Extracting HCP Campaign KPIs

Given an HTML file with pharma HCP campaign data:

1. Parse the HTML and extract every KPI name, value, and the surrounding table/section it belongs to.
2. Classify each KPI into a category (e.g., Reach, Engagement, Conversion, ROI/Financial, Compliance).
3. Build one small table per category.
4. Draw the relationship chain across categories (what feeds what).
5. Present: category tables → relationship map → plain-English summary.

Progress:

• Step 1: Open and parse the HTML file (tables, lists, divs, spans — KPIs in pharma reports often sit in <table> elements or labeled <div>/<span> pairs)
• Step 2: Extract every KPI: name, value, unit, time period, and any footnote/definition
• Step 3: Deduplicate and normalize KPI names (e.g., "HCP Reach" vs "Reach (HCPs)" → same KPI)
• Step 4: Categorize KPIs into standard HCP campaign buckets (see below)
• Step 5: Build one table per category with columns: KPI | Value | Definition/Notes
• Step 6: Identify cross-category relationships (causal or sequential flow)
• Step 7: Produce a relationship summary (narrative + simple flow diagram in text)
• Step 8: Flag any KPI that doesn't fit a category or seems orphaned/unexplained

Standard KPI Categories for HCP Campaigns

Use these as default buckets unless the file suggests otherwise:

• Reach/Awareness: HCP Reach, Impressions, Unique HCPs Targeted, Frequency
• Engagement: Email Open Rate, Click-Through Rate, Detailing Visits, Call Attempts, Content Views, Webinar Attendance
• Conversion/Behavior: Prescription Lift, New Rx, Rx Volume, Formulary Adoption, Sample Redemption Rate
• Financial/ROI: Cost per HCP Reached, Cost per Engagement, Marketing ROI, Return on Investment, Cost per Rx
• Compliance/Quality: Opt-out Rate, Consent Rate, MLR (Medical-Legal-Regulatory) Approval Rate

If the file uses different terminology, still map each KPI into the closest matching bucket and note the mapping explicitly.

Finding Relationships

For each pair of category tables, ask: "Does a KPI in Table A causally or sequentially lead to a KPI in Table B?" Typical pharma HCP chains:

Reach → Engagement → Conversion (Rx) → Financial/ROI
                 ↘ Compliance (gates the funnel)

Document relationships as: [KPI A] influences [KPI B] because [reasoning from data/definitions in file].

Example 1: Input: HTML file with a table showing "HCP Reach: 5,000", "Email Open Rate: 32%", "New Rx Written: 420", "Marketing ROI: 3.2x"

Output:

Reach Table:
| KPI       | Value |
|-----------|-------|
| HCP Reach | 5,000 |

Engagement Table:
| KPI             | Value |
|-----------------|-------|
| Email Open Rate | 32%   |

Conversion Table:
| KPI            | Value |
|----------------|-------|
| New Rx Written | 420   |

ROI Table:
| KPI          | Value |
|--------------|-------|
| Marketing ROI| 3.2x  |

Relationships:
- HCP Reach → Email Open Rate: more HCPs reached increases the pool that can engage.
- Email Open Rate → New Rx Written: engaged HCPs are more likely to prescribe, suggesting engagement drives conversion.
- New Rx Written → Marketing ROI: prescriptions generated are the primary revenue driver behind ROI calculation.

Summary: The campaign funnel moves Reach → Engagement → Conversion → ROI, with each upstream KPI acting as a leading indicator for the next.

Example 2: Input: File contains only "Cost per Detailing Visit: $150" and "Formulary Adoption Rate: 18%" with no explicit link.

Output:

Financial Table: Cost per Detailing Visit ($150)
Conversion Table: Formulary Adoption Rate (18%)

Relationship: No direct KPI bridges these two (e.g., no "Detailing Visits per HCP" or "Visits leading to Adoption" metric present).
Flag: Relationship is inferred, not data-backed — recommend requesting a bridging KPI (e.g., visit frequency) to confirm causality.