Deleting legacy (over one year old) unanswered posts?


(Gear Buzz) #1

I wonder if a cull of old, unanswered posts would give the remaining answered content more “Google Juice”.

Its hard to “interest and delight” visitors with questions unanswered and just left hanging.

A journey from google to a thread on exactly what you are looking for , only to find it simply your question - but unanswered… Is surely not very valuable. (i suppose you ucould register at the forum and then “bump” it)

I bet we have 10,000+ unanswered threads (perhaps more)

Please discuss?

(Richard Millington) #2

I’m doing what we always tend to do on these sorts of discussions and asking @dominic_woodman to help :slight_smile:

Is there an easy to delete or deindex them?

Might be an interesting feature for Discourse (@HAWK)

(Sarah Hawk) #3

It’s already easy to do. If you have console access you could do it that way. If you’re hosted it’s easy enough to pull that data using Data Explorer and then do a bulk deletion.

(Richard Millington) #4

Can you explain it in more detail? Easy enough is largely about how much experience we have doing this :slight_smile:

(Sarah Hawk) #5

It’s Discourse specific so more appropriate for Meta.

(Adrian Speyer) #6

No, unless you redirect specific and old unanswered posts via 301 (permenant redirect) to newer unaswered posts which are essentially similar.

If you are talking about user experience, you could consider to archive. At least for Vanilla, this means the content stays on the forum, so you can track pageviews from Google to see what is popular (and hopefully answer), but can be excluded from user search.

I would avoid deleting old content, unless it is truly outdated and not relevant to your company/niche.

(Gear Buzz) #7


(Mark Schwanke) #8


Unanswered as in no reply or simply no correct answer designation? If there are responses but no Correct or helpful answer designation and you delete the responses what will happen to the points for the people who responded. If there are replies and you simply delete them the people who responded may be pissed.

If there are multiple helpful answer designations on a given reply could you mark them as the accepted or correct solution using one of your moderators thereby moving them out of the category you mention.

Did you look to see if there are a top 100 of these that drive traffic and possibly registrations? By answering those top 100 you could change them into beneficial posts instead of just wiping them.

What percentage of these posts are from one hit wonders compared to people who are actively engaging in the site?

Can you archive them instead of deleting them? Some community tools let you archive after a certain date. I think it should go by last activity and not creation date. If by last activity you can hopefully allow old topics to still be alive. If by creation date then some people might get frustrated if a recently active discussion is turned off suddenly by a macro that disables the discussion simply based on the date it was created.

Have you asked some trusted members of the community their opinion? I’d ask before you have a fall-out from your community. They may see deleting content as censoring or something else…

Hope this helps…

(Gear Buzz) #9

Yes lots of possibilities

Amongst the admin team we feel a Google visit to an unanswered / un replied to post is a bad user experience.

I thin no indexing might be good

(Dominic Woodman) #10

Lots of good notes already here.

The scale of this issue matters, if it’s 1% of your site it’s probably not an issue. If it’s 40%, then you have more of an issue.

You tend to “deal with” these sorts of posts not because they’ll improve the link flow around your site, but because you’re trying to raise your sites general quality (i.e. you want your site to be viewed as high quality in Google’s eyes and not have issues with thin content, panda issues etc.)

(In theory if you were to redirect or canonical or them it could change your link flow, in practice that’s hard to measure, probably having a minimal effect and not the reason to do this.)

My general decision tree would be something like:

  1. Is this a high quality searchable question that isn’t answered anywhere else? I’ll try and get someone to answer it.
  2. Is it a high quality question, but answered elsewhere? I’ll try and consolidate it with the other question and get that one to rank.
  3. Is it a low quality question, on a unique topic? I’ll try and get the questioner to improve the question. Otherwise:
  4. It’s a low quality question, that can’t be improved or is on a topic that’s already covered. I’ll no-index it.

In practice depending on the size of the site, you often have to use automated thresholds to work with these things (i.e. less than 30 words, by a first time user etc.) and it’s very much a matter of experimentation.

(Gear Buzz) #11


What springs to mind is merging the question with an active thread.

However they may create a frankenstine non chronological addition mid thread or a non origional stay thread

But it’d enough to spark my imagination and get my minds donkey powered wooden cogs clunking round a bit longer on it.

We have elasticsearch and might be able to use that or its co-mention detection to identify thr best candidate threads fo merge with.

(Jason Hill) #12

Dominic’s response is excellent. It was the way I was thinking when I saw the initial post. Having thousands of unanswered threads suggests a significant problem that needs addressing,. From a community members point of view, if I make a post and never even get my post acknowledged, let alone answered, why would I ever return?

In terms of managing these posts, whether the “solution” is trying to surface them for community members to answer, providing an automated response, or archiving, I would like to see community platform vendors seriously tackle this problem. Take Lithium for example - one of the largest and most mature CM platforms out there, with customers who have massive, mature communities with millions of posts. Yet they provide nothing to help a CM with a problem like this. When you have a mature community, manual archiving of posts one by one is simply not feasible. Automated systems for handling old posts with configurable rules are desperately needed.

(Gear Buzz) #13

To be clear I mean years old single posts with zero responses.

We have had an “unanswered posts” area to appeal to members to answer but I don’t think it was ever a big success.

Auto No indexing sounds the best idea so far. (So after a few months…) we could calculate all average time a thread might drift off say, page 2 of our quietest sub forum and then no index.

We have over 1 million threads now.

(Dominic Woodman) #14

I’m not going to pretend I’m an expert on getting people to answer un-answered questions, it’s a very specific kind of person who picks on that sort of thing I imagine.

Depth is definitely one way to do it, the chronological nature of forums means that older posts tend to get less traction because they get buried deeper and deeper into the site.

If you’re running custom scrapes or processes to pull out bad questions, some possible ideas for criteria:

  • Use a service like TextGears to check for grammer errors
  • Anything un-answered after a time period
  • Length of post
  • Age of user
  • Number of edits to the original post

Particularly if you’re not getting organic traffic from any of these posts, you can often cut fairly deep without losing much.

(Mark Schwanke) #16

Jive doesn’t have an archive function that works well. One solution doesn’t work for everything. The criteria would have to be granular to the sub community level and only archive if the post meant certain criteria. But there would need to be something that hides unresponded to posts from search so you don’t show search results for content without replies while still suggesting other similar posts that do have responses.