Understanding Recommendations

Recommendations are actionable items generated from your audit findings. While AI insights explain what the data means, recommendations tell you exactly what to fix and in what order. They are designed to give you and your clients a clear improvement roadmap.

How Recommendations Are Generated

Recommendations are generated using a rule-based system. Each recommendation rule defines:

1. Conditions — What audit data triggers this recommendation
2. The recommendation itself — What to do about it
3. Impact and effort scores — How important it is and how much work it takes

Condition Operators

Each rule has one or more conditions that are evaluated against the audit data. Conditions use these operators:

Operator	Description	Example
greater_than	Value exceeds a threshold	Score > 80
less_than	Value is below a threshold	Percentage under 50
equals	Value matches exactly	Status equals “inactive”
contains	Value includes a substring	Name contains “test”
is_true	Boolean value is true	DKIM configured is true
is_false	Boolean value is false	SPF configured is false

A recommendation is triggered when all of its conditions are met. If a rule has three conditions, all three must evaluate to true for the recommendation to appear.

Recommendation Attributes

Each triggered recommendation includes:

Impact (1-5)

How much improving this item will affect the portal’s health, efficiency, or business outcomes:

Score	Meaning
5	Critical impact. Directly affects core functionality, compliance, or revenue tracking.
4	High impact. Significantly improves operational efficiency or data quality.
3	Moderate impact. Meaningful improvement in a specific area.
2	Low impact. Minor improvement or optimization.
1	Minimal impact. Polish or nice-to-have enhancement.

Effort (1-5)

How much work is required to implement the recommendation:

Score	Meaning
5	Major effort. Requires multi-day project, cross-team coordination, or significant technical work.
4	High effort. Several hours of focused work with some complexity.
3	Moderate effort. A few hours of straightforward work.
2	Low effort. Under an hour. Simple configuration or update.
1	Minimal effort. A few minutes. Single setting change or toggle.

Time Estimate

An estimated number of hours to complete the recommendation. This helps scope improvement projects and set expectations with clients.

Category

The functional area the recommendation belongs to (e.g., Email Deliverability, Pipeline Configuration, Data Hygiene, Workflow Optimization). Categories help group related recommendations for planning.

Priority Score

Recommendations are sorted by a priority score that combines impact and effort:

Priority Score = (Impact x 2) + (5 - Effort)

This formula favors high-impact, low-effort items. Examples:

Impact	Effort	Calculation	Priority Score
5	1	(5 x 2) + (5 - 1)	14
5	3	(5 x 2) + (5 - 3)	12
3	1	(3 x 2) + (5 - 1)	10
4	4	(4 x 2) + (5 - 4)	9
2	5	(2 x 2) + (5 - 5)	4

Recommendations are displayed in descending priority score order, so the most impactful and easiest fixes appear at the top.

Recommendations in Reports

The recommendations section of an audit report shows:

• Each recommendation with its description, impact, effort, time estimate, and category
• Sorted by priority score (highest first)
• Grouped or filterable by category

This gives clients a ready-made action list they can work through sequentially or hand off to their team.

Relationship to AI Insights

Recommendations and AI insights are complementary but distinct:

• AI insights are generated per-block and explain findings in natural language
• Recommendations are generated from rules across the entire audit and focus on specific actions

A single block insight might reference multiple data points. A recommendation targets a specific condition and action. Together, they give a complete picture: the insight explains the context, and the recommendation provides the next step.