msn.com showed up twice
msn.com ran about 3.7 percent CTR on the July 2026 learning flight and 5.35 percent on Supply Bench. The host repeated. The creative did not have to.
Published 2026-09-09 · Search job: repeat bad inventory domains
Repeat bad inventory domains show up because open-exchange leftovers are a stable set of hosts, not a new random draw every flight. msn.com is the host we can name twice from our own books. The July 2026 learning flight recorded about 3.7 percent CTR on msn.com against a flight average near 0.15 percent. The 2026-09-02 Supply Bench pull recorded 785 impressions and 42 clicks, a 5.35 percent CTR, against a flight average near 0.28 percent.
The creative changed. The offer changed. The host and the shape did not.
Two windows, one host
July 2026 was the learning flight we first read on clicks and CTR. That readout later had to be thrown out. The placement scan from that window is still useful as a repeat flag: msn.com was already loud against a 0.15 percent average.
Supply Bench is the later, grain-complete pull for campaigns 4791 through 4810. msn.com sits in the seven-domain set that produced 41.3 percent of clicks from 4.7 percent of impressions. 785 impressions is not a one-row fluke. 42 clicks at 5.35 percent is more than eighteen times the later flight average.
We keep both windows because a single flight can be luck. Two flights, months apart, with the same host in the same direction, is how we decide a domain belongs on a watchlist instead of in a one-off footnote. The public list is Click contamination watchlist.
What "bad" means on this page
Bad here means implausible click density on remnant display, repeated. It does not mean a certified IVT verdict. We did not have a vendor add-on on either flight. invalid_impressions of zero is a measurement gap.
msn.com is also a large consumer host. A real person can click a 300x250 there. That is why we will not write "msn is fraud." We will write that 3.7 percent and then 5.35 percent, against 0.15 percent and 0.28 percent flight averages, is the wrong shape for leftover inventory we are willing to keep buying without a visit meter.
Click-to-page-view is the meter that would settle the argument. Anonymized main-seat brand-name display flights sat at 15.85 percent when usable and 3.75 to 6.68 percent when not. Until that join exists for these flights, concentration plus repeat is the claim.
What we do after a repeat
Scan, then exclude at a grain the seat can actually compile. Placement and site are direct targeting keys. Domain exclusions still need seat-level reference lists, which is why a domain finding can sit in the report and still be an honest product limit. See Domain waste after we filter exchanges and Placement vs site vs domain.
Do not use the higher CTR as a reason to buy more of the host. CTR is a trap. The rest of the seven-domain table is Seven domains produced two fifths of the clicks.