Content examples of videos and copy we’ve created for customers across B2B industries: Tech, SaaS, Venture Capital, Enterprise Software, Consulting, Insurance, Logistics.
)
75 comments·51 reposts
Lessons From AWS: How I Saved 400M in 1 YearIn my last year at AWS, I was once tasked with finding $400 million in cost savings for cloud spending in just one year.
Generative AI can sometimes feel like talking to Elon Musk about politics.
It’s incredibly intelligent and confident, but you can’t trust it with everything.
ChatGPT, for instance, has already been trained on 1/3rd of all the information on the internet. So feeding the remaining 2/3rd won’t make it more accurate and unbiased.
So, how do we tackle this challenge?
One way of addressing this is:
Instead of a huge language model, you could go for a smaller one, maybe with a few billion parameters, but tailored to your specific local data.
You need to integrate the insights from the large language model (LLM) with the knowledge from your local data, giving more weight to the latter due to its relevance to your context.
That’s what Google is doing with its medical database.
This approach allows us to tailor the AI to handle context-specific information and reduce the impact of biases present in broader datasets.
Once you get that, the LLMs are great at:
![]()
Information Retrieval: It allows you to ask an English language question and get to the right document in your million corpus document file.
![]()
Summarization: You can have it read through a 100-page contract and give you the salient clauses.
![]()
Content Generation: It can simplify something down for a 10-year-old or introduce headers for each major paragraph in a document.
By leveraging LLMs for these tasks while incorporating domain-specific models, we can maximize the potential of generative AI and foster a more responsible and impactful AI ecosystem.
Share your thoughts and ideas in the comments below!
#chatgpt #generativeai #ai
It’s incredibly intelligent and confident, but you can’t trust it with everything.
ChatGPT, for instance, has already been trained on 1/3rd of all the information on the internet. So feeding the remaining 2/3rd won’t make it more accurate and unbiased.
So, how do we tackle this challenge?
One way of addressing this is:
Instead of a huge language model, you could go for a smaller one, maybe with a few billion parameters, but tailored to your specific local data.
You need to integrate the insights from the large language model (LLM) with the knowledge from your local data, giving more weight to the latter due to its relevance to your context.
That’s what Google is doing with its medical database.
This approach allows us to tailor the AI to handle context-specific information and reduce the impact of biases present in broader datasets.
Once you get that, the LLMs are great at:
By leveraging LLMs for these tasks while incorporating domain-specific models, we can maximize the potential of generative AI and foster a more responsible and impactful AI ecosystem.
Share your thoughts and ideas in the comments below!
#chatgpt #generativeai #ai
19 comments·1 repost
Generative AI Can Sometimes Feel Like Talking to Elon Musk About PoliticsIt’s incredibly intelligent and confident, but you can’t trust it with everything.
)
33 comments·4 reposts
What I’d Do as a New Head of MarketingIf I was hired as Head of Marketing at a growing b2b SaaS company, here's what I would do on day 1
)
13 comments
My 5 Secrets for a Great Work-Life BalanceStruggling to set boundaries in your work and life while working remotely?
)
16 comments·4 reposts
Lessons from AWS: Different Is Better Than BetterLessons from building a multi-billion dollar service at Amazon Web Services (AWS)
)
7 comments·5 reposts
Equal Geographical Pay vs. Location-Based PayIf you’re starting a remote-first company, I’d advise you to consider equal geographical pay.
)
1 repost
The 6-step system we used to book $102k with 1 adLast month, we booked $102,000 in net new revenue from 1 ad.
Software alone can’t solve your problems.
You can’t introduce software, equip your front-line practitioners with it, and expect everything to fall into place seamlessly.
This perspective originates from a somewhat naive buying process. Leadership sometimes assumes that choosing software and arranging a few demos will equip front-line practitioners with what they need to use the new solution. That approach might work for “power users”, but not for your “average user”.
Fast forward a year, and they’re surprised that most users haven’t embraced the software. They’ve spent money, but poor return on investment makes leadership question what went wrong.
The primary cause? A lack of proper engagement.
When introducing new software to the enterprise, whether a productivity solution like VisibleThread or something completely different, you can’t expect people to magically see its value and know how to make their job better or easier.
In every organization, there’s a wide spectrum of users. A small fraction might quickly adopt and exploit the software, but the average person won’t necessarily have that capacity. And you can’t blame them. They’re already preoccupied with daily tasks, personal lives, and other responsibilities.
To ensure successful software adoption, it's crucial to guide its integration into daily workflows. This requires an internal champion who oversees the adoption and holds others accountable.
What are internal champions holding practitioners accountable for?
- Firstly, it's about tracking usage. Simple metrics such as sign-ins or specific software-related tasks can provide insights.
- The second aspect is gathering feedback on the software's performance and utility. This internal change management coordinator should relay this data to the senior leadership.
After all, they were the ones who initially approved the purchase and will want evidence of its value.
The crux is that successful software deployments necessitate strong internal ownership post-purchase. Especially when a software's impact spans more than just one team, its influence is felt organization-wide.
Without a dedicated champion ensuring adoption and gathering metrics, investments in new software can quickly wither on the vine.
#softwareadoption #usagemetrics #userengagement
You can’t introduce software, equip your front-line practitioners with it, and expect everything to fall into place seamlessly.
This perspective originates from a somewhat naive buying process. Leadership sometimes assumes that choosing software and arranging a few demos will equip front-line practitioners with what they need to use the new solution. That approach might work for “power users”, but not for your “average user”.
Fast forward a year, and they’re surprised that most users haven’t embraced the software. They’ve spent money, but poor return on investment makes leadership question what went wrong.
The primary cause? A lack of proper engagement.
When introducing new software to the enterprise, whether a productivity solution like VisibleThread or something completely different, you can’t expect people to magically see its value and know how to make their job better or easier.
In every organization, there’s a wide spectrum of users. A small fraction might quickly adopt and exploit the software, but the average person won’t necessarily have that capacity. And you can’t blame them. They’re already preoccupied with daily tasks, personal lives, and other responsibilities.
To ensure successful software adoption, it's crucial to guide its integration into daily workflows. This requires an internal champion who oversees the adoption and holds others accountable.
What are internal champions holding practitioners accountable for?
- Firstly, it's about tracking usage. Simple metrics such as sign-ins or specific software-related tasks can provide insights.
- The second aspect is gathering feedback on the software's performance and utility. This internal change management coordinator should relay this data to the senior leadership.
After all, they were the ones who initially approved the purchase and will want evidence of its value.
The crux is that successful software deployments necessitate strong internal ownership post-purchase. Especially when a software's impact spans more than just one team, its influence is felt organization-wide.
Without a dedicated champion ensuring adoption and gathering metrics, investments in new software can quickly wither on the vine.
#softwareadoption #usagemetrics #userengagement
10 comments
Behavior Change Is Really HardSoftware alone can’t solve your problems
)
21 comments·8 reposts
How to Reduce Your Observability Costs by 2-4XI keep hearing from companies about their soaring costs of observability vendors.
)
18 comments·2 reposts
3 Lessons From Building $1B+ Services at AWSAt AWS, I had the unusual opportunity to create around half a dozen $1B+ services.
)
9 comments·4 reposts
Working From Home vs. Working RemotelyDon’t restrict your employees to working from within a 30-mile radius.
)
4 comments
How Zocks Allows Deeper Customer EngagementHere’s how Zocks helps you build better client relationships
)
15 comments·12 reposts
Lessons From AWS: How To Write A NarrativeI learned how to write narratives when I was at Amazon Web Services (AWS).
)
5 comments·8 reposts
Why Cyber Insurance Is Not Enough“My company has cyber insurance. Isn’t that enough to protect us?”
)
5 comments
How We Helped a B2B SAAS Generate 45 Sqls in 4 MonthsWe recently helped a B2B software startup launch a new product ($100k ACV - prev. $15-20k ACV) and go from 5 to 50 SQOs in just 4 months.
Last month, we booked $102,000 in net new revenue from 1 ad.
Here's the exact 6-step system we used:
1. Create a content engine to post daily on my Linkedin
- Ideate questions around our product, company, and expertise (b2b marketing & demand gen)
- Weekly recording session with Tobias to answer the questions in a conversational way
- Turn the recording & transcript of that conversation into bite-sized video, text and image Linkedin posts
- Repeat weekly for 5 posts/week or 20/month -> daily post
2. Determine best-performers
- Do monthly analysis using Shield to determine which posts & messages resonated most with our audience
- Main metric: "total weighted engagements (TWE)" = likes + 2*comments + 4*shares
- Best-performer = any post with TWE greater than at least 1 standard deviation above the average for that data set (you can use other thresholds or just subject feeling)
3. Feed best-performers into targeted Linkedin ad campaign
- Run towards 3 cold targeting groups
1. Targeting ICP based on job titles, geography and company size for top-down approach (for us: any Founder or Head of Marketing of a software business with 10-100 employees)
2. Targeting broad ICP based on job function/department for bottom-up approach (for us: anyone working in the Marketing, Sales, or Biz Dev department of a software business with 10-100 employees)
3. ABM/account-level targeting: I created a list of 200 target accounts who are our dream customers
- Retargeting layer: any website visitor, any Linkedin company page visitor and anyone who engaged with the ads in the cold layer (50% video viewers)
4. Objective: in-feed consumption of the content + drive interested people to website
- Linkedin campaign objective: Engagement
- CTA: Learn more
- Landing page: our website homepage (not a lead gen form, not our demo page)
- Tracking conversions: Linkedin Insight Tag with 30-day "Last touch - Last campaign" attribution
5. Wait for inbound
- We want to drive high-intent demo requests ONLY, not leads
- By sending people to the home page, rather than a lead gen form or a call booking page, we only get people who are REALLY interested
- Installing self-reported attribution: most demo requests are not tracked by Linkedin Insight Tag, so make sure you have a qualitative way to attribute
6. Convert & optimize
- Current cost per qualified demo from ads: $200-300
- In our case, 80% of the demo requests were generated by 1 outlier ad
- Outliers driving most of the conversions is a common pattern we see, that's why you need a content engine to test new messaging & creatives all the time, to find your next outlier
- Total ad spend here was $3,500, so that's a 30x ROI on revenue
- $102,000 revenue generated is just the initial contract value, LTV is higher
![]()
Btw, this is the exact system we implement for customers at Project 33.
#linkedin #demandgen #founderledmarketing #GTM #saas
Here's the exact 6-step system we used:
1. Create a content engine to post daily on my Linkedin
- Ideate questions around our product, company, and expertise (b2b marketing & demand gen)
- Weekly recording session with Tobias to answer the questions in a conversational way
- Turn the recording & transcript of that conversation into bite-sized video, text and image Linkedin posts
- Repeat weekly for 5 posts/week or 20/month -> daily post
2. Determine best-performers
- Do monthly analysis using Shield to determine which posts & messages resonated most with our audience
- Main metric: "total weighted engagements (TWE)" = likes + 2*comments + 4*shares
- Best-performer = any post with TWE greater than at least 1 standard deviation above the average for that data set (you can use other thresholds or just subject feeling)
3. Feed best-performers into targeted Linkedin ad campaign
- Run towards 3 cold targeting groups
1. Targeting ICP based on job titles, geography and company size for top-down approach (for us: any Founder or Head of Marketing of a software business with 10-100 employees)
2. Targeting broad ICP based on job function/department for bottom-up approach (for us: anyone working in the Marketing, Sales, or Biz Dev department of a software business with 10-100 employees)
3. ABM/account-level targeting: I created a list of 200 target accounts who are our dream customers
- Retargeting layer: any website visitor, any Linkedin company page visitor and anyone who engaged with the ads in the cold layer (50% video viewers)
4. Objective: in-feed consumption of the content + drive interested people to website
- Linkedin campaign objective: Engagement
- CTA: Learn more
- Landing page: our website homepage (not a lead gen form, not our demo page)
- Tracking conversions: Linkedin Insight Tag with 30-day "Last touch - Last campaign" attribution
5. Wait for inbound
- We want to drive high-intent demo requests ONLY, not leads
- By sending people to the home page, rather than a lead gen form or a call booking page, we only get people who are REALLY interested
- Installing self-reported attribution: most demo requests are not tracked by Linkedin Insight Tag, so make sure you have a qualitative way to attribute
6. Convert & optimize
- Current cost per qualified demo from ads: $200-300
- In our case, 80% of the demo requests were generated by 1 outlier ad
- Outliers driving most of the conversions is a common pattern we see, that's why you need a content engine to test new messaging & creatives all the time, to find your next outlier
- Total ad spend here was $3,500, so that's a 30x ROI on revenue
- $102,000 revenue generated is just the initial contract value, LTV is higher
#linkedin #demandgen #founderledmarketing #GTM #saas
52 comments·2 reposts
Last month, we booked $102,000 in net new revenue from 1 ad.Here's the exact 6-step system we used:
)
10 comments·3 reposts
Explaining AI Usage to Clients in FinanceHow should you talk to your client about using AI?
)
6 reposts
Why 90% Of Your Content Goes Unused90% of the content is never used. Ever.
)
30 comments·3 reposts
Building the First Country on the InternetI joined SafetyWing because I love its mission: Building the 1st country on the internet.
)
9 comments
What I Learned From My Boss at Morgan StanleyI got very little mentorship in my life for some reason.
)
1 comment·1 repost
The Biggest Challenge Financial Advisors FaceHow can Zocks help financial advisors resolve their day-to-day challenges?
)
4 comments·1 repost
The Role of AI/ML and Automation in Cyber SecurityLet’s talk about having automation tools and AI/ML for cyber security.
Ready to scale your B2B software company? Speak to someone on our team to find out if Founder-Led Marketing would make sense for your company.
Subscribe to our newsletter for monthly Founder-Led Marketing insights.
We use cookies to make things better.