The behavior-robot filter in Yandex Metrica cannot be undone
Contents
Share
Yandex Metrica's help says the filter "Filter robots by behavior" deletes those robots from source data. You cannot undo it, and the robot display mode no longer switches after you turn it on. For an ordinary report the help recommends leaving this filter off.
Below is how to take behavior-based robots out of one report and keep the history, and the one case in which the help itself says to turn the filter on. This is a counter setting, not the robotness share. What that percentage means is covered in Robotness in Yandex Metrica.
What the statistics count by default
By default, Metrica statistics include visits from robots identified by behavior. Those visits sit inside visits, bounces and goal completions until you remove them from a specific report. There is no separate column that says "already removed": the report figure simply includes the visits.
You can remove them from one report with a segment. For a comparison, the help points to the mode "Data: without robots" and the action "Compare with data with robots". You see the report as it opened, then the two pictures side by side. The gap is the visits Metrica classed as behavior-based robots in that slice.
Where to look up robots by name in the Robots report is covered in How to find bots in Yandex Metrica. The behavior filter does not replace that table. The Robots report stays available after behavior filtering is on.
How to compare a report without erasing history
The order follows the same help page. It does not require a change to the counter filters.
Open the report for the period you already reconcile with ads.
Write down visits, bounces or the goal in the mode that opened on its own. Behavior-based robots are inside that number.
Turn on "Data: without robots" and the action "Compare with data with robots".
If you only need the robots out of this report, set a segment. Do not turn on the counter filter at this step.
A teaching example, not a benchmark for a campaign. Over a week the report shows 8,000 visits. In the "without robots" comparison, 6,400 remain. The gap of 1,600 visits is what Metrica classed as behavior-based robots in this slice. Those 1,600 are not Yandex Direct spend and not a share of the budget. They are the difference between two modes of one report. You can open the same period again tomorrow: a segment does not rewrite source data.
The segment lives in the report you built. The full picture of the period stays on the counter. You can repeat the comparison next week on a new period and on the old one.
What the counter filter does
The option in the filters is named "Filter robots by behavior". It deletes such robots from source data. You cannot undo the switch. The robot display mode no longer switches after that: the "with robots" and "without robots" comparison you used above cannot be rebuilt for the deleted visits.
The help recommends turning the filter on for the Logs API, when the export should come without behavior-based robots. In other cases it recommends leaving the filter off.
If behavior-robot filtering is already on, additional filtering in Metrica becomes unavailable. The Robots report remains. You can still open the report about robots, but you cannot add another counter filter on top, and the deleted visits do not come back into source data.
| What you turned on | What changes | Can you get the old picture back |
|---|---|---|
| A segment or "Data: without robots" | One report's picture | Yes, open the report again |
| "Filter robots by behavior" | The counter's source data | No undo |
| That option plus another counter filter | Additional filtering is unavailable | You cannot add a new filter on top |
How to see whether the switch is already on
Before you edit anything, open the counter filters and read the line "Filter robots by behavior". If it is missing or off, the history is still there: report modes still switch.
If the line is already on, there is no way back for the deleted visits. From then on, read what the counter collected after the switch, and read the Robots report. The old period cannot be restored in "with robots" mode for those visits. That changes what you do: do not hunt for an undo button in the settings. The help puts the decision at "do not turn it on", made beforehand.
Additional filtering is unavailable after the switch. A filter by address, or any other counter filter, has to be planned before this one. After it, the help leaves you the Robots report and the data written already without behavior-based robots.
If the export goes through the Logs API
The help names the Logs API as the reason to turn the filter on: the export then arrives without behavior-based robots. That decision is about the raw rows you pull for yourself, not about the chart in the interface. The Robots report in the interface remains after the switch, so the robot names Metrica shows separately do not go away.
The order for this case is different. Set the other counter filters you need first: after behavior filtering, additional filtering is unavailable. Then turn on "Filter robots by behavior", and only then pull the export. You will not be able to put the deleted visits back into source data, so save the period you still want to see "with robots" as a report comparison before you flip the switch.
If you do not need the Logs API, this section ends for you at the previous paragraph. Leave the filter off and compare modes inside the report.
Where this breaks
A segment and the "without robots" mode take visits out of the report picture. They do not say that Direct charged money for those visits, and they do not replace a placement review. The counter filter is even less of an ad-account fraud control: it changes what Metrica stores.
The Logs API is the case the help names as a reason to turn the filter on. If you pull raw visits and you need the export without behavior-based robots, turning it on matches the recommendation. If you read reports in the interface, the same help asks you to leave the filter alone.
Turning it on "so the chart looks smoother" breaks the comparison the chart was for. After the visits leave source data, you can no longer measure the share you removed with a second mode. A segment shows that share and leaves the data in place.
Write down three lines before you click the counter filter: the period you are comparing; the "as it opened" figure and the "without robots" figure; whether you need the Logs API without these visits. If the third line is empty, do not turn the filter on. The comparison inside the report answers the question for this period.