How to Answer Pricing Questions
Pricing strategy questions assess your ability to set a product price that makes sense in a given business context. Pricing is much more complex than just setting a number, and answering pricing questions well requires a firm understanding of how the product serves users and the core business. Example questions include:
Asked at Google Asked at Google 
Asked at Flipkart To answer these, first consider the many ways product pricing can affect a product's success and the company as a whole. For starters:
- Price affects revenue and gross margin. Generally, higher-priced products sell fewer units, but the gross margin on each unit is higher.
- The pricing model is a strategic decision in its own right. Freemium, usage-based, subscription, and hybrid aren't interchangeable. How you charge shapes customer behavior and cost structure as much as how much you charge.
- Early-stage products are often priced for adoption, not profitability. Uber subsidized rides to build a habit; AI products like ChatGPT and Claude are doing the same now. Ask whether the company needs market share or margin before you set a price.
Keep these considerations in mind as you work to understand the problem context, the product and company goals, and the competitive landscape. Once you've covered those, you can hone in on a pricing strategy.
Why this matters now
Pricing strategy has become more complex in the AI era, not less. Traditional SaaS pricing assumes near-zero marginal cost at scale: add a user, collect their subscription fee. With AI products, that assumption breaks. Every query costs money. Every inference call incurs real compute costs. That fundamentally changes the pricing calculus, and interviewers at companies like OpenAI, Anthropic, and Perplexity are now asking pricing questions specifically to see whether you understand the cost structure underlying the product.
Inference pricing unit economics. At $50/user/month, a power user generating 500 completions a day can easily cost you $0.20–$0.30 per session in API calls. The users who love your product most destroy your margin. That's why AI pricing questions test whether you understand the cost side, not just your willingness to pay.
Step 1: Define the landscape
The goal of this step is to understand what pricing is meant to achieve for this product, and what levers you actually have. By the end, you should be able to say: "I think this company launched this product because X, which means pricing should prioritize Y, and the key constraints going into model selection are Z." Without that synthesis, any model you pick is a guess.
Start with clarifying questions to understand the product:
- What value does this product provide?
- Who is the target user?
- How does it differ from the company's existing offerings?
- Is this a brand-new product or a pricing change to an existing one?
The launch context matters. Setting an initial price is very different from changing an existing pricing model with established user expectations.
Then zoom out to company goals:
- Is the company optimizing for growth, profit, or market share?
- Is it entering a new market?
- Is the product meant to strengthen or reposition the brand?
Next, evaluate the landscape factors most relevant to this product:
- Competition: Pricing is positioning. While you may assume pricing lower is always the safer move, pricing far below competitors can signal weakness, not value. Apple doesn't undercut Samsung on flagship devices. Stripe didn't launch as the cheapest payment processor.
- Internationalization: Willingness to pay varies dramatically across regions. Western markets pay 3 to 5 times more for digital products than markets in India or Southeast Asia. Spotify charges ~$11/month in the US and ~$2/month in India. A subscription model that works in the US may require freemium or ad-supported elsewhere.
- Consumer sensitivity: How price-conscious are users? What lower-cost alternatives exist?
- Public perception: Could pricing create backlash or affect brand trust?
Before moving to Step 2, synthesize what you've learned. What is pricing trying to achieve here? What are the constraints? What factors will guide model selection?
Claude Design
"Claude Design is targeting non-designers (PMs, founders, marketers) who want to build UI without hiring a designer. Anthropic launched it to expand into a new user category and build a habit before competitors establish a foothold in the space. The goal right now is adoption, not margin. But they can't price at zero because visual compute is expensive: a single screen design burns tokens at a rate that would take many paragraphs of text to match.
The competitive set isn't Figma. It's the non-designer's current alternatives: hiring a contractor or skipping design entirely. Those are slow and expensive, so Claude Design has real pricing room. The main risk isn't pricing too high; it's users hitting a limit mid-session, never finishing anything, and leaving frustrated. That failure mode should shape the pricing structure.
So going into model selection: we're optimizing for adoption over margin, cost recovery is a floor constraint, not a ceiling, and we need a structure that gets users to a finished output before they run out of runway."
Step 2: Choose a pricing model
With your goals and constraints from Step 1 in hand, choosing a pricing model is actually two decisions made in sequence: first, what is your monetization approach (how does the business make money from this product at all?), and then, within that approach, what is the pricing structure (how exactly do you charge?). candidates often skip the first decision and jump straight to the second, which is how you end up recommending a subscription model for a product that probably should be ad-supported.
Decision 1: Monetization approach
Start by deciding which bucket fits. Or mix and match from the buckets. This is the top-level strategic question.
Decision 2: Pricing structure
Once you've chosen your approach, select the structure within it.
If everyone pays:
If free with indirect monetization:
If part free, part paid:
For consumer AI products specifically, usage-gating has a known failure mode. Users need to reach a finished output before hitting the wall, or they churn frustrated. Design the free allowance around completion, not just usage volume.
Claude Design
"First decision: monetization approach. Ad-supported is out; ads would destroy the creative context. Purely free as an ecosystem play doesn't work because Claude Design runs on separate visual compute that is significantly more expensive than text inference: Anthropic can't absorb that cost at scale. The real choice is between everyone pays and part free, part paid. Since Anthropic is in adoption mode and needs to establish habit before competitors do, part free, part paid is the right call.
Second decision: structure within that bucket. Feature-gating doesn't work well here because the core capability (generate a UI) is the same for all users. Usage-gating is the right fit, but the unit has to change. The current tiers are framed as '5x more usage' and '20x more usage,' which means nothing to a non-designer. The unit should be sessions or screens. A hybrid subscription with a session allowance and per-session overage beyond that gives most users predictability while keeping the cost structure defensible for Anthropic."
Step 3: Determine the price
With a model chosen, set the actual number. Use three anchors to triangulate, then commit to a specific recommendation you can defend.
Value anchor (ceiling): What is the customer willing to pay? What outcome does this product deliver, and how much is that worth? For B2B, tie it to measurable ROI. For consumer, tie it to alternatives: what does it cost to do this manually, hire someone, or use a competitor?
Cost floor: What's the minimum you need to charge to not lose money at scale? For AI products, this means inference costs per user, not just engineering headcount.
Competitive calibration: What does the market expect? Use competitive pricing to sanity-check your range. But don't let it make the decision. If the whole market is underpriced, that's an opportunity.
Value sets the ceiling. Cost sets the floor. Competition tells you where customer expectations sit within that range. Once you have a number, commit to it. Then stress-test:
- Cannibalization: Does this price create wrong incentives for existing products or tiers?
- Market signaling: Does the price say the right thing about quality?
- Long-term arc: Is this a launch price or steady-state? If you're pricing for adoption now, how do you raise prices later without triggering churn?
A complete pricing answer picks a price, anchors it to something real (usually competitive benchmarks), names a defensible range, and makes a value argument for where in that range to land. That's the minimum bar.
To signal seniority, triangulate all three anchors explicitly: cost-based, competitive, and value-based. Explain what the chosen price signals to the market and how pricing strategy should evolve as the product matures. Also, flag cannibalization risk before the interviewer has to ask because a senior+ candidate knows that's the first follow-up.
Claude Design
"The value anchor is the alternative: hiring a contractor for a day of design work runs $300-500. Claude Design doesn't need to beat that on quality; it needs to beat it on speed and accessibility for someone who just needs something good enough. That puts willingness to pay somewhere in the $30-75/month range for a professional user.
The cost floor is real: visual inference is expensive, and Anthropic can't give unlimited usage away. A $20 Pro plan inclusion for light use makes sense as a trial, but a serious user will burn through it fast.
Competitive calibration: v0 by Vercel charges $20/month for a generous allowance, Canva Pro is $15/month. Those are loose comps but set user expectations.
My recommendation: a $50/month Professional tier framed in sessions or screens, not tokens. Add a contextual upgrade prompt mid-session when users are close to their limit ('You have about two screens left: upgrade to finish this project'), and make the per-session value improvement between tiers explicit so the jump to $100 feels like an obvious choice rather than a mystery.
Stress-test: the main risk is setting a free tier so generous that users never hit the wall. The moment of frustration is also the highest-intent conversion moment. Calibrate the free allowance to get users to their first real output, then stop. Not before, and not far after."