Google Scholar shows today's h-index and i10-index, plus a bar chart of citations received each year. It does not show what the h-index was on an earlier date. CiteVector rebuilds that history from archived copies of the public profile in the Internet Archive's Wayback Machine.
Each time the Wayback Machine archived a Google Scholar profile page, it kept a copy of that page's metrics table: total citations, h-index and i10-index on that day. CiteVector asks the Wayback Machine's index for every archived URL of the profile (any hl= language, with or without view_op=list_works), keeps one snapshot per calendar month, fetches the raw archived HTML and reads the metrics table with the same parser it uses for live profiles.
The result is a dated series. Each point is a real value that appeared on the profile on that date, not a reconstruction. Hover a point in the dashboard to see the figures and open the archived page it came from.
The history is only as dense as the archive. Some profiles were captured every month for years; others were captured rarely or never. Two things help:
Snapshots whose metrics table cannot be read are listed but not plotted.
The Wayback series gives the h-index Google Scholar actually displayed on each archived date. That is different from a modelled h-index by year, which CiteVector also draws in the projection charts by applying a citation model to each paper. The modelled values are estimates and are labelled as such. For the forward-looking version, see the h-index predictor.
Geoffrey Hinton's profile has archived copies going back to 2012: open the example dashboard.