the geo market report + the TTS master service list + served city names. No SEMrush volume/KDI. |
| Validated audience map (1.7) | content plan; blog audience targeting | SUBSTITUTE. Audiences come from the geo avatar report + the reviews crawl. |
| Keyword priority matrix + pillar buckets + ITP (1.8) | website architecture; content plan; blog writing (one-keyword-in-H1 rule, SEO assets) | SPLIT. Mini-site: replace with the fixed page set + SAB per-city; the TTS "ITP" is the contractor's master profile / lead CTA. Blogs: open decision (Q1 below). |
| Website pillars + ITP step | Phase 3 buildout | REPLACE with the fixed mini-site template. |
| Design prototypes + call-2 design choice | Phase 3 buildout | REPLACE with the mini-site template. |
| Content plan (keyword-mapped 12-month interview calendar) | blog writing (reads a plan row: topic, keyword, pillar, shortlink) | REPLACE with a lighter calendar (4 blogs/month, topics from geo + services + case studies). Fix the engine config (it assumes the 12-month calendar + a current interview). |
| Monthly expert interview (Phase 4 genesis) | expertise questions; interview recording; blog genesis | REPLACE. Genesis = the one onboarding call + ongoing case studies + geo research. Blog writing must accept a non-interview genesis (Q2 below). |
Where a TTS run hits a hard-stop today, regardless of intent.
Loads the Brand Anchor (fine, TTS has Cornerstone), then queries the wiki and applies keyword targeting + reads a content-plan row. It needs: (a) the wiki populated for TTS, (b) a content-plan row source, (c) the keyword decision (Q1).
The standard flow fills the wiki from three research files TTS does not have. TTS needs an ingest path that takes its real inputs: the geo-research output (done), the reviews crawl (audiences / hooks / stats), and the call transcript (topics / hooks). The crawl + call ingest is to build.
Hard-stops if brand docs, client CSS tokens, or the wiki are unpopulated. For TTS, design/CSS comes from the mini-site template / brand-extract and the wiki from geo+crawl+call. The gate must become track-aware so it does not demand keyword-strategy artifacts a TTS run never makes.
"Any Phase 4 work" hard-stops on brand docs + CSS + wiki universally. Same track-awareness need as the preflight gate.
Three moves, only the third is real engineering. These are engine work, written generically, landed in an engine session.
Lean: yes but light. Derive local keywords from geo + services + city (the per-city pages are a local-SEO effort), and drop the SEMrush-grade volume/KDI rigor. Alternative: relax the one-keyword-in-H1 gate for TTS blogs entirely.
Lean: TTS blogs derive from the one onboarding call + ongoing case studies + geo research; no monthly expert interview. This is the biggest structural divergence from the standard model.
Recommend a per-track manifest the engine reads (and un-linking or overlaying the TTS delivery master), plus track-aware gates. Ties into the cloud move and the Cornerstone naming work.
A lighter calendar with no interview mapping, and a matching fix to the engine config, which today assumes the 12-month interview calendar.
Decide Q1 and Q2 first (they define what the blog step needs). Then build the two missing engine pieces in an engine session: the crawl+call wiki ingest path, and the track manifest + track-aware gates. The geo-research library is already in place as the research substitute. Do not refactor the gates blind; this map is the input to that engine spec.