# Apify Cost Reconciliation — Balloon Decor Ocean County Project

**Project:** Balloon Decor Ocean County — website SEO brief
**Budget cap (project):** $2.00 USD
**Budget cap (Apify FREE plan, monthly):** $5.00 USD
**Period:** 2026-09-22 to 2026-10-21 (monthly cycle)
**Project duration:** 2026-09-29 04:32 → 05:21 UTC (~50 min)

---

## Final spend

| Metric | Value |
|---|---|
| Project budget cap | $2.00 |
| Project spend (T1 + T5) | **$0.4404** |
| Project remaining | **$1.5596** |
| Project utilization | 22% of $2 cap |
| Apify FREE plan cap | $5.00 |
| Apify FREE plan spend | $1.2451 |
| Apify FREE plan remaining | $3.7549 |
| FREE plan utilization | 25% of $5 cap |

---

## Per-task breakdown

| Task | Apify actor | Spend | Notes |
|---|---|---|---|
| T1 (Competitor discovery) | `s-r/free-google-maps-scraper` (BicdJfvb7zWHUt0Ks) | $0.0752 | 3 production runs + 4 schema probes. 30 raw → 10 curated competitors |
| T2 (Site audit) | none | $0.00 | Pure web crawl (web_extract + curl) |
| T3 (Gap matrix) | none | $0.00 | Analysis-only on existing artifacts |
| T4 (Niche vocab) | none | $0.00 | web_extract + grep on existing audits |
| T5 (GBP benchmarks) | `compass/crawler-google-places` (nwua9Gu5YrADL7ZDj) | $0.3652 | 10 place details. Includes $0.20 romy smoke tests abandoned mid-run |
| T6 (Final brief) | none | $0.00 | Synthesis on 6 input artifacts |
| **TOTAL** | | **$0.4404** | **22% of project cap** |

---

## Methodology

- Per-run costs pulled from Apify REST API `/v2/users/me/limits` after each completed task.
- Subagent transcripts at `/home/ubuntu/.hermes/cache/delegation/live/deleg_*/task-*.log` include per-call pricing.
- Raw Apify run JSONs stored in `raw-apify-runs/` (12 files, ~1.5 MB total) for reproducibility.

## Findings

1. **Stayed 4.5x under the project budget cap.** $0.44 spent of $2.00 allowed. T2-T4 + T6 spent $0 because web tools were sufficient; only T1 and T5 actually needed Apify.
2. **Cheapest actor wins.** `s-r/free-google-maps-scraper` ($0.0025/place) was 8x cheaper than `compass/google-maps-extractor` ($0.020/place) for the T1 use case. Skipping premium actors on simple lookups matters.
3. **Smoke-test discipline worked.** T5 burned $0.20 on three `romy/google-maps-extractor` runs before switching to `compass/crawler-google-places`. In retrospect, validating the right actor up front would have saved ~$0.18. Updated `apify-actor-runner` skill with this finding.
4. **No over-spend incidents.** All sub-agents respected the per-task budget caps.

## Recommendations for future projects

- For T1-style competitor discovery: default to `s-r/free-google-maps-scraper` (cheapest, returns name + rating + reviews + website + phone + address).
- For T5-style GBP detail enrichment: default to `compass/crawler-google-places` (accepts ChIJ place IDs directly, returns photos count + posts + Q&A + claim status).
- Build a $0.20 smoke-test budget into any task that picks an unfamiliar Apify actor. Don't burn full runs before validating the schema.