Just as how graphs and funnels count things differently, they also group results differently because they are measuring different things.
If a single user has completed an event multiple times, that user will be grouped into multiple different buckets in a graph. However, when the same user or event is analyzed in a funnel, Heap will only represent that unique user once, so it will be included in the bucket that occurred closest in time to the subsequent funnel event.
As an example in Heap, we’ll graph how many unique users clicked the Define Event button grouped by which page in Heap they were on when they clicked it.

The results for this query give us a total of 915 unique users bucketed into groups based on which page they were on when they clicked the button.

The same unique user can have clicked that button on multiple different pages, and thus, they will appear in multiple categories.
In comparison, let’s set up a funnel for users who clicked the Define Event and then went on to click Run Graph Query.

The results give us different, lower totals from the graph. This is because each unique user is only represented once, and is only included in the bucket that occurred closest in time to the subsequent funnel event.

For example, a user who may have clicked the Define Event button from the Events, Live, and Event Visualizer page over the past week would be counted in all three of those buckets in the graph. That same user would only be counted in one of those buckets in the funnel, thus lowering the count.
For more information about funnel analysis, see our Funnel analysis overview guide.
For more information about graph analysis, see our Graph analysis overview guide.