Automation: Monthly Multi-Client SEO + AI Ranking Reports (DataForSEO, GSC, GA4, claude code)

Overview We run a local-SEO agency and produce a monthly SEO report per client by hand. We want it fully automated: pull every data source, generate AI-ranking checks, assemble a styled per-client report, and drop it in a folder with a draft cover email ready for us to review and send. This is a paid test project scoped to one sample client. If it's clean, the follow-on rolls it out to all ~10 clients on a monthly schedule, at a proper budget. Required skills - DataForSEO API (this is central - rank tracking, SERP, on-page/crawl, and backlinks all pull from here; must be comfortable with its Labs, SERP, On-Page, and Backlinks endpoints) - Google Search Console API and GA4 Data API (OAuth or service account) - Perplexity API and Google AI Overviews for AI-ranking checks - Automation/orchestration (n8n, Make, or Python/Node) with scheduling - API cost, rate-limit, caching, and error handling - Report templating/rendering (HTML or PDF/Google Doc) What the finished system does (per client, config-driven) 1. Search Console - last-month vs prior-month clicks, impressions, CTR, avg position; top queries and top pages with deltas. 2. GA4 - organic sessions, engaged sessions, conversions/events, landing-page performance, month-over-month deltas. 3. Rank tracking (DataForSEO) - keyword positions and movement via DataForSEO Labs/SERP, including local map-pack where relevant. 4. Technical / on-page (DataForSEO On-Page) - crawl summary: indexation, broken links, Core Web Vitals, title/meta issues, schema presence. 5. Backlinks (DataForSEO Backlinks) - new/lost referring domains and anchor summary for the month. 6. AI ranking - run a niche query bank through the Perplexity API and Google AI Overviews and record whether the client is cited/mentioned, and which competitors are. (We provide the query bank and report logic direction.) 7. Assembly - merge all of the above into a single styled report per client (HTML we can host, or a clean PDF/Google Doc) using a template we provide. Every number must trace to a source; no invented metrics. 8. Delivery - write the report to a per-client folder and generate a draft cover email (do NOT auto-send anything - drafts only, we send manually). Engineering requirements (this is where we're testing you) - Config-driven, not hardcoded - one config entry per client (domain, GSC property, GA4 property ID, DataForSEO keyword set/location/language, competitors). Adding client #11 is a config edit, not a code change. - Idempotent re-runs - re-running for the same month must not duplicate or corrupt output. - Coverage gate - if any data source fails or returns empty, the run flags it explicitly (PASS/WARN/FAIL per source). Silent skipping is an automatic fail of this project. - API cost + rate-limit handling - respect DataForSEO and Google/Perplexity quotas, back off/retry on limits, cache expensive calls (DataForSEO SERP/backlink pulls especially). Print an estimated/actual API cost summary per run. - Dry-run mode - a flag that runs the whole pipeline against sample data with zero paid API calls. - Logging + error handling - a run log showing what ran, what it cost, and what failed, readable by a non-engineer. - Scheduling - built to run on a monthly cron/n8n schedule (demonstrate one triggered run; full scheduling is the follow-on). - Handover - a Loom walkthrough plus a README: how to add a client, run it, read the coverage report, and rotate credentials. Stack (your call, justify it) - n8n, Make, or plain Python/Node are all fine. Tell us what you'd use and why. DataForSEO for rank/on-page/backlinks/SERP. GSC + GA4 via Google APIs. Perplexity API + Google AI Overviews for AI ranking. Out of scope for this test (these are the paid follow-on): rolling out to all 10 clients, live production scheduling, and connecting to our live client accounts. This test runs against one sample property we provide with scoped, read-only, revocable credentials. Acceptance criteria - One full report generated for the sample client with every section populated from real (or clearly-labelled sample) data, DataForSEO sections included. - Coverage gate output present and correct (deliberately break one source and show it flags WARN/FAIL, not silence). - Idempotent: two runs for the same month produce one clean result. - Dry-run mode works with zero paid API calls. - Draft cover email produced; nothing sent automatically. - README + Loom delivered; a non-engineer can add a second client from the docs alone. Timeline: 7-10 days. Budget: $700 fixed (test project; larger follow-on for the full 10-client rollout if this lands well). To apply, answer these three (generic proposals are skipped): Which DataForSEO endpoints have you used, and how did you handle its task-post/task-get async pattern and pricing? Have you pulled GSC and GA4 data programmatically? Which auth method (OAuth vs service account) and why?