Comp research is a rabbit hole and I lose every link

Comp research breaks because saving is cheap and recall is expensive. You open Levels.fyi, two Blind threads, a Substack from a recruiter you respect, three Reddit comments, and a Glassdoor page on a Tuesday night, star them all in Chrome, and by offer day Friday you cannot find the one that listed L5 base for Series B fintechs.

The scene is always the same. It is Friday at 9:14 a.m. The recruiter says base is $198k with 0.18 percent equity over four years and a $25k sign-on. You smile, you ask for the day, you hang up. Then you open the laptop and start hunting. Your Chrome bookmark bar has three folders called comp, comp final, and comp REAL. You scroll Blind on your phone. You search "L5 fintech base" in Gmail because you forwarded one of the threads to yourself last week. The clock keeps moving. By noon you are no closer to the number you saw on Tuesday than you were when the call ended, and the recruiter wants a response before 5.

Why comp research punishes savers

Comp research is a forum problem wearing a database costume. Levels.fyi gives you medians by company and level, but the comment that explains why a specific Series B is offering 30 percent over band lives in a Blind thread from six weeks ago with 47 replies. You read it once on a phone, screenshot two of the replies, bookmark the URL, and forward the page to yourself with the subject "READ". Five surfaces, one signal, zero retrieval path. The same pattern shows up in the browser bookmark graveyard: starring a tab is a vote that the page might matter, not a way back to the sentence that did. By Friday morning you have 60 bookmarks and you cannot remember which 60 they were.

The Tuesday-to-Friday gap

The gap between Tuesday research and Friday recall is where the work disappears. On Tuesday you read a Blind comment that said: "if they are under 50 engineers and post Series B, expect base in the 180 to 205 range with a real refresh." You felt the number. You did not write it down because you bookmarked the thread, and you trusted that bookmarking the thread was enough. On Friday the recruiter says $198k and you are trying to remember if that comment was about a 40-person company or a 90-person company, because the gap matters for how aggressive your counter can be. Levels.fyi will not tell you that. Glassdoor will not tell you that. The signal is in the thread, and the thread is one of 47 you skimmed in three nights. Recall, not storage, is the problem you actually have. The hard drive fallacy names it bluntly: saving and remembering use different muscles.

What recall actually looks like

What you want on Friday is not the bookmark. It is the sentence. You want to type "what was the comp range that Blind comment listed for Series B fintechs under 50 engineers" and get the comment back with the URL attached so you can re-read the context. That is a different shape of tool than a folder. A Notion comp tracker is a good idea on Sunday and a bad idea on Wednesday at 11 p.m. when you have read your fifth thread of the night and the friction of opening Notion, finding the right database, and typing in three fields is the friction that made you bookmark the tab instead. The save surface has to live where you read. For most US comp research in 2026, that is Chrome.

You also want search by meaning, not keyword, because Friday's question is "what did people say about counter strategy at this size" and Tuesday's clip did not contain the word "counter" anywhere. It contained "push back" and "they bent" and a story about an engineer who waited two extra days and pulled in an extra 12k.

How recall-first capture fits the scene

The shape that works for comp research is recall-first: capture wherever you are reading, then ask in your own words later. dEssence is one option for this. You clip the Blind thread with the Chrome extension, forward the recruiter Substack from your phone to the Telegram bot, or paste the Levels.fyi row into the web app at dessence.ai. On Friday you ask in your own words, "what did Blind say about counter strategy at Series B fintechs under 50 engineers", and the thread comes back with the URL. No folders, no tags, no organizing. Save it, forget it, ask for it later. It is memory you don't have to maintain.

Honest about dEssence

A few things worth saying before you lean on it for a job search. dEssence is in beta. There is no native iOS or Android app yet, so capture from your phone goes through the Telegram bot or the mobile web at dessence.ai, not a system share sheet. The free tier caps how much you can archive, which matters if you have been hoarding sources for six months. There are no team or shared-collection features, so you cannot hand a comp folder to your partner or your career coach. The paid tier is not finalized, which means the price of keeping a multi-year comp archive there is still unknown.

If your comp workflow is already alive in Notion and you actually open it, keep using Notion. The point of recall-first capture is to remove a step you were skipping, not to add a tool you will forget about.

Frequently Asked Questions

Q: How do I save a Blind thread without losing the context of the replies? Clip the whole thread URL, not a screenshot of one reply. The thread page already contains the comments inline, so the saved version preserves the conversation. With a Chrome extension that captures the open tab in one click, the page you read is the page you get back. Screenshots strip the URL and freeze the thread before the late replies that often have the real numbers.

Q: What is the right way to track comp numbers from Levels.fyi? Save the row, not the number. A Levels.fyi row has company, level, base, equity, bonus, location, and the date of the submission. If you copy only the base into a spreadsheet, you lose the qualifiers that matter on offer day. Save the URL of the filtered view you were looking at when the number jumped out.

Q: Can I search saved comp research by meaning instead of by keyword? Yes, and that is the whole point of switching off bookmarks. Friday's question is rarely worded the same way as Tuesday's clip. You want to ask "what did people say about counter strategy" and get back a thread that used the words "push back" and "they bent". Recall by meaning solves the gap that Ctrl+F cannot.

Q: How do I keep comp data separate from other saved articles? You do not need to. The retrieval question carries the topic. Asking "what was the L5 base for Series B fintechs" pulls comp threads; asking "what did the recruiter Substack say about sign-ons" pulls the newsletter. The shape of the question filters the archive without any folder work.

Q: Is dEssence safe to use for sensitive job search material? Treat it as a research notebook, not a document vault. Use it for public sources like Blind, Levels.fyi, Reddit, and recruiter Substacks. Do not put your actual offer letters, signed paperwork, or anything covered by an NDA into a beta tool. Keep the legal documents in your own files.

Comp research is a recall problem in a storage costume. The fix is to save where you read, then ask in your own words later. dEssence is free during beta, no card, with the caveats above: no native mobile app, a free-tier archive cap, no team workspace. Worth a look if your Chrome bookmark bar already has a folder called comp REAL and another called comp REAL final.