Card 1
Pipeline by stage
Deal funnel report
Question it answers: Where is the money sitting right now, and where is it stuck?
OPTIMIZE · Chapter 6
Every HubSpot implementation I have walked into has the same problem: 40-card dashboards that nobody looks at after week two. The fix is not a better dashboard tool. It is fewer cards mapped to fewer questions. Four is the right number for an SMB founder.
Card 1
Deal funnel report
Question it answers: Where is the money sitting right now, and where is it stuck?
Card 2
Contact created vs original source
Question it answers: Which channel is actually producing leads this month?
Card 3
Funnel report on lifecycle stage
Question it answers: Where in the funnel are conversions dropping versus last quarter?
Card 4
Time in deal stage report
Question it answers: Are deals moving faster or slower than they were 90 days ago?
HubSpot reporting tells you what is happening inside the funnel. It does not tell you which AI surface is sending the traffic. For the full picture — AI referrals from ChatGPT, Perplexity, Claude, and Google AI Overviews — pair this dashboard with the GA4 custom channel-group work in the AI Marketing Playbook's measurement chapter. HubSpot answers "is the pipeline healthy?" GA4 answers "where did the pipeline come from?" Both, weekly.
Two reasons. First, they have too many cards — 30+ charts no one knows the story behind. Second, they answer questions nobody is asking on a Monday morning. A four-card dashboard tied to four real founder questions outperforms a 40-card one in every implementation I have run.
Monday morning, by email, at 7 AM local time. HubSpot's scheduled dashboard delivery handles this natively. The founder reads it with coffee before the week's calls start. Daily is too noisy, weekly arriving Friday is too late to act on.
Out-of-the-box for the first 60 days. HubSpot's standard reports cover pipeline, source attribution, lifecycle funnel, and deal velocity well enough to start. Build custom reports only when you have caught the standard reports lying — usually because your data model has a quirk the defaults do not account for.
HubSpot's native source attribution does not break out ChatGPT, Perplexity, or Claude as separate channels. Two paths: tag inbound URLs with UTM parameters when promoting on AI surfaces, and pair HubSpot reporting with GA4 custom channel groups for AI referrals (covered in the AI Marketing Playbook's measurement chapter).
Breeze Copilot can pull data and summarize trends conversationally, which is useful for ad-hoc questions ('show me deals stuck more than 21 days'). The repeating four-card founder dashboard is still better built once and scheduled, rather than re-prompted weekly. Save Breeze for the questions you did not anticipate.