This is the feedback tool link authority data as a file. Seventy domains in this category, one row each, carrying each domain's harmonic centrality position in the newest published crawl, the same position a year earlier, the change between them, and how many distinct domains link to it now.
The source is Common Crawl's own domain level hyperlink graph, which is published free at data.commoncrawl.org. Nothing here was bought and no ranking tool was used. Lower is better, so position 4,916 means four thousand nine hundred and fifteen domains on the crawled web are more central than that one.
domain,harmonic_2026_jun_jul_aug,pagerank_2026_jun_jul_aug,harmonic_2025_aug_sep_oct,change_in_harmonic,referring_domains_newest
uservoice.com,1796,1636,2632,836,6451
canny.io,4916,4011,5449,533,3089
frill.co,18471,25939,386420,367949,211
votefirst.app,44127014,16688064,,,5
Take the file. It may be copied, quoted and used commercially with no attribution and no link back.
What is feedback tool link authority measured from here?
Common Crawl's hyperlink graph, seven releases of it, streamed rather than stored.
Each release publishes a ranks file and a vertices file covering every domain the crawl saw, and an edges file naming every link between two domains.
The edges archive for the newest release is 9.4 gigabytes, so it was streamed through a filter that kept only the rows naming one of the domains in this category and discarded the rest as it passed.
Two numbers come out of the ranks file. Harmonic centrality is a measure of how close a domain sits to everything else on the graph, and PageRank is the familiar one. Both are published as positions rather than scores, so the smallest number is the strongest domain.
The third number is counted rather than published. The referring domains column is the number of distinct domains the newest edge stream shows pointing at that domain, and 51,266 such rows were kept for the seventy targets.
Which feedback tool has the most link authority?
UserVoice, at harmonic position 1,796 with 6,451 domains linking to it.
Pendo follows at 4,453, Canny at 4,916, Productboard at 5,336 and UserGuiding at 7,575. Five of the six domains in this category inside the top ten thousand are those, with Aha! at 8,375 the sixth, and the distance between them and the rest of the field is the fact this dataset exists to show.
Their pricing pages are read separately, because being central is not the same as being expensive.
The median domain here sits at harmonic position 1,510,523 with 261 referring domains. So the gap between the category's strongest domain and its typical one is about three orders of magnitude, and a page that looks beatable on a domain that is not beatable is the mistake this table prevents.
This product's own domain sits at 44,127,014 with five referring domains. That figure is published here for the same reason every other one is.
Which domains gained and which lost over the year?
Twenty seven of the sixty seven measurable domains improved and forty got worse.
The largest single rise is quackback.io, up six million positions, and productlane.com and ducalis.io are close behind it. A rise of that size on a domain that starts near the bottom of the graph is a small absolute change dressed in a large number, which is worth saying before anybody quotes it.
The largest falls are harvestr.io, savio.io and suggested.co, each down between six and eight million positions. Frill is the clearest rise among the established domains, from 386,420 to 18,471.
Three domains have no row in the older release, being featureos.com, idealift.app and this product's own domain, so their change column is empty rather than zero.
One domain in the category, logchimp.dev, is absent from all seven releases and therefore absent from the file entirely, and a fetch of it on 15 September 2026 did not resolve at all.
How was this built, and when does it change?
By streaming seven published archives once, and it is rebuilt when Common Crawl publishes a new release.
The sweep is cc-graph.sh in this project's own research directory, which streams each release and keeps only the rows a filter file names, and authority-dataset.py, which joins the streams and writes the CSV.
The whole sweep took a little over two hours at about 3.2 megabytes a second. The file names and sizes were checked with a head request before anything was read, and all three archives returned 200.
The run was started once and restarted after its working directory was deleted by another process. The restart began every stage again from the first byte, so no figure here comes from a partial stream.
Common Crawl publishes a graph release roughly every quarter. The refresh rule is one sweep per release, and the next is due when the release after cc-main-2026-jun-jul-aug appears.
Who is this dataset for?
Anybody comparing this category who needs to know whether a domain is reachable before they spend a month trying to outrank it.
That includes people who are not going to buy anything here. Nobody will be asked to link to this page and no message will be sent about it. The sweep had to be run for this site's own planning, and a table of seventy domains is more useful in public than in a private file.
If you are shopping rather than researching, the prices for the same category are published as a second file, and the vendor pages themselves are grouped by what each pricing page does.
Three more files count the same sixty nine domains a different way, being every page each of them publishes, every heading at least two of them write and how much vocabulary their comparison pages share.
14 days, a card at signup, then $9 or $39 a month.
