the gsc angle is the sneaky one. had a client where impressions jumped like 30% with zero extra clicks and we almost reported it as a win, turned out to be fanout noise not real demand. cant set targets off raw impressions anymore, gotta separate human queries from llm fan-out first
@MaxScore@manakoai vs pages are quietly the best geo play right now. had a client whose x vs y page got pulled into chatgpt answers way more than any product page we had. one thing we learned tho, if the comparison reads too one sided the models tend to skip it, the balanced ones get cited
@lilyraynyc been doing this too. the thing that caught us off guard is one keyword maps to like 4-5 different prompt phrasings and only some of them actually pull our client into the answer. so now we log which phrasing triggers a citation instead of assuming the rewrite is one to one
@CyrusShepard yeah we saw this on a client too, a category term spiked in trends but actual gsc clicks stayed flat. the fan-outs inflate the curve so youre chasing demand that isnt even human. been cross checking trends against gsc impressions before trusting any spike now
been pulling which pages chatgpt and perplexity actually cite for our clients. its almost never the google #1. its whoever answers the question cleanest in the first 2 lines.
ranking and getting cited are two different games now. anyone else seeing this split
Indonesia will allow police officers to legally hold civilian administrative positions, a move that could increase the role of security officials in government https://t.co/ICtt0wpn9q
Ada satu yayasan yang dalam waktu kurang dari setahun mengelola 1.179 dapur MBG di 38 provinsi. Satu yayasan, tanpa batas maksimal jumlah dapur yang boleh dikelola , sementara yayasan lain dibatasi maksimal 10.
Namanya Yayasan Kemala Bhayangkari. Strukturnya melekat langsung di hampir seluruh Polda dan Polres Indonesia , 419 kepengurusan dari pusat sampai cabang. Pucuk pimpinan di setiap wilayah otomatis diisi oleh istri Kapolda atau Kapolres setempat. Bukan dipilih. Otomatis.
Hitungan ICW: dari insentif BGN saja : Rp6 juta per hari per dapur, 313 hari operasional , perputaran dananya bisa menyentuh Rp2,2 triliun per tahun. Di luar dana awal Rp500 juta per dapur dan biaya operasional.
Ketika ICW coba menelusuri profil 419 kepengurusan yayasan ini di sistem AHU Kemenkumham, sebagian besar datanya tidak bisa ditampilkan. Yayasan yang mengelola triliunan rupiah uang program pemerintah, tapi profilnya tidak bisa diakses publik.
ICW sudah surati KPK sejak 24 Februari 2026 untuk minta pengawasan. Sampai hari ini belum ada tindak lanjut yang signifikan.
Program makan bergizi gratis harusnya soal anak-anak yang perlu makan.
Tapi kalau satu yayasan yang strukturnya melekat di institusi penegak hukum bisa dapat privilege tanpa batas, kelola triliunan rupiah, dan datanya tidak bisa diakses publik , siapa yang mengawasi?
@zodchiii being named isnt random though. the models lean on whoever has the clearest structured info and third party mentions. had a client go from never cited to the default rec in their niche just by fixing how they were described across the web, not their own site
@alexgroberman the verification loop framing is the cleanest take ive seen on this. the stage that breaks most for our clients is the middle one. AI cites them, buyer googles to verify, and a competitor outranks them on the branded comparison query. you win the citation and still lose at verify
@VadimStrizheus this gap is the part clients underestimate most. had one ranking #1 on google but chatgpt kept recommending 3 competitors instead. wasnt authority, their pages just never said who its for or how it compares so the model had nothing to cite
@Curlh1 nice one. worth checking youre in the AI answers too not just google. seen clients crush organic but go invisible in chatgpt/perplexity cause their pages had no comparison or use case language for the models to pull. quick thing to audit while youre hot
hot take, GEO is just SEO with a new name
all these "GEO experts" popping up teaching stuff we been doing for 10 years. structured content, entity clarity, being the source worth citing
real question, whats actually NEW in GEO that wasnt already good SEO. genuinely curious
@rustybrick the part that stands out to me is google screening for this at the hiring level. when the search engine itself is writing interview questions about generative ai optimization you know its not a fringe tactic anymore. kinda changes how we scope what an seo hire even does now
@alexgroberman the verification loop point is the part people sleep on. we stopped treating an ai citation as the win and started asking what the page does once the buyer lands to verify. cited but weak page just sends them back to google to find a competitor who closes better
@algomizercom this gap is real. we audit a clients llm visibility separate from their google rankings now cause theyre almost never the same list. had one b2b client ranking top 3 on google for their category and chatgpt didnt name them once. fixing it needed a totally different content play
@sukie234 if youre rebuilding from scratch just bake the structure in from day one. clear answers to real questions, clean schema, a solid about page so AI knows what you are. so much easier than retrofitting later. happy to share what works for us if you want
@askOkara good one. the real follow up, when you dont show up go find which sources the LLM is pulling from and earn a mention there. the check is easy, the fix is reputation work. thats the part everyone skips