Where do users come from? (write-up)

[lizconlan]

Multiple commits preserved to make it easier to reverse out some of my odder decisions and to give the reviewer the option of not having to read everything in one go (although it might be simpler just to look at the rendered pages on the branch view.

I have regrets, chiefly about having stopped including link titles on the inbound links page after the first section so now it feels inconsistent. I should probably fix this but would appreciate advice on whether to take out the few that are there or fill in the rest. Assuming that the drilldown info is remotely useful (again, separate commits so relatively easy to ditch).

I like the sector information on the referrers page but am really not convinced that it's correct (and very conscious that I just made it up). Happy to get rid of it if preferred - it could be that it was more useful as a working note in the margin to help me spot the patterns in the data than as additional cruft for readers to wade through.

Resisted the temptation to add the Ukrainian data (suddenly available this morning) at this stage as there's enough danger of never finishing this as it is. Perhaps a new ticket once this has been merged?

Apologies if I've not knocked all the bugs out of this yet but I've reached the stage where I kinda know how it works and am mostly reading what I expect to see 😞

Closes #4

[lizconlan]

PR title is the one I deliberately avoided giving to the page itself as it's in an uneasy place on the humour/"author appears not taking this seriously" boundary. Very hard to unthink though.

[garethrees]

Broadly I think we have most of the information that's interesting, but I don't feel that the top-level page makes it any easier to understand the situation.

• The Answer section doesn't point out that we're talking about 7-13% of traffic, which is an important framing point.
• We could go further in making the data easier to digest. +1 to the "Sector" columns; I think these are good because they mean you don't have to understand what each individual site is (but yeah, please do check that they're correct…). The interesting thing that we can do is compare across sites to find trends and differences.
• Remember that these aren't academic papers. We're trying to be as terse as possible with actionable information.

I've made further comments inline.

[lizconlan]

Bit of a braindump at the moment - rewrite in English 😐 :taxi:

[lizconlan]

I think that's covered all the inline comments. Possible room for improvement in the backlinks top section but I couldn't work out how to phrase the relationship between the sites and 2 (of 3) of the referral/backlink crossover domains. May not be immediately obvious to all readers, may or may not be important (and in at least one case is probably a blip because of the relatively low traffic/early stage of the site at the time the data samples were taken).

[lizconlan]

Intention is to squash most of these commits before rebase/merge BTW

[garethrees]

Worth @crowbot giving it a quick look over, but on the whole looks good to me.

[crowbot]

Looks good to me! :shipit: