G2 reviews to Power BI

Most review sources give you a star and a paragraph, and every question about that paragraph is a text search somebody keeps re-running. G2 asks four questions, and the feed keeps the answers apart: Title, Overall, Pros and Cons, one column each.

That single fact changes what the table is for. Cons is a field, so the share of reviews complaining about reporting is a measure, not an afternoon. Pros is a field, so the thing your happiest customers name first is a bar chart rather than a claim the sales deck has been making since 2024.

Alongside the four sit the ordinary ones: the rating, the date, the language, the link back, the reviewer’s name, and the tags, topics and subtopics Reviewflowz fills in as reviews arrive. One row per review, with column names a saved report can bind to.

And the reviewer is a person, not a handle. G2 verifies through a work email or LinkedIn before a review goes live, so the name on the row is real. The company behind it is not in the feed, and we say so further down rather than letting you find out in week three.

Ask a product team what buyers complain about on G2 and you get three anecdotes and one loud review from March. The Cons column turns that into arithmetic: sixty-one reviews out of four hundred and twelve name reporting, and each one is a sentence you can read beside the bar.

It works because G2 prompts for the complaint separately. Where the text is one blob, a filter for “reporting” also catches the reviews praising it, and the meeting becomes an argument about the query. Here the field only holds the criticism.

The same shape works on Pros. What your advocates volunteer first, unprompted by any survey you wrote, is a ranked list for whoever writes the next landing page.

Tags, topics and subtopics ride along as their own columns, filled in by Reviewflowz as reviews land, in whatever language they were written in. So a theme you already track and one you spot in the Cons text sit on the same visual, counted the same way.

G2 review volume is not a steady trickle. Most G2 reviews are solicited, so they land in campaign-shaped bursts: six flat days, then thirty-one in a week because somebody sent the request. Averaged over a month that texture disappears, which is why the monthly number keeps surprising people.

On a chart it stops being a surprise. Reviews per day with the campaign opening marked tells you what a push produced and, more usefully, what it produced that you did not want. A burst of four-star reviews naming the same missing feature is a roadmap item with a date on it.

A burst is also when reviews move. People edit them, and G2 takes some down. The feed carries a change stream of inserts, updates and deletes, so a review that disappears reaches your model as an event, not as a total that quietly stops adding up.

The change stream is a signal, not a diff. A row says that a review changed and when, never what it changed to, so a report that re-reads it by id can never publish a stale snapshot. A boring guarantee, and the one that matters at quarter end.