Inspired by the classic 2012 Forbes post “Top 10 Questions Every Business Plan Should Answer” — updated for the world we actually build in today.
I came across that old Forbes post the other day. The advice has aged surprisingly well, but reading it in 2026 felt a bit like listening to a great song on a cassette tape — timeless, but unmistakably from another era.
Back in 2012, “SaaS” was still something you had to explain to half the people in the room. AI meant IBM Watson winning Jeopardy. Nobody priced anything per-token. Nobody was worried about an OpenAI update vaporizing their roadmap overnight.
So here’s my version. Same spirit — ten questions any IT company (SaaS startup, boutique agency, integration shop, AI product) should be able to answer before they write a plan, build a deck, or talk to a single investor or customer.
1. What problem are you actually solving?
The bar is higher than it was in 2012. ChatGPT can handle most “I wish there was a tool that…” complaints with a prompt. If your problem can be solved in 30 seconds by someone pasting into Claude or Copilot, it’s not a business — it’s a feature of something bigger.
Be specific. Not “we help companies be more productive.” Who loses sleep over this? How much does it cost them today in time, money, or risk? Write it so a person who has the problem would recognize themselves in your first sentence.
2. What’s your solution — and why can’t a weekend hacker with an LLM clone it?
This is the defensibility question, and in the AI era it matters more than ever. Shipping is no longer a moat. Anyone can vibe-code an MVP in a weekend.
Your defensibility probably lives in one of these:
- Proprietary data or workflows you uniquely have access to
- Deep domain integration (the kind that takes years of client work to understand)
- Distribution — you own the customer relationship
- Network effects
- Trust, compliance, and regulatory fit (especially critical in EU enterprise)
“We built it on Next.js and it’s really well designed” is not a moat. Say the quiet part out loud: why are you defensible a year from now?
3. How big is the pain — and is it a real market, or a LinkedIn trend?
TAM/SAM/SOM still matters, but the trap has changed. Today the trap is pitching a market that mostly exists in VC Twitter threads. “AI agents for X” is a category with €0 in validated annual spend until someone actually signs a contract.
Show real numbers: how many target customers exist, what they already spend on adjacent solutions, and how fast that line is moving. If you can’t find existing budget lines to redirect, you’re not entering a market — you’re creating one. That’s a much harder pitch.
4. How do you make money — and does your pricing survive the AI cost curve?
Business models in SaaS used to be simple: per-seat, monthly. That’s getting messy.
- Per-seat pricing is under pressure when one “seat” uses an agent that does the work of five people.
- Usage-based (tokens, API calls, runs) is clean for you but scary for buyers.
- Hybrid (platform fee + usage) is where most of the market is landing.
And here’s the ugly question nobody asked in 2012: what are your COGS when a customer goes heavy on AI features? If 40% of your revenue goes to OpenAI or Anthropic, your gross margin looks like a hardware company, not software. Plan for it.
5. Who buys it — and how do you actually reach them?
Your go-to-market motion needs to match your deal size. This is where a lot of early IT companies quietly burn through runway.
- Under €5K ACV: product-led growth, self-serve, content, community.
- €5K–€50K: inbound + inside sales, hands-on demos, short cycles.
- €50K+: enterprise sales, pilots, procurement, security reviews, RFPs.
I’ve seen a €200/month SaaS try to run enterprise field sales, and a €100K/year platform try to close deals with a “Start free trial” button. Neither works.
Bonus in 2026: SEO is shifting to AEO/GEO — being found by AI search agents, not just Google. If your buyer asks Claude or Perplexity “who does X?”, is there any chance your name comes up?
6. Why is your team the right one — especially now?
In IT, being technical isn’t a differentiator anymore. AI flattened that. Domain depth is the real signal.
If you’re building compliance tooling for audiology clinics, does your team know what a REM measurement is? If you’re selling into Danish patient associations, does someone on your team actually speak the language of the people signing the contracts?
Investors and customers both want to see three things: technical credibility, domain credibility, and the ability to ship without drama. If you only have two of the three, be honest about which one’s missing and how you’ll plug it.
7. Who are your competitors — including the ones that didn’t exist last quarter?
The old answer was a 2×2 grid with you conveniently in the top-right quadrant. Update it.
Your competition in 2026 includes:
- Direct competitors (obvious)
- Incumbents — the Microsofts and Salesforces bolting on AI features every quarter
- Horizontal AI platforms — OpenAI, Anthropic, Google — that might swallow your use case with their next release
- In-house builds — customers now genuinely ask “why wouldn’t we just build this with Cursor and three engineers?”
- The status quo — spreadsheets and email. Still your biggest competitor in most B2B markets.
Your answer isn’t “we’re better.” It’s “here’s where we sit when the big players move, and here’s why the customer still picks us.”
8. What have you shipped — and what’s next?
Investors and serious customers want proof you execute. In the IT world that means concrete things:
- Customers live in production (with names, if they’ll let you use them)
- ARR, retention, NPS, any usage data that shows stickiness
- A public product — not just a Figma file
- A credible 12-month roadmap tied to revenue milestones, not feature lists
“We plan to launch in Q3” landed fine in 2012. Today it’s a red flag unless you can show why shipping is blocked on something other than your own team.
9. What are the unit economics — honestly?
SaaS has its own vocabulary for a reason. Know yours cold:
- CAC and CAC payback period
- LTV / CAC ratio
- Gross margin — and what it looks like after AI inference costs
- Net revenue retention — the single most important SaaS number there is
- Burn multiple / Rule of 40 if you’re scaling
If you’re pre-revenue, at minimum have realistic modeled assumptions with sources. “We’ll get 2% of the market” is not a model. “Comparable companies at our stage convert inbound demo requests at 8–12%, and we’re assuming 5%” is a model.
10. How much do you need — and what milestone does it buy?
The final one, and the one most founders get wrong.
Don’t raise for 18 months of “stuff.” Raise to hit a specific, credible milestone — the one that either (a) gets you to break-even, or (b) unlocks the next round at a meaningfully higher valuation. Anything else is a slow bleed.
Break the ask down: which hires, when, why; infrastructure; sales motion investment; and — this is the new line item — AI/compute budget if that’s real for you. And show what you won’t spend on. Discipline is signal.
One extra question Forbes couldn’t have asked in 2012
What happens to your business if the major AI labs ship your core feature next Tuesday?
If the honest answer is “we’re dead,” that’s a problem. If it’s “we’re fine, because our value lives above or around the model,” that’s a defensible business in 2026.
Every IT company — mine included — should be able to answer these ten (okay, eleven) questions in a few clear sentences each. If you can’t, you don’t have a plan yet. You have an idea.
And in this market, ideas are the cheapest thing around.