Claude Code vs Clay (Honest Review) 1) Claude Code is the news. Yes, you ship fast custom workflows or automate operational bottlenecks with 2-3 hours of coding. Ex. I built an AI Agent that… | Michael Saruggia | 33 comments

Each stage of the B2B funnel has a different job. When one part breaks, everything after it suffers: Here's how to build a strong funnel in 2026. 1/ AWARENESS - Get in front of the right people Goal: Decision-makers at your target accounts know you exist. → Define your ICP (Ideal Customer Profile) tightly → Post consistently on LinkedIn → Contact target accounts directly Track: Reach among target roles, website traffic from ideal accounts. 2/ CONSIDERATION - Build trust before asking for anything Goal: Show buyers you understand their problems better than anyone else. → Share case studies with real numbers → Deliver value in outbound before asking for a meeting → Address objections before they come up Track: Repeat site visits, inbound inquiries, email replies. 3/ INTENT - Capture the people who are ready Goal: Make it easy for interested buyers to take the next step. → Make your offer specific and clear → Show proof with measurable outcomes → Be clear about who you're not a fit for Track: Consultation bookings, qualified meetings, show rate. 4/ CONVERSION - Close the deals in your pipeline Goal: Turn qualified pipeline into closed revenue. → Align what marketing promises with what sales delivers → Cut poor-fit prospects early → Review pipeline performance weekly Track: Close rate, average deal size, sales cycle length. 5/ LOYALTY - Grow what you've already built Goal: Increase lifetime value and generate referrals. → Report on outcomes, not just activity → Turn wins into case studies → Stay in contact with decision-makers year-round Track: Retention rate, expansion revenue, referral-driven deals. Fix the structure first. Then scale. Looking to generate attention and scale your business through outbound? Apply for our free pilot campaign. We'll book 3-4 ICP-fit meetings, no commitment. Find out more here: https://bit.ly/C17Pilot Repost to help others grow. Follow Enzo Carasso 🧲 for systems that hold up at scale. | 98 comments on LinkedIn

GTM tech that moves the needle in 2026 We’ve divided them across 13 categories 👇 1. Data → Get this wrong and everything downstream suffers. - AI Ark - ListKit - DiscoLike - saasyDB - BuiltWith 2. Scrapers → If the data isn't sitting in a clean format somewhere, you go and get it. - Apify - ZenRows - ExportApollo - Boomerang - Instant Data Scraper 3. Email Finder → We waterfall all four of these. One source is never enough. - Prospeo – Findymail - Datagma - LeadMagic 4. Email Verification → Validate before anything goes near a sequencer. Every time. - ZeroBounce - BounceBan (Verify risky emails without sending messages) 5. Intent Signals → Track the right signals and you'll know exactly when to reach out. - Clay - Common Room - Trigify.io - Champify - UserGems 💎 6. Website Visitors → Someone's already on your site looking around. These tell you who. - RB2B - Vector 👻 - Warmly, 7. AI and Automation → Build it once, let it run. - Make - n8n - OpenAI - Anthropic 8. Email Infrastructure → Land in spam and you've lost before you've started. - Smartlead - Instantly.ai - Saleshandy - Zapmail.ai 9. Spam Testing → Check your domain reputation regularly. - GlockApps - EmailGuard 10. LinkedIn Outreach → Email alone isn't enough. These handle the LinkedIn side cleanly. - GetSales.io - HeyReach.io - La Growth Machine 11. LinkedIn Branding → Your content should be doing some of the selling for you. - Taplio - Kleo - Supergrow 12. Meetings and Comms → Once the meeting's booked, show up prepared. These handle the rest so you can stay focused on the actual conversation. - Demodesk - Fathom.ai - Fireflies.ai 13. Content Designs → Assets, organisation, consistency. - Figma - Canva StackOptimise is a GTM Engineering agency. Today, we send ~2M emails and generate 3,000+ positive replies every month. Which of these are you already using? 👇 StackOptimise ⚙️ | 20 comments on LinkedIn

For the longest time while cold emailing local businesses, I removed “info”, “support”, and “help” @ [company name] emails. Wow, what a mistake that was. I thought that those were generic emails and nobody checks them. So, I’d of course favor actual people’s (generally the owner) emails. Except over time, I realized how small this made our list. It became clear that not all SMB business owners actually set up personal email accounts, and sometimes do operate out of the generic names (some friends in the space made this clear to me!). And, it just…worked (see below for confirmation). This changes a lot for us. It’s almost like we unlocked new TAM by confidently reaching out to addresses we never would have before, being that we now know they actually may be fruitful. All that to say: if you’re cold emailing local businesses, don’t shy away from the generic addresses. They might just be exactly what will work. | 40 comments on LinkedIn

Signals won’t magically make someone buy. But they 𝘸𝘪𝘭𝘭 make your outreach feel relevant, timely, and researched 👇 Here’s 10 we leverage daily at StackOptimise. 1. They just raised funding - A company that's just closed a round has cash to burn and targets to hit. - They're actively looking for ways to grow. Tools: Crunchbase, Coresignal, PredictLeads, CB Insights 2. They're hiring for relevant roles - Posting jobs for SDRs or a Head of Sales? They're investing in growth. - Extract extra insight from the job descriptions Tools: TheirStack, Clay, Leadsotters 3. A key decision-maker just started - New CROs and VPs want to make an impact fast. - Way more open to new approaches in their first 90 days. Tools: UserGems 💎, Common Room, Clay 4. They're using a competitor - If they're already paying for something similar, they get the problem you solve. - Even better if they've left a bad review somewhere. Tools: BuiltWith, HG Insights, TheirStack 5. They visited your website - Someone browsing your site is way warmer than a random cold lead. - Especially if they're hitting your pricing page. Tools: RB2B, Vector 👻, leadpipe.com 6. They engaged with your content - Liked your post. Commented. Downloaded a lead magnet. - These are soft signals showing they're already thinking about the problem. Tools: Trigify.io, Myteamfluence, ScrapeLi 7. They left a negative review for a competitor Someone frustrated enough to write a bad review? Already considering switching. Tools: G2, Capterra, Trustpilot (scrape with Apify) 8. A past champion moved companies Someone who bought from you before just started somewhere new. Warm intro waiting to happen. Tools: Champify, UserGems 💎, Common Room 9. They're running ads If they're spending money on ads, they've got budget. And they're actively trying to grow. Tools: Apify, Adyntel, Ahrefs 10. They look like your best customers Find companies similar to your top accounts. Reach out before your competitors do. Tools: Ocean.io, DiscoLike, Tamtam We use most of these signals across our client campaigns at StackOptimise. Stack a few together and your "cold" outreach starts feeling pretty warm. What signal's working best for you right now? 👇 StackOptimise ⚙️ | 27 comments on LinkedIn

Small Claude trick I've been using lately. If you have a bunch of skills set up in a project, Claude will usually figure out which one to use. But not always. A quick fix: add a trigger command to the skill itself. Something like [weekly-analytics] or [campaign-check]. Now it only fires when you type that specific command. I use brackets because Claude doesn't use them natively. Slash commands it already does, so there's potential for conflict. The other nice thing is you can chain them. I have a weekly analytics skill and a campaign optimization skill that I combined into a single [weekly-campaign-check] that runs both in one go. Nothing groundbreaking. Just makes managing a growing list of skills a bit cleaner. Walked through it in today's Instantly Academy video. Want to join the academy when it goes live? Let me know!

How to write a great cold email (in 5 steps): This got me 9% reply rates every month 1. Trigger Get straight into the reason you are sending them an email, make sure it is relevant to the issues you solve for and personal to the company the email is to, example: “Was reading through your FY24 annual report John and saw you are expanding to the US from the UK” 2. Implication Based on that key event, what could be a key focus for them, plant the seed of the issue or “Poke the bear” as Josh Braun says, example: “With the expansion, curious how you’re planning to manage currency conversion from GBP to USD?” 3. Pain: Now is when we get into the main issue and more importantly, what happens if they stay the same and don’t fix it, example” “Most companies use high street banks with 3% FX fees, meaning if you hit the revenue forecast from your report for FY25 that could be as much as $300,000 in fees. 4. Social Proof + Solution People don’t care about what you do, they care about what you’ve done for others. SO show how you helped people like them avoid the issue. “Deel paid 0.4% on average with our online bank & saved $50,000 in extra fees.” BONUS: The bigger the gap from the current state (their pain and what it leads to) to the desired state (where you’ve helped similar companies get to) the better! So use metrics or stories to help it resonate 5. Soft CTA Don’t dive straight into trying to book the meeting, instead try & start the conversation, tie it back to your value prop too to help it stand out. “If we could save you costs on transactions, would that be worth a chat?” This framework helped me: - Average 134% over 18 months - Self-source $1m+ in qualified pipeline - Teach 347+ SaaS sellers how to book more meetings every month Go use this today to book more meetings through cold email! ➕ Follow me for B2B SaaS cold outbound tips ♻ Share if you like this post | 37 comments on LinkedIn

People are taking spintax too far and it's ruining their cold emails. They're trying to spin individual words to increase variance, but this often leads to certain sentences just not making sense or sounding weird when read out loud. This is not the way. You clearly show your prospect that a human did not write the email. The right way to run spintax: 1. Identify the right variables to do it with in the first place. Certain ones are over the top and risk-prone for messing up the email. Especially with certain Clay workflows, people are going overboard. Easy, less risk-prone ones: - Case studies (switch out company name and result) - CTA sentence (spin 4-5 different ways of saying "Can I send you more info?") 2. Instead of spinning specific words, do it with specific phrases all together that still mean the same thing. This reduces the chance of the email sending platform creating a sentence that just looks weird or doesn't make sense. This is “boring” stuff but it's important. | 10 comments on LinkedIn

Cold emails not getting results? Here’s 7 ways to 10x your performance. 1/ Strengthen the Offer → Low reply rates are often a result of weak offers. → Lead with a measurable outcome tied to a real problem. → If your prospect can’t picture the result in 10 seconds, it’s not strong enough. 2/ Narrow Your ICP → Tighter targeting = cleaner replies. → Use revenue range, role, and intent signals. → Broad ICPs dilute your message before it lands. 3/ Improve Data Accuracy → Bad data kills deliverability before you start. → Verify emails and phone numbers before launching a campaign. → Clean data means you actually reach the right inbox. We use BetterContact, a data enrichment tool that cross-checks 20+ data sources to find and provide you with the most accurate emails and phone numbers. 4/ Write Direct, Specific Emails → One problem. One outcome. One next step. → If your email could apply to any company, it’s not strong enough. → No generic statements. Be specific about the pain you’re solving. 5/ Increase Speed to Respond → An MIT study found responding in 5 minutes vs. 30 minutes = 21x better qualification. → Slow response = lost deals. 6/ Qualify Before Booking → Not every reply deserves a meeting. → Qualify role, urgency, budget, and authority before you book. → Unqualified meetings tank your close rate. 7/ Build Infrastructure for Scale → Warm domains. Multi-touch sequences. Reply workflows. → If your system breaks when you increase volume, you can’t rely on it. → Performance improves when infrastructure can handle pressure. Ready to get results from your outbound efforts? Apply for our free pilot campaign: https://bit.ly/C17Pilot Repost for anyone running outbound. Follow Enzo Carasso 🧲 for outbound systems that work at volume. | 93 comments on LinkedIn

My secret outbound hack to get +7% reply rate. I call it the Signal Stacking Effect. Whenever you’re building a prospecting list, filter by intent first, not job title. It works because when you’re prospecting off real signals, you: - Reach buyers who are already in motion - Anchor your message to a real “why now” - 3–5x your reply rates without increasing volume 99% of reps start with: “VP of Sales, 200–1000 employees, US.” That’s not intent. That’s a demographic. Intent looks like: - Recently raised funding - Hiring for RevOps or SDRs - Following your competitor’s page - Just implemented HubSpot or Salesforce - Engaging with specific industry influencers That’s the difference between BS outbound and the Signal Effect The magic happens once we integrate Alta | AI Revenue Workforce as a working system. Now your message isn’t random. It’s contextual. Instead of: “Hey, we help sales teams improve pipeline.” You say: “Noticed you’re rolling out HubSpot and hiring 3 SDRs. Most teams at that stage struggle with ramp time in the first 90 days and suffer low quota achievement. Worth comparing notes?” That’s how you start real conversations. We’ve seen social + tech stack signals alone push reply rates to 10–15%. Same rep. Same copy skill. Different list. It’s not how you write cold emails. It’s who you decide to write them to. And let Alta | AI Revenue Workforce do the heavy lifting. | 80 comments on LinkedIn

This guy re-invented himself at 40+ y.o. And landed a 6-figures GTM Engineer job Keith Gill didn't get lucky He's been working 1-1 with me to: > Hone his GTM Engineering skills > Create a better and more appealing positioning We live in a world full of negativity: 'AI replaced 456 programmers' 'Amazon fired 3848 employees' 'Banding Spoon fired ⅓ of Vimeo in one night' Well, I prefer to tell you the nice news. I prefer to tell the stories of the people I had the pleasure to mentor 1-1 that at any age (youngest is 18, oldest is 60+) decided to change and put in the work. Thanks Clay. Thanks to everybody who decided to reinvent themselves and crush it You can get there in 60 days.. clayoperator.com/program | 32 comments on LinkedIn

From my experience, you need at least 1 of these 3 things for a outbound to work in 2026: 1. Truly differentiated offer This is hard because it has to tie in fundamentally to your product or solution. But, you should work this in your GTM Messaging Strategy by thinking about compartmentalized offers. 2. Outstanding social proof Outstanding is the key, not just 'some' social proof. Again, difficult because if you don't have it, well, you dont have it. But you should work with in your GTM MS by thinking about social proof matching which you can layer at granular levels in Clay. 3. Creative lead magnet strategy This is where most companies should be because they don't have 1 or 2. If you dont have 1 or 2, and your top of funnel is an ebook or whitepaper about something they'd find in about 3 mins from a Google search... Get more creative. Btw, if you want to work with Spring Drive on your outbound, you can apply for a pilot at springdrive.co. | 25 comments on LinkedIn

I used to worry about silly things in my cold email campaigns. 5 years later, this is what I worry about and what I don't worry about. 1. Which tool I'm sending from. My brother and sisters in outbound, the underlying technology of how Smarltead and Instantly send email is the largely the same. Pick who you like the feature set of more and move on. 2. Does email warmup work and what are the perfect warmup settings. If email warmup didn't work, the platforms offering it unlimited for free would be the first people to ask you to shut it off so they could lower their AWS bill. I don't think there are perfect email warmup settings. As long as it's sending something and getting replies back it's fine. 3. Triple Validating Leads We validate our emails with million verifier before sending and enjoy the usage of bounce ban verified emails by using the Prospeo API. It's all going to be alright. 4. Spam words in my copy. Ok if you send "Bitcoin" or "jesus" or "oprah" you're going to have issues. But I see people asking questions in chats all the time about the word free or something. And it's ok in moderation. If you think you're having issues, pull it out and see if it improves. 5. The quest for email deliverability knowledge that no one else has. Any gimmick or strategy outside of the tried an true things we talk about on here all the time turns out to be a quick fad if it works at all or really not true. 6. 3rd party tools to tell if you're emails are going to spam Just check the reply rates of your domains... 7. Focusing on everything all at once in the copywriting vs spending 80% of my time on how I pitched the offer of what I'm helping to promote 8. Wondering if different email providers get better results than the others. Of course we use Hypertide.io and Zapmail.ai but if the provider you're using is fast and gives you access to the admin console, you're all set. 9. This might not fit as a worry, but you should run more tests than you think and see how the best brands run facebook ads. It's no different. Cold email is just a private ads network. 10. Working with a client where their website/case studies/social content wasn't dialed in. Yes that all helps but we run free tests where people can't look up the company and are just responding to the offer and it works all the time. What would you add to the list? | 16 comments on LinkedIn

We're cutting our AI token spend by over 90% with one platform. Taking screenshots of technical websites is all you need to know how to do. Vast(.)ai has made it really easy to upload any open source model (I went with OSS 120B and OSS 20B) and we are calling it with the API. On one GPU that I can rent for less than 75 cents per hour, I cranked it up to 16 requests per second with my test prompts that had about 10,000 input tokens in them. At these rates, 90% is actually an underestimate. It's probably approaching 95-99%. I can't even give you a tutorial because I literally just asked ChatGPT how to set it up and took screenshots along the way when I had issues so here's some use cases we are using it for. 1. Finding the business owner of a local business. They don't often have linkedin profiles but will tell their story on their page. Basically free to go 2-3 levels deep on the website and get the content for OSS 120B to find the owner's name. 2. ICP classification of companies. We can make the prompt as long as we want with as much input as we want in order to get the best results. 3. Org Chart Mapping Lots of data can be passed in to the AI with the JSON of linkedin profiles and connections can be drawn by department, location, and self reported team affiliation on their profile. 4. Finding the right target to reach out to. Now you can get every contact in a department and pick the best one in order of your preferences. 5. Keep going further than ICP Classification and get all of the content from websites and learn anything about your prospect's public facing content. With basically unlimited AI throughput, it feels that almost anything is possible now. | 20 comments on LinkedIn

An underrated Clay benefit is “how it teaches you to think” A wide gap exists in almost every GTM org I’ve met with, between conceptual ideas of where they want to be Vs what their current day to day workflows and tasks look like Tools do not fix this alone, you need to learn to think differently Given the most powerful algorithm for calculating micronutrient deficiencies, most people cannot solve for “getting more l-theanine” on their own You could say that to them - and they’d move on - no benefit derived - even if it is true Had a convo today and was challenged by a customer: “what if their data isn’t better? What if their signals aren’t better than other signals?” It’s a fair impulse to ask this sort of thing, but highlights the gap in worldview Many companies today are nerfing themselves They are losing ground, because of ground gained in years prior What winning looks like today isn’t the same as 2025 and isn’t the same at all vs 2020 Back to Clay (not a sponsored post) Clay has changed the lives of more “used to be good at sales development and want to go to the next level” people than any other tool on the market The big crms aren’t doing this - the market was too mature already - revops technical solutions come from known Revops ecosystems Mature frameworks, copy pastable workflows, etc It’s comfy to buy a known future… contained boundaries Clay is wall to wall blackboard, a fresh box of chalk, and an eraser You walk up, you start from the first line, and over time you can do whatever you can imagine The limiting factor for sales dev people is often technical acumen - using the tool will guide you, with iteration, like those early cold calls you stumbled through Don’t give up The limiting factor for technical people is often creativity and tactical familiarity (aka all the stuff 💜 🔮 Will Allred posts, among many others) This batch needs to just send it - let themselves fall on their face - and iterate forward as well; similar to making those first cold calls, stumbling through the ups and downs, and one day finding yourself quite alright at this weird little skillset If your company doesn’t want to fund those early stumbles - that’s where this big bright ecosystem of service business exists - many inside GTMCafe.com where we convene weekly for a mastermind of GTM excellence (free to all, no better space) The future will be orchestrated by talented, in the know people, with more asymetrical impact than ever before Weapon of choice? Clay (among others - but this is sincerely where you should start) | 19 comments on LinkedIn

New Tech stack for GTM Leaders to look at. Here's some of what we use or recommend at Spring Drive: (In no particular order) 1. Find leads - Ocean.io (lookalikes) - DiscoLike (lookalikes) - Crunchbase (great for saas, investors, and sem traffic proxy) - D7 Lead Finder (local lead king) - The big 'A' (static data still great here) - Apify (amazing marketplace for custom leads) - Linkedin Sales Navigator Premium - (the original pool, also great for 'posted content' filter for linkedin outreach) 2. Find Emails - Prospeo.io - LeadMagic - Findymail - BetterContact 3. Sending - Smartlead - Instantly.ai - EmailBison - HeyReach.io 4. Website tracking - RB2B - Vector 👻 5. Meeting Co-Pilot - tl;dv - AI Meeting Assistant 6. Video - Sendspark 7. No-Code Dev/IDE - Lovable - Cursor - Windsurf These are just a few, the list is getting long --- Anyone got q's about tech stack for their team? Comment below or DM me | 24 comments on LinkedIn

Here are a few things agencies are doing in 2026 to get ahead: 1) Using Supabase: caching emails, verifications, results. Saving significant time and money by not paying for every API call. 2) Vibe coding in LOVEABLE. A few weeks ago, a client was left high and dry from a previous CRM vendor. We rebuilt their CRM in a day. To be fair it wasn't that complex but we integrated it into Instantly.ai to capture their replies in real time. 3) Digging deep on AI...using tools like Manus AI to understand layers of complexity with browser based and deep layer computer agents. 4) Understanding tools like Firecrawl and Cursor and how they can dramatically change your approach to finding data and lead generation. 2026 will be the year the AI alters work for the better, and if done right clients and results will benefit.

22 tips to stay organized in Clay (so you don't burn credits like an idiot) ​ 1. Label inputs, steps, and outputs clearly. Input = what you need to start the Clay table (full name, LinkedIn URL, etc.). Steps = enrichments (Step 1, Step 2 etc). Outputs = final action (email, CRM push). Makes troubleshooting 10x easier. ​ 2. Color-code everything. Green = inputs. Yellow = needs tweaking. Red = something's broken. ​ 3. Use folders. SOP templates, client work, in-progress tables, temp tables. Delete untitled workbooks immediately. ​ 4. Know what Clay does / doesn't. Clay is the middleman. Always ask: What are my inputs? What does Clay do with them? What's the output? Etc, etc. ​ 5. Enable auto-deduplication. Bottom-right corner. NEVER spend credits twice on the same contact. ​ 6. Turn the table OFF when importing data. "Auto update columns" will burn credits before you're ready. Switch it off. Bottom-right corner. ​ 7. Test on 1 row. Then 2. Then scale. Run 1. If it works, run 2. If it works, run all. ​ 8. Switch off auto-update on complex tables. Stops enrichments from auto-triggering the next step and draining credits. ​ 9. Focus on ONE table at a time. Shiny object syndrome kills productivity. Start to finish. Then move on. Oh, and do not try to replicate the giveaway Clay tables from Linkedin posts. They are 99% BS and won’t work. ​ 10. Save other people's templates separately. Use them as inspiration only. Everyone's tech stack is different. ​ 11. Save YOUR templates for reuse. Spent 30 minutes writing a killer prompt? Save it. You'll use it again. ​ 12. Build waterfall templates. Email finders, enrichment chains - set them up once and then reuse them forever. ​ 13. Use Clay Functions. Automates repetitive enrichment flows. Will save you HOURS. ​ 14. Create a Do Not Contact (DNC) list. Separate table. Cross-reference with "Lookup Multiple Rows" so you never contact the same person twice. ​ 15. Push contacts to a CRM table. After adding leads to campaigns, write them to a master CRM table in Clay. Keeps everything organized. ​ 16. Define scope before building. Write down: Step 1, Step 2, Step 3. Then match each step to an enrichment + prompt. ​ 17. Prompt AI properly (6-step framework): → Input (what you're starting with) → Action (what you want AI to do) → Constraints (word count, format) → Prefix (sentence structure) → Example (show AI what good looks like) → Click “Generate” to generate the mega-prompt inside Clay ​ 18. Mark broken columns in red. Mental shortcut when you re-open the table. ​ 19. Check credit consumption. Settings → Credits Usage. See which tables are burning credits and replace them using external APIs. ​ 20. Buy tools outside Clay Cheaper to use Prospeo or Lead Magic directly than through Clay enrichments. ​ 21. Delete unused tables. Clean workspace = clear head. ​ 22. Don't overcomplicate it. Most people add 20 enrichments trying to look fancy. Start simple. Add complexity only when needed.

2010 was Salesforce 2015 was HubSpot 2025 is Clay These kinds of companies are only born every 5 to 10 years. They represent a paradigm shift in how marketers and salespeople do business, and they open up an entirely new business category. > Salesforce → CRM and cloud data storage > Hubspot → Inbound Marketing and RevOps > Clay → Clay Agencies and GTM Engineering Huge consulting businesses have been built on top of these ecosystems. (Some Salesforce implementation businesses are nine figures.) The number one advice if I wanted to skyrocket my career in marketing and sales is always the same. Niche down on a rocket that's booming and already in demand. Closing what's already hot is the number one business cheat code. And sell the new way of doing things. | 48 comments on LinkedIn

The tests are in and we are not allowing generic AI personalizations based on company descriptions as first lines anymore. I could prove this raised reply rates but recently, in most markets, it falls flat. We try to always lead with some kind of signal or pain point that we research but if we can't do that, we would fall back to a company description based personalization. Something like, "I saw you help fitness coaches retain clients and I wanted to get connected." We used to see increased reply rates when we personalized first lines with AI-generated content scraped from company websites. Now the data is pointing the opposite direction. Sophisticated buyers recognize generic AI personalization instantly which hurts your message: - "Hey, I saw how you helped fitness enthusiasts breathe better..." - "I noticed you're focused on helping local businesses..." - Generic references to their company description They've seen it a thousand times. Here's what we do instead: If we have a real signal: We use it. Something specific, custom-scraped, that ties directly to their problem and our value prop. **If we don't have a signal:** We treat cold email like a banner ad. Put the offer in the preview text. Make it punchy. No fake personalization. Think about it: When someone's scrolling on their phone or web browser, they see the subject line + preview text together. That's your click opportunity. Stop wasting time generating first lines from company descriptions. Unless you have custom research that goes well beyond what's publicly available on their homepage, foregoing AI personalization and writing punchy, split-tested first lines performs better. We've seen this deteriorate over the past six months across our customer base and as someone that would suggest it based on the data, I want to correct the record and say we're out of it now.