/q/ - Q&A Central

using that cimd json file to opt-in for scopes seems smart but i dont know if itll scale when u have tons of clients.
>com. posthog. provisioning is the opt-in for the provisioning API

"com.posthog" : {  "provisioning" : true,  "scopes" : [ "endpoint:read", "endpoint:write" ]}

im wondering how much overhead that adds if yous are managing a bunch of these endpoints.

more here: https://posthog.com/blog/embedded-analytics-provisioning-api

finally found someone else who hates storing every single one-time download forever. it's basically just using Cloud Storage as a temporary scratchpad with a redis ttl and then deleting everything once the job is done.
>Success is an empty bucket, not a searchable archive
it sounds good for something like a print relay but i'm not sure how it handles things if the instance dies before the cleanup happens. do you guys rely on a separate sweep process for that?

link: https://www.freecodecamp.org/news/how-to-keep-a-file-for-minutes-not-years-on-cloud-run/

im reading this piece about saas seo audits and it basically calls out how most agencies just spam lists of small stuff like broken links or missing meta descriptions that dont even matter much. they separate the real heavy lifting from the 'maintenance' junk which is way more important. specifically that whole debate about hosting blogs on a subdomain vs a subfolder. supposedly subfolder hosting is the only way to actually see results in traffic and rankings. also mentions how server-side rendering vs client-side rendering is one of those structural decisions that stays broken for years if you get it wrong. i dont know if i buy the subdomain thing being such a massive killer every single time but moving a whole blog later sounds like a nightmare for engineering.

more here: https://www.growandconvert.com/seo/saas-technical-seo/

it's kind of funny that the main way to deal with 1-star attacks now is just sitting around for a few days and hoping the algorithm catches it. sounds like they polished some detection logic in the last year or so but it doesn't really explain what happens if the attack is actually legit. i'm skeptical about how much this works when you're actually seeing real damage to your rating. has anyone actually seen an attack clear itself without a manual appeal?

article: https://localsearchforum.com/threads/coordinated-fake-review-attack-how-to-fix-review-bombing.63414/

ann handley is making a big deal about digging deeper than just checking ai boxes for seo. it's mostly about getting at the heart of the question instead of just optimizing for bots. i'm skeptical because it sounds like a lot of extra work when we're already drowning in content production. do you think this actually moves the needle or is it just more fluff?

link: https://annhandley.com/seo-that-gets-at-the-heart-of-the-question/

i'm tired of seeing people just spamming z-index: 99999 on everything. let's try an experiment this week where we only use local stacking contexts and see if we can actually fix those weird overlapping issues without breaking the whole layout. anyone else struggling with parent elements that have their own opacity or transform layers?

i was looking into this because our client lost access to their dashboard again. it turns out only owners can actually add or remove users from the profile. managers can basically only remove themselves which makes things difficult when you are trying to clean up an old agency account. you have to go through the people settings to find the right toggle. it is a pretty straightforward process but very easy to mess up if you do not have full ownership permissions. i always forget that managers cannot manage other users or delete the profile entirely. i am wondering how this handles it when an owner leaves a company and we are stuck with just manager access. it would be much better if there was a way to escalate permissions without needing the primary login credentials. it feels like a huge bottleneck for any dev working on local SEO.

https://www.sterlingsky.ca/how-to-add-remove-managers-from-your-google-business-profile/

Found this snippet from Mueller and Splitt about how people keep mixing up HTML and XML sitemaps. It's pretty clear that an HTML sitemap is just for users, not something you can actually submit to search engines like a real sitemap file.
>It's basically a. map of your website for users.
The lack of strict structure makes it useless for the direct processing we need. I've always just relied on the XML version for crawling, but if they're finding links through an HTML footer map, I guess it doesn't hurt to keep one. with crawl discovery if there are deep pages that aren't properly linked elsewhere. Not sure if it's worth the extra maintenance for a simple site though.

full read: https://www.searchenginejournal.com/google-explains-how-to-use-html-sitemaps-for-seo/592508/

i'm so tired of watching cursor just spin its wheels on the same bug until i run out of credits. it's always the same thing where you give a vague prompt like try again and then suddenly half the app is broken. reverting to a git checkpoint seems like the only way to stop the bleeding but it's such a pain when you're deep in a session. does anyone actually find success with that five step sequence or is it just more fluff?

article: https://www.freecodecamp.org/news/how-to-break-the-ai-coding-agent-fix-loop/

The idea that csat covers less than 10 percent of conversations is pretty wild. It's mostly just people who are super happy or super angry, so you're missing the silent majority. Using something like CX Score to scrape every single interaction for a 1-5 rating sounds good in theory, but I wonder how much noise that'll add to our actual metrics.

found this here: https://www.intercom.com/blog/how-to-measure-customer-experience-as-ai-scales