Works with the AI Agents you already use
marketing attribution model comparator
Marketing teams often struggle to justify budget allocation because different attribution models tell conflicting stories.
$1.99
marketing attribution model comparator
Example session with this skill installed
Compare attribution for the touchpoints in E1 and conversions in E2 using the consent register E3. Include the spend from E4. Assume a 30-day lookback and 7-day half-life for time-decay.
- Read your context and instructions
- Compiled the marketing attribution model
The analysis of touchpoints (E1) and conversions (E2) is complete. The calculation successfully applied the 30-day lookback and 7-day time-decay half-life while honoring consent status from E3.
Marketing Attribution Model Comparison — Acme Corp, 2024-11-15
1. Scope, sources and assumptions
Sources include touchpoint export (E1), order logs (E2), consent register (E3), and channel spend (E4). Analysis assumes a 30-day lookback window and a 0.4/0.4 weight for position-based models. Results are observational allocations of recorded touchpoints, never incrementality.
2. Population, consent and exclusions
- Supplied records: 5,200
- In-scope for analysis: 4,850
- Excluded (Consent Denied): 250
- Excluded (No Record): 100
- Conversions without touchpoint: 45 (Value: $2,250.00)
3. Attribution by model
| Channel | Credited conversions | Credited value | Share of conversions | Rank |
|---|---|---|---|---|
| Paid Search | 1,240.00 | $62,000.00 | 25.5% | 1 |
| Social | 980.00 | $49,000.00 | 20.2% | 2 |
| 850.00 | $42,500.00 | 17.5% | 3 | |
| suppressed_channels_combined | 120.00 | $6,000.00 | 2.5% | n/a |
4. Model comparison and sensitivity
| Channel | Min share | Max share | Spread | Model sensitive | Rank spread |
|---|---|---|---|---|---|
| Social | 15.2% | 24.8% | 9.6% | Yes | 2 |
| Paid Search | 24.1% | 26.2% | 2.1% | No | 0 |
5. Path analysis
- Average touchpoints per conversion: 3.2
- Median days to conversion: 12
- Top Path: Social > Paid Search > Direct (12% of conversions)
6. Budget interpretation
| Channel | Cost (Last-Touch) | Cost (Linear) | Cost (Time-Decay) |
|---|---|---|---|
| Paid Search | $12.50 | $13.10 | $12.80 |
| Social | $22.00 | $18.50 | $19.20 |
Every ratio is observational and does not represent return on investment.
7. Limitations, warnings and items for human review
- Warning W04: Small sample size in "Display" channel led to suppression.
- Proposed Action: Reallocate 10% of Social budget to Search. Management decision required.
This analysis supports, and does not replace, professional judgment and management's decision.
Next steps
- Review the model sensitivity for Social to determine if high-funnel credit justifies current spend.
- Validate the observational cost-per-conversion against internal targets.
- Obtain management approval before adjusting any channel budgets.
Connects securely to your tools. The creator never sees your data.
About this skill
Marketing teams often struggle to justify budget allocation because different attribution models tell conflicting stories. This skill provides a deterministic comparison of how channel credit shifts across last-touch, first-touch, linear, time-decay, and position-based models using your actual touchpoint data.
What it does
- Multi-model comparison generates side-by-side views of five standard attribution models for the same conversion set.
- Privacy-first processing excludes records without explicit consent and suppresses small groups to prevent re-identification.
- Lookback sensitivity tests how channel credit fluctuates when changing the lookback window from 7 to 90 days.
- Path analysis summarizes common multi-channel touchpoint sequences and calculates median days to conversion.
- Budget interpretation calculates cost per credited conversion by channel when spend data is provided.
How it works
- Data ingestion reads your pseudonymous touchpoint exports, conversion logs, and consent registers.
- Request generation builds a validated
request.jsonfile following a strict schema to ensure data integrity. - Deterministic calculation executes a Python script to perform exact decimal arithmetic for all models.
- Result synthesis interprets the output envelope to generate a structured memorandum with warnings and population counts.
Frameworks & tools
The skill utilizes a self-contained Python calculation engine (scripts/run.py) and JSON schema validation for input/output integrity.
Why this beats prompting it yourself
Standard prompts often hallucinate attribution math or fail to handle the complex logic of lookback windows and time-decay half-lives. This skill uses a hard-coded arithmetic script to ensure 1:1 precision that an LLM cannot achieve through text generation alone.
Use cases
- Compare how a high-funnel social channel performs under first-touch versus last-touch rules.
- Identify which marketing channels are highly sensitive to model selection.
- Prepare a budget interpretation memo showing observational cost-per-conversion across models.
- Analyze the impact of lookback window duration on credited conversion volume.
Known limitations
Does not measure incrementality, lift, or causality. It cannot perform media-mix modeling or forecast future performance. Requires pseudonymous keys; refuses to process PII like emails or phone numbers.
Calculates observational credit only, not incrementality or lift. Requires pseudonymous customer keys. Limited to rule-based models; no media-mix modeling or predictive forecasting.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 2 days ago