Explain Event Influence on Pipeline

Matches event attendees to accounts and opportunities, compares win rate, deal size and cycle time for engaged vs non-engaged accounts, and writes an honest narrative for leadership.

  • Measurement
  • advanced
  • CSV
  • Google Sheets
  • HubSpot
  • Salesforce
  • Slack

Example prompts

Once the skill is added, start a chat with one of these and attach your file if you have one.

  • Match everyone who attended our field events this year to Salesforce accounts and show influenced pipeline and win rate vs non-attendees.

  • Our CFO thinks events don't work. Write a one-page narrative from this data that's honest about what we can and can't prove.

  • Which accounts had 3 or more people at our events and an open opp? Rank by deal size.

What this skill does

Events reach buying groups, so giving credit to one contact on one campaign undercounts them. This skill matches every attendee, booth visitor and meeting across your events to the right account, by email domain first and normalized company name second, and reports the match rate so you know how complete the picture is.

It then classifies each opportunity as engaged or not within a stated window, compares win rate, deal size and sales cycle between the two groups and by segment, and writes a plain-language narrative that separates what the data shows from what it can't prove.

When to use it

  • Leadership asks "do our events actually work?"
  • Event credit is missing or untrusted in the CRM
  • Measuring buying group coverage across a year of events
  • Finding open deals with no event touch to invite next
  • Preparing a board or budget defense of the event program

How it works

  1. 1IntakeCollects engagement lists and an opportunity export, and confirms the window and which campaign member statuses count as engaged.
  2. 2MatchRemoves duplicates and competitors, matches people to accounts by domain then name, and reports the match rate by event.
  3. 3ClassifyMarks each opportunity as engaged or not using a 270-day default window, with an optional Python script that writes both tables as CSVs.
  4. 4CompareCompares win rate, average won deal and cycle time for engaged vs non-engaged accounts, then by segment, flagging small samples.
  5. 5ExplainWrites the short answer, accounts to act on, the event mix in won deals and a section on what the data can and can't prove.

What you'll need

  • Engagement lists across events (email, company, event, date, campaign member status)
  • Opportunity export with account, domain, amount, stage, created and close dates
  • Attribution window and engaged statuses, or approval to use the defaults

What you get

An attribution document with a short answer for leadership, what was measured, the engaged vs not engaged comparison, segment cuts, a ranked list of accounts to act on, the events that show up most in won deals, honest caveats and steps to strengthen next year's answer, plus CSVs of the account matches and opportunity classification.

Tips

  • Capture company domain on every registration form; it lifts the match rate more than anything else.
  • Log meetings as a campaign member status so they count as engagement.
  • Hold out a random slice of target accounts from invitations next year to test cause, not just association.
  • Agree the attribution window with finance before you run the numbers.

What's in the download

  • event-attribution-analysis/SKILL.md8.4 KB
  • event-attribution-analysis/assets/sample-event-engagement.csv3.0 KB
  • event-attribution-analysis/assets/sample-opportunities.csv2.5 KB
  • event-attribution-analysis/references/benchmarks.md1.1 KB
  • event-attribution-analysis/references/examples.md1.9 KB
  • event-attribution-analysis/references/matching-rules.md2.0 KB
  • event-attribution-analysis/references/narrative-guide.md2.0 KB
  • event-attribution-analysis/scripts/account_match.py8.5 KB
Read the SKILL.md
---
name: event-attribution-analysis
description: Analyzes how events influenced pipeline at the account level. Matches attendee, meeting and badge scan lists to accounts and opportunities, applies a stated attribution window, compares engaged and non-engaged accounts on win rate, deal size and cycle time, and writes a plain-language attribution narrative with honest caveats. Use when leadership asks whether events work, for influenced or multi-touch event attribution, buying group coverage, or when event credit is missing in the CRM.
license: MIT
metadata:
  author: eventmarketingsoftware.com
  version: "1.0"
  category: measurement
  source: https://eventmarketingsoftware.com/skills/event-attribution-analysis
---

# Event Attribution Analysis

Events reach buying groups, not single leads, so contact-level campaign credit undercounts them. This skill rebuilds the picture at the account level: who from each account engaged with which events, which opportunities those accounts had, and how engaged accounts compare with the rest. Then it writes a narrative that says what the data shows and what it can't prove.

Use event-roi-report for a single event's cost and return. Use this skill across a program of events, or when the CRM's campaign influence is missing or untrusted.

## Intake

If the user has an `event-context.md` file (CRM, campaign naming, engaged statuses, attribution window, segments), read it first.

Ask for whatever is missing in ONE message:

1. Engagement lists across events: email, company, event, event date, campaign member status. Badge scans, meeting lists, dinner RSVPs and webinar attendance all count.
2. Opportunity export: ID, account name, account website or domain, amount, stage, created date, close date. Optional: segment (size, region, new vs expansion), owner, primary campaign source.
3. Attribution window. Default: a touch counts if it happened no more than 9 months (270 days) before the opportunity was created, up to its close date (or today if still open).
4. Which statuses count as engaged. Default: Attended, Visited Booth, Meeting Held. Registered, Invited and No Show don't count.
5. The question being answered (for example "do events work?", "which accounts to prioritize", "why isn't event credit in the CRM") and the audience (CFO, CMO, sales).

## Connectors (all optional)

- If a Salesforce or HubSpot connector is available, read campaigns, campaign members, accounts, contacts and opportunities. Read only.
- If a Google Sheets connector is available, read or write the match tables there, after a preview.
- Otherwise, work from uploaded CSV/XLSX files or pasted tables and return Markdown tables plus CSVs of the account matches and opportunity classification.

## Method

```
- [ ] 1. Clean and dedupe engagement rows; remove competitors, vendors, students, employees
- [ ] 2. Match each person to an account; report the match rate
- [ ] 3. Build the account engagement table (people, events, first and last touch)
- [ ] 4. Classify opportunities: engaged in window or not
- [ ] 5. Compare engaged vs non-engaged accounts (and within segments)
- [ ] 6. Answer the user's specific question (rankings, lists)
- [ ] 7. Write the narrative with method and caveats
- [ ] 8. Validate
```

1. **Clean.** Remove exact duplicate rows (same email, event, status). Remove competitor, partner-employee, student and internal domains the user names. List every removal.
2. **Match** in this order: email domain to account domain; then normalized company name (drop Inc, LLC, Ltd, Co, punctuation, "the"; expand "Mfg" to "Manufacturing"; "&" = "and"); then manual review. Free-mail addresses (gmail, outlook, yahoo and similar) can only match by company name. Report: rows, removed, matched by domain, matched by name, unmatched. A match rate under 70% means results undercount influence; say so. Full rules in [references/matching-rules.md](references/matching-rules.md).
3. **Account table.** One row per account: engaged people, distinct events, first touch, last touch, events list.
4. **Classify opportunities.** Engaged = at least one engaged person from the account touched an event within the window: on or after (created date minus 270 days) and on or before close date (or today if open). Everything else counts as not engaged. This is influenced pipeline as used across this library; sourced pipeline needs a primary campaign source and is covered in event-roi-report. Never add sourced and influenced together.
5. **Compare.** For engaged vs not engaged opportunities: count, pipeline, closed won, closed lost, win rate (won / (won + lost)), average won deal size, average won cycle (created to close, days). Then repeat by segment if available (size band, region, new vs expansion), because engaged accounts are often bigger or further along. Flag any group with fewer than 10 closed deals as directional.
6. **Specific asks.** Typical ones:
   - Buying group coverage: accounts with 3 or more engaged people, with open opportunity amount, ranked by deal size.
   - Accounts with open opportunities and no event touch (who to invite next).
   - Event mix: which events show up most in won deals (count of won opportunities touched per event; not a credit split).
   - Missing CRM credit: opportunities that are engaged here but have no event campaign influence in the CRM. List them; don't fix them.
7. **Narrative.** Use the structure and wording rules in [references/narrative-guide.md](references/narrative-guide.md). Correlation isn't causation: engaged accounts may differ in size, intent or sales attention. Say what would make the case stronger (a holdout group, consistent statuses, pre-event baselines).

Optional script: `scripts/account_match.py` does steps 1 to 5 deterministically and writes `account_matches.csv` and `opportunity_influence.csv`:

```
python3 scripts/account_match.py engagement.csv opportunities.csv --as-of 2026-10-18 \
    --exclude-domain competitor.example.net --lookback-days 270
```

If code execution isn't available, do the same steps in chat, show the match table, and recount totals before writing the narrative.

## Output template

```
# Event Influence on Pipeline: [program or period]

## The short answer
3 to 5 sentences in plain words, with the key numbers and the main caveat.

## What we measured
Events, period, engaged statuses, window, match rate.

## Engaged vs not engaged
## By segment (if available)
## Accounts to act on
## Events that show up in won deals
## What this can and can't prove
## How to make next year's answer stronger
## Appendix: account match table, opportunity classification
```

Table columns:

- **Match rate:** Source rows | Removed | Matched by domain | Matched by name | Unmatched | Match rate
- **Comparison:** Group | Opps | Pipeline $ | Closed won | Closed lost | Win rate | Avg won deal $ | Avg won cycle (days) | Note
- **Accounts to act on:** Account | Engaged people | Events | Last touch | Open opp $ | Stage | Suggested next step
- **Event mix:** Event | Engaged accounts | Opps touched | Won opps touched | Won $ touched
- **Opportunity classification:** Opp ID | Account | Amount | Stage | Created | Close | Class | Engaged people in window | Events

Worked example: [references/examples.md](references/examples.md). Context statistics with sources: [references/benchmarks.md](references/benchmarks.md).

## Validation checklist

- [ ] Engagement rows in = matched + unmatched + removed
- [ ] Every opportunity appears once in the classification table
- [ ] Win rates use closed deals only; open deals aren't in the denominator
- [ ] Window, engaged statuses and as-of date are stated
- [ ] Small groups flagged as directional
- [ ] The narrative uses "influenced" or "associated with", never "caused" or "generated"
- [ ] Every number in the narrative appears in a table

## Approval gates

- Never create campaign members, change primary campaign source or edit opportunities in the CRM. If the user wants missing credit fixed, produce an import-ready list and stop for an explicit yes, then hand it to their admin or their CRM import process.
- Show any Slack post or shared-sheet write in full and wait for approval. Default to drafts.

## Privacy

Engagement lists hold personal data. Keep the work in the user's tools. Report at account level; include person names only in the appendix and only if the user wants them. Don't use the match results to add EU or UK contacts to marketing lists without a recorded opt-in.

## Related skills

event-roi-report (single-event cost and return), event-portfolio-scorer (use the event mix to pick next year's events), event-debrief.

FAQ

How do I measure event influence on pipeline?
Match everyone who engaged with your events to their account, then mark each opportunity as influenced if someone from that account engaged within a set window before it was created and before it closed. Compare win rate, deal size and cycle time for influenced and non-influenced deals, and show how many deals sit behind each number.
Why do events get so little credit in the CRM?
Most CRM attribution credits individual contacts on campaigns, but events reach several people at an account, often people who aren't contact roles on the deal. Badge scans with personal emails and unlogged meetings also fail to connect. Matching at the account level by email domain recovers much of that missing influence.
Can this prove that events caused more deals?
No, and the skill says so. Engaged accounts may already be bigger, warmer or getting more sales attention. It shows association and explains the limits. To test cause, hold back a random group of target accounts from invitations and compare their results with invited accounts over the same period.
What attribution window should I use for events?
Use a window that fits your sales cycle. The default here counts a touch up to 270 days before an opportunity was created and any time before it closed. Shorter cycles can use 180 days and very long ones 365. Keep one window for every event so comparisons are fair.