William

Is BI dead? The CEO of a BI company just called dashboards a plague. Ketan Karkhanis (@karkhanis) runs @thoughtspot. When I asked him where BI is going on the first episode of The Decision Layer, instead of defending dashboards, he talked about trust. And I think he’s right. For 20 years BI was the answer, but now it’s becoming the receipt. My bet: the company that wins AI analytics will be the one whose CFO actually believes the numbers. An agent that says “revenue dropped 12% because of EMEA renewals” is worthless if nobody can see how it got there. People will open a dashboard to double-check. Congrats, you’ve built an AI that creates more work. That’s why I think most “AI for analytics” projects launching right now are dead on arrival. They ship answers when they should be shipping evidence. Ketan put it simply: the hottest feature in AI is trust. And trust only comes from transparency built into the data. So the real AI work in your company is happening in the boring layer underneath, in metric definitions and lineage. The stuff nobody puts on a keynote slide.

I helped a $20M company cut its CAC by 35% in just 90 days. What happened: The company's marketing director was about to get 40% more money to spend on ads. But she wanted to know which ads were actually bringing in customers. The problem: - Meta reported 2,400 sales - Google reported 1,800 sales - But the company's store records showed only 1,100 actual customers Every platform was taking credit for sales. So we started with one simple rule: Use your own sales records to count customers, then figure out which ads brought them in. This helped the company make better decisions about where to spend its money. 90 days later, they were saving $16 on every new customer. One lesson to learn here: Make sure you know which campaigns bring in real customers before increasing your budget. The better you understand where your customers come from, the more you can get from every dollar you spend.

Your next $1M in ARR might come from the SaaS customers you already have. I worked with a $25M SaaS company that kept signing new customers, but its existing customers were spending 5% less each year. The company was winning new customers while losing ground with the ones it already had. So we focused on 3 moves: - Getting every team to agree on the same revenue numbers - Identifying customers at risk of leaving - Finding opportunities to grow existing accounts Within ONE quarter, revenue from existing customers went from shrinking 5% to growing 8%. The team found new growth opportunities by understanding the customers they already had. If you're a SaaS founder, ask yourself: How much more revenue could you generate if your existing customers stayed longer and spent more? You might already have the data to find out.

Your SaaS company has data on everything. So why are you still guessing? Say, you want to know why MRR dropped last month. So Finance sends you a spreadsheet. Sales sends you another report. Now you're comparing numbers, checking definitions, and asking your analyst why they don't match. An hour later, you still don't have an answer. Next week, the same thing happens with a different question. Well, your problem is simply that your team has to piece together the answer every time someone asks a business question. As your SaaS company grows, these manual workflows become harder to maintain. - Finance pulls the numbers - Sales explains what changed - An analyst checks the definitions and reconciles the reports Every question becomes another reporting task. The next-level SaaS companies have made those answers easier to find. Their teams use trusted metrics, shared definitions, and reliable data sources to answer common business questions without rebuilding reports from scratch. Start with the 5 business questions your team asks most often. For each one: - Define the metric - Agree on the trusted source - Make the answer easy to find Your team should spend less time piecing together numbers and more time deciding what to do with them.

If I asked you why your SaaS charges what it does, could you defend your pricing? If you can, skip this post. If you can't, I’ll solve your problem right here. Know that your SaaS could be leaving $10,000s on the table every month because your pricing is wrong. This happens because you're copying what your competitors charge without checking what YOUR customers are willing to pay. I've seen this happen at one survey software company that was using competitor pricing. Then they finally analyzed their own customers. They compared: - Revenue per customer in each country - How many customers upgraded from Basic to Pro - How often customers canceled in each country - Which acquisition channels brought in higher-paying customers And found something big: Customers in some countries were willing to pay significantly more than customers in others. Yet everyone was being charged the same price. Think about how much money you could be missing because you're charging your highest-value customers the same price as everyone else. Here's what I'd do with your pricing data: 1. Revenue: Compare average monthly revenue per customer in the US, UK, Germany, and other markets you serve. 2. Upgrades: Calculate what percentage of customers move from Basic to Pro in each country. 3. Churn: Compare the percentage of paying customers who cancel each month, by country and plan. 4. Pricing complaints: Count support tickets mentioning “too expensive” or “pricing” in each market. 5. Opportunities: Find countries with high upgrade rates, low churn, and few pricing complaints. Those are the markets I'd investigate first for a potential price increase. Then test a higher price with new customers and measure how many still buy. Stop copying competitors who know NOTHING about your customers. Your own data can tell you where to test charging more.

Stop hiring people just because they know how to use AI. Knowing 50 prompting tricks means ABSOLUTELY nothing if you can't produce good work. I'd rather hire someone who understands finance, marketing, data, or engineering and knows how to use AI to get results. Because AI can write a report in seconds. - But can you tell if the numbers are wrong? - Can you spot a bad decision? - Can you tell AI what actually matters to the business? That's where real expertise comes in. Next time you interview someone, ask 3 questions: 1. What have you actually built with AI? 2. What results did it produce? 3. How did you check if the output was correct? Stop asking candidates which AI tools they know. Ask them what they've actually accomplished with them.

There’s a BIG mistake SaaS founders make that we need to address: Too many SaaS companies wait until a free trial ends to figure out why someone didn't buy. By then, you’ve already lost the customer. The real problem usually happens in the first few days. - Someone signs up - Tries the product - Doesn't get value - Never comes back And your team doesn't realize it until the trial expires. I've seen a subscription company find something interesting in its data: Customers who completed fewer than 2 classes in their first month were 3X more likely to cancel. So they changed their onboarding emails to help new members complete more classes. Their retention efforts became 2X more effective. But they could only find this pattern after every team agreed on how to measure customer activity. Here's what you should do (if you’re running SaaS): 1. Open last month's free trial signups 2. Sort users by their last login 3. Find everyone who stopped using the product after day 1 4. Identify the first important action your paying customers took 5. Count how many trial users never completed that action Your product already tracks what users do. You just need to connect those actions to who eventually pays. Find out where users stop getting value, then fix that step before spending more money acquiring them.

Most SaaS founders think their churn problem is a product problem. Wrong. I’ve looked at the data from 50+ B2B and SaaS companies. There's a reason why 36% of your new signups feel confused and cancel by day three. The protocol to reduce churn for SaaS companies:🧵

Before you waste another $10K trying to get more SaaS leads, check the customers sitting right under your nose. Your SaaS could already have hundreds of people worth reaching out to. 7 places I'd look: 1. Free trial users who disappeared after day 1 2. Customers who canceled in the last 12 months 3. Good leads your CRM ranked too low 4. People who started signing up but never finished 5. Small customers who use your product heavily 6. Deals you lost because the timing was wrong 7. Ad channels that bring customers who stay the longest I saw an online marketplace get huge results from using its existing customer data. They sent 4M emails a week. Only 3% were sent after a customer took a specific action. That 3% generated 35% of all email revenue. Over 20 months, email grew from 2% to 12% of total revenue. No new hires. The opportunity was already sitting in their data. Pick one of these 7 groups and reach out this week before spending more money on new leads.

The craziest thing that ever happened to me: I ran my own consulting firm for 3 years without knowing which clients actually made me the most money. And I work in data. For my first 3 years running Valiotti Data: - I had no reports on my own business - I guessed which projects were profitable - I made decisions based on gut feeling Eventually, I hired an assistant. Today, she's my operations manager. She keeps our documents, finances, and cash in order. And I finally know where the money is going. Here's a quick test for every founder: 1. Write down your 3 most profitable customers from memory 2. Open last quarter's financial reports 3. Check how many you got right If you can't name your most profitable customers, how do you know which ones you should be getting more of?

Your SaaS MRR has been flat for 6 months. You have 3 options: 1. Win more new customers 2. Keep more of the customers you have 3. Collect all the revenue you already earned Many founders pick option 1 first, because ads can buy it. If the leak is in option 2 or 3, more ad spend raises your costs, and revenue stays flat. Option 3 sounds unlikely until you check it. I know a $6M marketplace with no data team kept its revenue in 3 systems: - A payment processor - The product database - An accounting tool In the first 30 days, we found $16K a month missing between them. That's $192K a year that no number of new customers would have recovered. The fix took about 40 hours over 3 weeks. Today an alert fires whenever the 3 systems disagree by more than $500. Run one check before you raise the ad budget. Pull last month's revenue from your payment processor, your database and your accounting tool, and write the 3 numbers side by side. If they differ by more than $500, start with option 3.

Before managing $2B+ in client revenue, I started businesses I barely knew how to run. A craft brewery An online coffee store A software company with ZERO experience 17 years and 50+ companies later, here are the lessons I learned the hard way since 2009: 🧵