How to stop ad impressions by IP in Direct
Contents
Share
The ban on impressions by IP is turned on in the additional settings of a Yandex Direct campaign. In the campaign parameters, the block “Disable impressions by IP address” allows up to 25 addresses. The help page places this block among ad experiments and says that excluding addresses can make the campaign work inefficiently. Twenty-five is the limit of one campaign’s field, not the size of an account blacklist.

On the same page the help does not recommend an IP ban even when activity looks suspicious, and it points to Direct’s own fraud protection. Below is which addresses the field will not take, why one IP is not one person, and where the help page actually tells you to look up addresses.
What the field will not take
Private subnets cannot be banned. The help examples are 10.x.x.x and 192.168.x.x. These are not “suspicious ranges”. They are addresses Direct will not put in this field. Filling the block with an office’s internal address and expecting impressions to that office to stop does not follow from the page: a private subnet is not accepted.
A dynamic address can change for the same person. The address you entered in the morning can belong to another device by evening, and the person from the morning click receives a new one. The ban stays on the old line. A provider or company proxy gives one address to many people. Turning it off turns the impression off for everyone who exits through that address, not for one visit.
That is why the help page does not recommend an IP ban even when the activity looks suspicious. The recommendation does not depend on whether you found the address in a report or in an email. Direct’s own fraud protection is what that page names instead of the list.
Where an address comes from at all
The help page suggests looking up addresses in the server logs, and marking visits from Direct with your own link parameter. A server log and a Metrica grouping are not the same line. How to read a cut by IP and provider is covered in how to check traffic by IP and provider.
The grouping ym:s:ipAddress returns IPv4 and IPv6 with the last byte and the last 8 bytes zeroed. The help examples are 77.88.21.0 and 2001:db8:85a3::. That illustrates the zeroing. It is not an observation from a campaign. The last IPv4 byte in such a line is already zero. Copying 77.88.21.0 into the ban block enters the truncated address from the grouping example, not “the” device address. If a full address is needed for an experiment, the ban page sends you to the server logs.
Your own link parameter marks a visit from Direct separately from the IP. The impression ban does not replace it. The parameter is visible on the visit. The “Disable impressions” block decides whether to show the ad to an address before the visit.
Twenty-five lines and an inefficient campaign
The block lives in the campaign parameters, not in a shared account bin. The twenty-fifth address is the last line of this field. The page does not promise to accept a twenty-sixth in the same campaign. Moving a “competitor list” from a spreadsheet into this field and closing every campaign in the account with it does not match the page: the limit is described for a campaign, and the help page does not advise this as protection in the first place.
Excluding addresses can make the campaign work inefficiently. The page does not spell out how far the click price will rise. It ties the block to experiments on ads: remove a few addresses so they do not interfere with checking an ad, rather than keep a standing blacklist. A dynamic address and a shared proxy are exactly what makes a standing list crooked. One person leaves the address you banned and stops seeing the ad. Dozens of people stay on the proxy address and stop seeing it with that person.
Suspicious activity does not cancel this. The help page says it outright: even in that case an IP ban is not recommended. Look up addresses in the server logs. Mark Direct visits with your own parameter. Cut invalid impressions with Direct’s fraud protection, not with twenty-five lines.
| Question | What the help page says |
|---|---|
| Where it is turned on | additional campaign settings, a block of up to 25 addresses |
| What cannot be entered | private subnets, examples 10.x.x.x and 192.168.x.x |
| What happens to the campaign | exclusion can make it work inefficiently |
| What to do about suspicious activity | the ban is not recommended; fraud protection exists |
| Where to take the address | server logs; mark a Direct visit with your own parameter |