LSI is showing that the most commonly used search term by users of our community is "20" - meaning our users are inputting the number 20 in the search function & pressing enter .
Does anyone have any idea why this might be the case? Is this a known issue with LSI?
I was looking at this today and our most common search is "1", which is just as strange. Apparently in the last two weeks 5% of our overall searches were this value (over 1200 searches or to put another way 90 a day).
Definitely appears to be some kind of error.
Just took a quick squizz at ours and nothing looks amiss for the past month or last few weeks. "breast cancer" and "bowel" are our top two, people searching for their usernames appear to be the top without results along with the letter a :rolling_on_the_floor_laughing:
Hope you can get it sorted!
This is possibly the result of a bot. LSI has bot filters that keep the metrics bot-free as much as possible but sometimes it fails and certain bots/crawlers cloaked as actual user accounts lead to search results like this. You may file a ticket with Khoros Support and get some more details on the bots, provided the details are exposed (such as their IP Address and geo-location etc.) but usually they aren't exposed. I wouldn't say this is an issue, but rather a rare case where bot/crawler metrics get mistaken as user metrics (because they are represented that way) and LSI cannot differentiate this. The bot Database does get updated so that LSI doesn't count metrics from these bots in the future but if something new comes up that's not already recorded in our Database, then it will show up in the metrics unless it gets detected and then recorded in our Database. Hope this helps.
This is becoming a bit of an issue for us since there are hardly any "real" searches displayed
Should the bot filter automatically pick this up and clean it or is it better to log a ticket with support?
Sorry about the delay here. I would highly recommend filing a Support ticket. Like I mentioned earlier, sometimes, some bots do slip by the filter if they were developed in a way to not get detected as a bot but rather get detected as actual human users. The engineers used to make a change in the dictionary of the bot filter and add the information of such bots there so the filter can detect that specific bot in the future. But it is no longer the process now.
This was discussed with Engineering and as the process of manually adding the details of the bot is not really a solution but rather a temporary relief, the team no longer manually updates it. Instead, when you create a ticket with Support and let us know about such bots that have bypassed the filter, we record the data for case studies so that we can come up with a permanent solution to this which will require some extensive development work. I hope this helps.
Update ( @Wendy_S :(
I was mistaken. I discussed this with Engineering again. We never actually made changes to the bot traffic in LSI by updating the DB to detect bots. Any metric that has been fetched by LSI once, will stay there for good. If you do come across suspicious traffic then you can file a support ticket and let us know about it. The Analytics team can provide you with details of where the bot traffic generated from (user agent details and IP address etc.) and you can block them from the community so no future traffic gets accepted from there.
Welcome to the Technology board!Curious about our platform? Looking to connect on social technology? You've come to the right place!