ABM account engagement report

Purpose

Turn account-level ad data and pipeline data into one self-contained HTML dashboard for a client. Answer three questions for a VP Marketing or CRO audience:

1. Which accounts are progressing through the funnel?
2. Where do buying signals appear in more than one channel?
3. Which segments are engaging, and which are spending without response?

When to use

• A client needs an account-level dashboard that shows which accounts are progressing, where signals appear in more than one channel, and which segments engage.

Route elsewhere when:

• The question is planning the program. Use abm-strategy-planner.
• The question is the measurement method. Use abm-measurement-framework.
• The question is one account's background. Use account-research-brief.
• The question is a monthly recap or a diagnostic fix list. Use linkedin-ads-monthly-report or linkedin-ads-audit.

Inputs

• Rate accounts CSV (required, or pipeline data required). Expected columns: Company Name, Website, Industry, Company Size, Headquarters, Clicks, Impressions, Cost, CPC, Engagement Level, Conversions, Leads, Influenced Revenue. Any signal tool export that matches these fields works.
• Pipeline data (strongly recommended). One row per contact: company, contact, title, source channel, funnel stage (Lead, MQL, Opportunity), search keyword, campaign.
• Client context (ask first). Ask: "Which client is this for? A company name or URL helps me set the ICP." If context exists in context/client-*.md, use it. If the user declines, run without ICP framing and say so on the dashboard.
• Optional: search-term reports, CRM deal exports.

Procedure

1. Validate. Require a Company Name column. Blank numeric cells become 0. Strip currency symbols and whitespace. Read the file with BOM handling. Map "N/A" industry or size to "Unknown". Log every row that fails and its reason.
2. Engagement flag. Ignore the export's Engagement Level column when it is uniform (often every row says "Low"). Use actual behavior instead: clicks at or above 1 counts as Engaged. Note this on the dashboard.
3. Match pipeline to accounts. Exact case-insensitive match first, then partial match. List every unmatched pipeline company. Do not drop it silently.
4. Multi-channel. An account present in two or more pipeline sources is multi-channel. These are the strongest signals and lead the dashboard.
5. Funnel. Count unique accounts at each stage: Reached (every account in the export), Engaged (1+ clicks), Lead, MQL, Opportunity. An account at several stages counts once, at its highest stage. The stage counts must be non-increasing from top to bottom. If they are not, stop and report the error.
6. Score. Score = (clicks x 10) + (impressions / 100) + stage bonus, with bonuses Lead 25, MQL 50, Opportunity 100. These weights are the default. They are not benchmarks. Record the weights used at the top of the dashboard. Sort descending.
7. Segments. By industry: accounts, impressions, clicks, cost, CTR, pipeline accounts. By size band: same measures. Always show the number of accounts behind each figure. A segment with one or two accounts is not a pattern and must be labeled as such.
8. ICP alignment (only with client context). Classify each industry as ICP or non-ICP. Report CTR for each group, the CTR multiple, and the share of impressions going to non-ICP.
9. Heatmap. Clicks by industry and size, for accounts with at least one click.
10. Revenue, if present. Keep three measures separate and never sum them into one number:
    • Exposure: impressions on the account.
    • Influence: exposure that came before pipeline creation, within a stated window.
    • Attribution: a click or conversion linked to the pipeline event. Remove renewals before claiming influence.

Output

One self-contained HTML file named [Client]_ABM_Account_Engagement_Dashboard.html, with sections in this order:

1. Header: client, channels, account count, pipeline account count, date, and the scoring weights.
2. Funnel: six stat cards (Reached, Engaged, Leads, MQLs, Opportunities, Multi-Channel) and a log-scale funnel chart.
3. Active pipeline: multi-channel callout first, then pipeline cards grouped Opportunity, MQL, Lead, each with company, contact, source, keyword, and campaign.
4. Leaderboard: tabs for All, Pipeline Only, and Ad Engaged Only.
5. Segments: clicks by industry and size, impressions against clicks, CTR by size, with a one-line insight.
6. Heatmap, with a note on where engagement is thin.
7. Action plan: one card per stage (Opportunity, MQL, Lead, Ad Engaged), targeting optimization, and signal-tool levers. Name real accounts and keywords from the data. Do not write generic advice.

Chat summary, three to four sentences: funnel progression (accounts reached, in pipeline, at opportunity), the multi-channel finding, the most useful segment finding with its sample size, and one account-level insight.

Quality bar

• Funnel stages are non-increasing.
• Every unmatched pipeline company is listed.
• No unfilled template tokens remain.
• No em dashes or emoji in the output.
• Segment figures show account counts.
• Revenue is not summed across exposure, influence, and attribution.

Rules

• Do not present scoring weights as industry benchmarks.
• Do not use a single client's results to set a benchmark for others.