How do you use ChatGPT Deep Research for market sizing?

Quick Answer: ChatGPT Deep Research can produce structured TAM, SAM, and SOM estimates for B2B SaaS markets by synthesising public data, industry reports, and competitor signals in a single session. Feed it a well-structured prompt with your target segment, geography, and assumptions, and it returns a documented market sizing report in minutes rather than days.

B2B market sizing is one of those tasks that looks simple on a slide and takes forever in practice. You end up bouncing between Statista paywalls, analyst PDFs, and LinkedIn guesswork, then stitching it all together manually.

ChatGPT Deep Research changes that workflow. This guide shows SaaS Hackers readers exactly how to use it to build a defensible TAM/SAM/SOM breakdown, including the prompt template we use and where to sanity-check the output before it goes in front of investors or your board.

What Is ChatGPT Deep Research (and Why It Matters for Market Sizing)?

ChatGPT Deep Research is a mode within ChatGPT (available on paid plans) that runs multi-step web research autonomously before generating a response. Instead of answering from training data alone, it searches, reads, and synthesises sources in real time, then produces a structured report with citations.

For market sizing, this matters because:

• It pulls from live industry sources rather than data frozen at a training cutoff
• It cross-references multiple inputs to triangulate estimates
• It documents its reasoning, so you can audit the logic

The output is not a replacement for commissioned research. It is a fast, structured first pass that would previously take a junior analyst two to three days to produce.

Understanding TAM, SAM, and SOM Before You Prompt

Before writing a single prompt, get clear on what you are actually asking for.

TAM (Total Addressable Market): The total global revenue opportunity if you captured every possible customer in your category.

SAM (Serviceable Addressable Market): The portion of TAM you can realistically reach with your current product, pricing, and go-to-market motion.

SOM (Serviceable Obtainable Market): The share of SAM you can win in a defined timeframe, typically three to five years, given your resources and competition.

Each layer requires different inputs. TAM is usually top-down (industry reports, analyst estimates). SAM requires your ICP definition (company size, sector, geography). SOM requires competitive data and growth assumptions.

ChatGPT Deep Research handles TAM and SAM well. SOM needs your internal assumptions baked into the prompt.

How to Set Up ChatGPT Deep Research for a Market Sizing Session

Step 1: Open Deep Research Mode

In ChatGPT, select the model dropdown and choose Deep Research. This is available on ChatGPT Plus, Team, and Enterprise plans. The session will indicate it is running research before responding.

Step 2: Prepare Your Inputs

Before prompting, write down:

• Your product category (be specific: "AI-powered contract review software" not "legal tech")
• Your target customer profile (company size, industry, geography)
• The revenue model you are sizing (ARR per seat, per transaction, per company)
• The time horizon for your SOM estimate
• Any known competitors or market benchmarks you want included

The more specific your inputs, the tighter the output.

Step 3: Run the Prompt (Template Below)

Paste your prepared prompt into Deep Research mode. Expect the session to take two to five minutes as it runs searches.

Step 4: Audit the Sources

Deep Research cites its sources. Open each one. Check that the underlying data is from credible organisations (Gartner, IDC, Grand View Research, government datasets, or reputable trade bodies). Discard any figure sourced from a vendor's own market claims.

The Prompt Template: TAM/SAM/SOM for B2B SaaS

Copy and adapt this prompt. The bracketed sections are where you insert your specifics.

Prompt:

You are a B2B market research analyst. I need a structured market sizing report for a SaaS product in the following category: [describe your product in one sentence, e.g. "AI-powered invoice processing software for mid-market logistics companies"].

Please research and produce a TAM/SAM/SOM analysis using the following parameters:

Target customer profile:

• Company size: [e.g. 50-500 employees]
• Industries: [e.g. logistics, freight, supply chain]
• Geography: [e.g. UK and Western Europe]
• Buying role: [e.g. CFO or Head of Finance]

Pricing assumption: [e.g. £500/month per company, billed annually]

TAM: Estimate the total global market for [product category] software. Use top-down methodology, citing analyst reports, industry data, or credible market research where available.

SAM: Narrow TAM to companies matching the target customer profile above. Estimate the number of qualifying companies and multiply by the annual pricing assumption to produce a revenue-based SAM figure.

SOM: Assume a realistic market penetration rate for a well-funded Series A SaaS company over five years. Cite comparable SaaS benchmarks where possible.

For each layer, show your working: the data source, the assumption applied, and the resulting figure. Flag any estimates where data quality is low or where I should verify independently.

Format the output as: (1) Executive Summary, (2) TAM Analysis, (3) SAM Analysis, (4) SOM Analysis, (5) Key Assumptions and Risks, (6) Sources.

This structure forces Deep Research to show its reasoning at each layer, which is what makes the output auditable rather than just a number you cannot defend.

What Good Output Looks Like

A well-executed Deep Research response to this prompt will include:

• A TAM figure with a source (e.g. "The global accounts payable automation market was valued at $2.8 billion in 2023, according to Grand View Research")
• A SAM calculation showing the number of qualifying companies multiplied by your pricing assumption
• A SOM estimate with a stated penetration rate and a comparable benchmark
• A clear list of assumptions and where the data is weakest

If the output gives you a single number without showing the working, regenerate with a follow-up prompt: "Show me the step-by-step calculation for each layer with the source for each input."

Where ChatGPT Deep Research Falls Short

Deep Research is strong at synthesis. It has real limits you need to plan around.

Data freshness varies. Some sources it pulls will be two to three years old. Always check publication dates on cited reports.

It cannot access paywalled databases. Gartner and Forrester full reports, Bloomberg, and Pitchbook are behind paywalls. Deep Research will cite summary data where it appears publicly but cannot pull full analyst reports.

SOM is inherently speculative. No AI tool can tell you what share of a market you will win. The SOM figure it produces is a benchmark-informed estimate, not a forecast. You own the assumptions.

It can hallucinate citations. This is the biggest risk. Always verify that the source it cites actually contains the figure it attributes to it. Open the URL. Check the number.

How to Sanity-Check Your Market Sizing Output

Run these checks before using the figures in any external document.

1. Bottom-up cross-check. Take your SAM calculation and build it from the bottom up independently. Use LinkedIn Sales Navigator or Apollo to count qualifying companies in your ICP. Multiply by your ACV. If the number is within 30-40% of the top-down SAM, you have a reasonable estimate.

2. Competitor revenue check. If a public or well-reported competitor exists in your category, use their disclosed ARR and market share claims to sense-check your TAM. If a competitor claims 5% market share and reports £10M ARR, the TAM implied is £200M.

3. Investor benchmark check. Search for investor memos or analyst notes that reference your category. Y Combinator, Bessemer Venture Partners, and OpenView Partners publish market data publicly. If your TAM is wildly different from what investors are citing, find out why.

4. Ask Deep Research to challenge itself. Run a follow-up prompt: "What are the three strongest arguments that this TAM estimate is too high? What would a sceptical investor say?" This surfaces weak assumptions before someone else does.

If you need outside validation beyond internal checks, SaaS Hackers also maintains a vetted list of B2B SaaS SEO agencies and specialist partners across growth disciplines via its top agencies directory, which can be useful when pressure-testing your category positioning and go-to-market assumptions.

Using the Output in Pitch Decks and Board Packs

Market sizing in a pitch deck needs to be defensible, not just impressive. When presenting TAM/SAM/SOM figures produced with Deep Research:

• Always attribute the underlying source, not the AI tool. "According to Grand View Research" not "according to ChatGPT."
• Show the methodology. Investors who see a £5B TAM with no working visible will ask how you got there. Have the answer ready.
• Present SAM as your real operating market. TAM is context. SAM is what you are actually going after.
• Anchor SOM to a named comparable. "We are targeting 2% of SAM in five years, consistent with [Competitor X] at a comparable stage."

SaaS Hackers recommends treating the Deep Research output as a first draft, not a final deliverable. It gets you 70% of the way there in 10% of the time. The remaining 30% is your judgment, your market knowledge, and your verification work. If you are building the broader growth system around that research, the SaaS Hackers resources section and guides hub are useful next stops.

FAQs

What is ChatGPT Deep Research and how does it differ from standard ChatGPT?

ChatGPT Deep Research is a mode that runs autonomous multi-step web searches before generating a response. Standard ChatGPT answers from training data alone. Deep Research actively retrieves and synthesises current sources, making it significantly more reliable for tasks like market sizing where data recency matters.

Can ChatGPT Deep Research replace a market research agency for B2B SaaS sizing?

No, but it replaces the first-draft phase that typically costs days of analyst time. Deep Research produces a structured, cited starting point. A market research agency adds primary research, proprietary databases, and validated methodology. For early-stage sizing to inform a pitch or product decision, Deep Research is sufficient with proper verification. If you do want specialist support, you can browse SaaS Hackers’ curated find an expert directory.

How accurate is ChatGPT Deep Research for TAM estimates?

Accuracy depends on data availability in your category. Well-documented markets (CRM, HR tech, cybersecurity) return tighter estimates because more public data exists. Niche or emerging categories will have wider uncertainty ranges. Always cross-check figures against at least one independent source and use ranges rather than point estimates in any formal document.

What ChatGPT plan do I need to use Deep Research?

Deep Research is available on ChatGPT Plus, Team, and Enterprise plans. It is not available on the free tier. As of 2024, Plus users have a monthly usage limit on Deep Research queries, so batch your market sizing sessions rather than running multiple exploratory prompts.

Is the output from ChatGPT Deep Research suitable for investor due diligence?

The output is a starting point, not a final submission. Investors will probe the methodology and sources behind any market sizing figure. Use Deep Research to structure your analysis and identify sources, then verify each figure independently before including it in materials that will face scrutiny.