Objective:
This assessment evaluates ETL platforms against LMNT’s actual volume profile and operational needs.
Note:
- Analysis covers revenue-related sources; SPINS estimates still pending.
- Excludes marketing, CX, support, finance systems, and other non-revenue sources.
Considerations:
- ware2go connector is missing in both recommended tools.
- Spins connector is missing in both recommended tools.
1. LMNT Data Ingestion Volume Overview
| Source | Est. Monthly Rows | Est. Annual Rows | Notes |
|---|---|---|---|
| Emerson | 250k–260k | ~3M | POS + Omni; store-item & traits refreshes total ~36M static rows |
| Shopify DTC | 700k–1M | 8M–12M | Based on 150–160k orders/month (estimate) |
| Shopify Wholesale | 20k–60k | 250k–700k | Directional |
| Amazon Marketplace | 1.5M–2M | 16M–20M | Based on ~300k orders/month (estimate) |
| Ware2Go | 360k–1M | 700k–1.4M | Shipping spikes up to 12k/day |
| Spins | Unknown | Unknown | Unknown |
Note: Shopify, Amazon, Wholesale, Ware2Go estimates based on order volume + typical connector schemas. Actual number may vary.
2. ETL Platform Comparison Matrix
| Criteria | Fivetran | Polytomic (ETL) | Polytomic (Reverse ETL) | Dagster |
|---|---|---|---|---|
| Scalability | High | High | High | Medium |
| Reliability | Very High | High | High | Variable |
| Maintenance | Very Low | Low | Low | High |
| Connector Coverage | Broad | Good | Marketing destinations | Custom |
| New Connector Speed | Slow | Fast (~1 week) | N/A | Engineering-built |
| Cost | High ($50 / 100k rows) | Low (~6M rows ≈ $500) | $400/connector | Low base + engineering |
| Best For | Shopify, Amazon, Stripe, Walmart POS | unsupported sources in Fivetran/High Volume for Fivetran | Klaviyo & Ads sync | Internal APIs |
3. Other Tools Considered
| Tool | Why Not Selected for LMNT |
|---|---|
| Airbyte | High maintenance; unstable for multi-million row retail datasets |
| Hevo Data | Limited track record with complex omnichannel retail loads |
| Matillion | Heavy ETL architecture; requires extensive engineering |
| Stitch / Talend | Deprecated or low investment in new connectors |
| Rudderstack | Strong CDP, weaker ETL ingestion for LMNT’s needs |
4. Recommended ETL Architecture
Primary ETL → Fivetran
Use for revenue-critical, daily ingestion:
-
Shopify DTC + Wholesale
-
Amazon
-
Stripe
-
Ads
-
Walmart POS + Omni
Why: Strongest reliability & connector quality.
Supporting ETL → Polytomic (ETL)
Use for high-volume or unsupported datasets:
-
Ware2Go
-
SPINS (future)
-
Any connector missing in Fivetran
Why: Cost-efficient, flexible, rapid connector turnaround.
Activation → Polytomic (Reverse ETL)
Use for syncing Snowflake →
-
Klaviyo
-
Meta Ads
-
Google Ads
-
GA4
-
CRM
Price: ~$400 per connector.
Custom Pipelines → Dagster (Optional)
For internal APIs or pipelines where pre-built connectors aren’t available.
5. Final Recommendation
Fivetran + Polytomic is the optimal ETL architecture for LMNT.