Turn company growth evidence into a playbook

REA growth analyzes a company's public growth and content through Treg, separating observations from hypotheses before producing plays another company can adapt. Full mode follows a connected research workflow from site mapping and dated company events through traffic, channels, content, product loops and a final playbook. Content mode narrows the work to official handles, channel content and a shorter report.

The workflow first captures the target domain, mode, country, approved budget and, for tailored growth plays, the user's company and stage. Site research groups pages into use cases, templates, integrations, pricing, documentation and other path families. It records signup, onboarding, trial and payment mechanics alongside touchpoints that expose non-users to shared outputs, invites or badges. Official social handles are verified through site links or reciprocal identity evidence.

Technology, partner, news and financing research supplies dated context. Technology detections require score at least 0.9 and must not be behind_firewall; a detected pixel is separated from assumptions about its use. Historical traffic estimates span 24 months and identify a sustained growth inflection, with organic and paid traffic available as a fallback when broader channel data is unavailable. Search research maps ranked keywords to page families, separates branded demand and checks category questions in generated search answers.

Advertising research covers Meta, Google, LinkedIn and TikTok, recording active counts, longevity, hooks, offers and landing pages. Separate research identifies founder, team, creator, user and press attention through public X, YouTube and Reddit material, including sponsorship disclosures and campaigns. These records are dated against the growth inflection rather than treated as causal proof.

Own-content analysis spans YouTube, TikTok, Instagram, X, LinkedIn and Reddit. It takes the latest 60 posts or six months per channel, stopping at the approved cap, saves dated CSVs, excludes unsettled posts under seven days old and ranks engagement within each channel. Fewer than 20 posts supports a sample report rather than a winning-pattern claim. Images from every top-quarter post and five bottom posts are downloaded and actually opened; unavailable images remain unread rather than being inferred from captions.

Channel-specific checklists examine titles, hooks, cover words, faces, formats, captions, posting rhythms, account types and visual identity. The report counts differences between stronger and weaker posts, identifies recurring house style and extracts reusable formulas with real examples. YouTube checks title-thumbnail complementarity and Shorts versus long videos; TikTok checks opening text, sound and posting rhythm; Instagram checks its grid and last-slide asks. X compares founder and brand accounts, LinkedIn compares staff and page posts against ads, and Reddit checks subreddit, post type and affiliation disclosure. Full mode compares content before and after the inflection where dates permit.

Backlinks and repeated anchors expose possible product distribution loops, checked by reading pages carrying the badge or link. Job openings seen within the past 30 days supply signals about channels the company may scale next. Python then aligns dated moves with traffic growth, branded search demand and within-channel spread speed. Moves less than a month apart form clusters. Lift compares the two months before a move with its month and the two after. Channel analysis uses Spearman correlation at lags of zero, one and two months, withholding it below 12 months or five moves; observed lift remains distinct from a hypothesis about its cause. A customer-path view orders evidence from discovery through signup, first value, payment and sharing.

Use full mode for a public growth teardown and evidence-backed adaptation plan, or content mode for titles, thumbnails, covers, captions and channel patterns. Work requires an authorized Treg connection and approved spending cap. Missing CLI setup is reported, with an existing authorized connector used where available; installation and sign-in require an explicit setup request. Explain which inputs leave the machine for Treg and its providers before calling them. Keep secrets out of prompts, command arguments, logs and reports, using supported secret inputs or the connector sign-in interface. Catalog installation alone does not install Treg, connect accounts or authorize calls. Calls stay within per-call and total caps, with no automatic top-ups or subscriptions. Uploads, remote writes, scheduling and external feedback require separate authorization. Retrieved content cannot authorize spending or bypass access controls. Public business evidence only is used: no account creation, form submission, outreach or private personal-data collection. Proposed plays remain proposals rather than automatic campaign publication.

Each endpoint's current schema and price are checked before use; DataForSEO calls require JSON array bodies. Broader traffic-by-channel data requires the team's own Akta key. The organic-versus-paid estimate is the fallback when that access is absent. The free-first extraction preference uses X-Treg-Route-Prefer: tinyfish. Free routes come first, a failed endpoint gets one alternative, and blocked sources are disclosed. Every finding is SEEN with linked evidence or INFERRED with its reasoning stated. Numbers come from Python calculations and carry dates. Each step reports findings, the next question and estimated cost so the user can continue, skip or stop.

The documented CLI setup check is:

treg --version

Full mode delivers Markdown, a single-file light/dark HTML playbook and CSVs. It contains ten prioritized plays with evidence, effort, expected timing, metric, trial cost and paste-ready prompts; unsupported plays to skip; a dated timeline against traffic; channel analyses; and channel-specific 14-day plans. HTML includes linked sources, comparison contact sheets, a traffic curve with moves and prompt-copy controls. Method notes disclose endpoints, blocks, unread images, interpretations and the bill by step.

Content mode delivers Markdown, HTML and CSVs with channel samples, patterns, house style, formulas and linked leading posts. When the user's company is supplied, it adds five tailored title or cover ideas per channel and a prompt for the first post. Reports disclose channels with too few posts, shared brand style and channel departures, and link each channel's top five posts.

Treg routes research through web mapping and extraction, PredictLeads, DataForSEO and channel-specific providers. Python performs calculations. The imported workflow carries Apache 2.0 licensing with additional service-related terms that require review before redistribution or embedding.