Purpose of Sheets

Mental Model

Think of Sheets as a structured workspace where data meets AI. Unlike a regular spreadsheet that only stores numbers or text, Sheets combine rows, columns, and AI skills to help you enrich information and take action.

• Columns = inputs and outputs Each column represents either raw input (like name, email, domain) or an AI-generated output (like job title, lead score, personalized email).

• Rows = records Each row is one entity you’re working with — such as a lead, customer, ticket, or expense.

• Skills = enrichment logic Skills are the AI-powered functions that transform inputs into outputs. For example, a “Lead Scoring” skill might take name, email, and company and output a score and summary.

• Sheets = collect → enrich → act The role of Sheets is to capture structured data, run AI on it, and produce outputs you can review, approve, or push into other systems (like Salesforce, HubSpot, or a cadence tool).

Best Practices for Using Sheets

Sheets are most effective when you treat them as structured AI workflows rather than static spreadsheets. These practices help ensure accurate, scalable results:

Best Practice	Why It Matters	Example
Validate on small samples first	Running Skills on a few rows helps confirm prompts and logic before scaling to the whole sheet.	Test lead scoring on 10 rows before processing 1,000.
Keep inputs consistent and normalized	AI performs best with clean, standardized inputs. Avoid free-text clutter or mixed formats.	Use domain.com instead of www.domain.com or pasted LinkedIn URLs.
Version your Skills with clear names	Label updates clearly so you can track iterations and avoid overwriting good logic.	“Lead Scoring v2 — tightened ICP criteria.”
Log changes to columns and formulas	Transparency makes it easier to audit results and troubleshoot when outputs shift.	Add a column noting when a scoring prompt or enrichment column was updated.
Document mappings in the Sheet description	Clear documentation ensures teammates know what each column and Skill is for.	In description: “Column E = Lead Score; Column F = Notes; pushed to Salesforce when >7.”
Review and refine iteratively	Skills improve with feedback. Adjust prompts or mappings based on outputs.	Tweak a cold email Skill until tone matches brand voice.