Public-source market research, scoped first
An AI market research tool that keeps claims tied to sources
Start with a decision, a market boundary and the evidence you need. ZeroTwo's market-research workflow is designed to organize accessible public information into a landscape, a competitor matrix and a source ledger. Treat the result as time-bound secondary research and check important claims against the original sources.
Disclosure: ZeroTwo publishes this page and sells the product it describes. No finished example study is published yet, so nothing below is a result. It is the specification a run should be held to.
Scoping brief
Template, not a result- Decision
- The choice this research has to inform
- Market boundary
- Geography, category, audience, use case
- Out of scope
- What should be excluded, and why
- Comparison fields
- What every competitor row must answer
- Evidence bar
- What counts as proof, what stays an inference
Fill these in before the run. Each row is a rule you can check the result against.
The workflow at a glance
- 3
- Files in the target output: landscape, competitor matrix, source ledger
- Public
- Evidence base: pages anyone can open, not paywalled or private data
- Secondary
- Research type: no surveys, interviews or fieldwork
- $14.99/mo+
- Work is on paid plans, starting at Plus; the Free plan does not include it
Choose the decision and the research boundary
A narrow brief beats a broad one. These six fields settle most of what a run will read, what it will skip and how you can judge it.
The decision
Name the choice the research must help you make. A decision tells the run what to leave out; a topic does not.
- Weak brief
- Research the project-management software market.
- Sharper brief
- Decide whether to build for agencies of 10 to 50 people or integrate with an existing tool they already use.
Market boundary
Geography, category, buyer and use case, written as inclusions and exclusions.
- Weak brief
- Global, all sizes.
- Sharper brief
- English-speaking markets, agencies, billed per seat.
Comparison fields
The questions every competitor row has to answer, fixed before research starts.
- Weak brief
- Compare features.
- Sharper brief
- Target customer, pricing unit, integrations, positioning claim.
Exclusions
Say what is not a competitor, so the run does not quietly widen the field.
- Weak brief
- (left blank)
- Sharper brief
- Enterprise suites and open-source projects with no hosted plan.
Evidence bar
Decide which claims need a first-party page and which may rest on secondary coverage.
- Weak brief
- Use good sources.
- Sharper brief
- Pricing and features: the vendor's own pages. Market patterns: label as inference.
Source rules
Ask for stable URLs, a retrieval date for each, a source type (first-party, press, analyst, community), and an explicit label wherever the answer is inference or a gap. Rules written into the brief are rules you can check the result against.
- Weak brief
- Add citations.
- Sharper brief
- One source ID per claim, with URL, retrieval date and source type in the ledger.
The example wording is illustrative, written for this page. It is not taken from a real study.
Scope a market studyKnow when you need research this workflow cannot do
Public secondary research is good at mapping what vendors say about themselves. It is weak at everything behind that. And a gap in public evidence is not proof that something is absent.
Choosing a tool, not scoping a study? Compare market research tools by research job.
Why customers choose, and what they would pay
Public pages show what vendors say, not what buyers think. Motives and willingness to pay need interviews or surveys.
Customer research and interviewsPaywalled or proprietary data
A run reads what is publicly accessible. Licensed reports or internal numbers are only usable if you supply them, and only if your licence permits that.
Market size
Any size figure rests on assumptions. Ask for the method and each input with its source; a range or a plain “not established” beats a precise number nobody can trace.
Market-sizing research
Three files that keep the evidence attached
A recommendation is only useful if a reviewer can trace what is known, what is inferred, what is missing and what is likely to change. This is the output the workflow is designed to return.
- 01Report
Market landscape
A document that sets out the boundary, the evidence, the competitor clusters, a recommendation, risks, assumptions and gaps.
Holds
- Boundary and exclusions
- Competitor clusters
- Recommendation, marked as interpretation
- Assumptions and gaps
- 02Table
Competitor matrix
A structured table with one row per competitor and the same fields on every row, so differences are comparable.
Holds
- Target customer
- Positioning
- Pricing basis and unit
- Capabilities
- Source, date, confidence
- 03Record
Source ledger
A row-level record of every public page used, so a reviewer can reopen the evidence instead of trusting the summary.
Holds
- Source ID
- URL
- Retrieval date
- Source type
- Accessible or not
Follow one claim back to its source
Every material claim should walk the same five steps. If one is missing, the claim is a lead, not a finding.
Claim
One statement the report relies on.
Source ID
Points to a row in the ledger.
URL + date
Where it was read, and when.
Evidence type
Direct, interpretation or inference.
Confidence
High, medium or low, with a reason.
- Claim
- [one sentence the landscape depends on]
- Source
- S-[n] · [publisher] · [URL]
- Retrieved
- [date the page was read]
- Evidence type
- direct | interpretation | inference
- Confidence
- [high | medium | low] because [reason]
Direct evidence
The source says it. A price on the vendor's pricing page, a feature in its documentation. Quote it, link it, date it.
Interpretation
What the pattern across sources suggests. Useful, but it is reasoning, so it is labeled as such and never presented as a fact about the market.
Compare competitors without inventing gaps
The matrix has the same fields on every row. When a field cannot be filled from public evidence, the cell says so.
| Matrix field | What each row records | When it is unknown |
|---|---|---|
| Target customer | Who the vendor says it serves, in its own words, with the page it said it on. | “Not stated publicly”. Not guessed from the logo wall. |
| Positioning | The headline claim, quoted or closely paraphrased, marked as the vendor's claim. | Left blank if the homepage makes none. |
| Pricing basis | The unit (per seat, per usage, flat, quote only), not a forced monthly figure. | “Quote only” or “not published”. Units are never silently converted. |
| Capabilities | What the vendor's own pages state, row by row. | “Not found in public sources”, which is not the same as “does not exist”. |
| Source and date | Source ID, URL and retrieval date for the row. | A row without a source is flagged, not filled. |
| Confidence | How direct the evidence is, with the reason. | Low, with the missing evidence named. |
Competitor inclusion is a method decision, not a ranking. Write the rule into the brief (for example, direct substitutes for one buyer in one segment), and ask the run to list the candidates it left out and why. Pricing units differ between vendors, so the matrix records the unit rather than forcing a single number.
From an open question to inspectable research
The run gathers evidence for the scope you defined and carries its provenance into each file.
Plan the investigation
The decision criteria become research questions, comparison fields and a source strategy.
Gather current evidence
Publicly accessible sources are collected, with first-party pages favored for product and pricing facts, and retrieval details recorded.
Separate fact from inference
Factual claims are tied to source IDs, gaps are shown, and synthesis or recommendation is marked as interpretation.
Build the decision files
The landscape, structured competitor matrix and complete source ledger come back together for review.
A source list versus a decision-ready landscape
Search-and-summarize is fine for orientation. It stops short when the result has to survive a reviewer asking “where did that come from?”
| Criterion | ZeroTwo workflow | Search-and-summarize |
|---|---|---|
| Starting point | Market boundary, decision, questions and source rules | A broad topic or a search query |
| Evidence record | Source IDs, URLs, retrieval dates, source types and gaps | Links gathered during individual searches |
| Finish line | Landscape, competitor matrix, source ledger, recommendation | Notes or a narrative that may need restructuring |
| Review | Trace claims, separating fact, synthesis and inference | Reopen sources and reconstruct how conclusions were reached |
This describes how the workflow is designed. The quality of any research result still depends on source access, scope, freshness and review.
Research cost follows the scope and the evidence bar
This runs as Work on a paid plan or with credits, not as a sponsored report. Work is listed for Plus ($14.99/mo) and above on the pricing page. We have not published an observed source-count or cost example, so use the levers below to keep scope in check.
Breadth
Each extra competitor, segment or geography multiplies the pages that have to be read.
Depth
More comparison fields per competitor means more sources per row.
Evidence bar
Requiring a first-party page for every claim costs more than accepting secondary coverage.
Revision rounds
Each pass over the files uses allowance again, so settle the scope before the first run.
Narrow the market first: a clear category, geography, buyer and decision removes irrelevant research before it starts. This workflow prepares research for review. It is not legal, financial or investment advice, and it does not claim exhaustive coverage.
Review it the way a skeptical reader would
A run gives you files to check, not a verdict to trust. These are the checks to apply when the files come back.
Claims trace to sources
Every material product, feature, pricing and market-pattern claim names one or more source IDs.
Facts and recommendations differ
Direct evidence, synthesis, inference, recommendation and missing evidence are visibly distinct.
Gaps stay honest
Market size, customer proof, pricing and prevalence are not invented to fill a weak source set.
Primary pages first
Product and pricing claims lean on the vendor's own product, pricing, documentation or policy pages where they exist.
Dates are visible
Each source carries a retrieval date, because pricing and product pages change after it.
Access limits are named
Inaccessible, paywalled or stale sources are listed as such rather than skipped.
Research like this is a snapshot. When the question becomes “what changed?”, move to ongoing competitor monitoring.
What people ask
Is this public-source research or primary customer research?
Public-source, secondary research only. ZeroTwo works from pages it can access. It does not run surveys, interviews or panels, and text an AI writes is not a response from your customers. For that, see customer research and interviews.
How are competitors selected?
By the rules in your brief. State what counts as a competitor and what does not, then ask the run to list the candidates it found and which it excluded, with a reason. Without inclusion rules the field drifts wider than the decision needs.
How current are pricing and capability findings?
As current as each source's retrieval date, which the ledger should record. Pricing and product pages change, so recheck anything that decides the outcome against the vendor's page before you act on it.
What happens when a source is inaccessible?
It should be listed in the ledger as inaccessible, with the reason, and the related claim marked as a gap or low confidence. A claim should not be filled in from memory because the page would not load.
Can the tool estimate market size without inventing numbers?
Treat any size figure as an estimate built on stated assumptions. Ask for the method, each input and a source per input. If public sources do not support a figure, the right output is a range or “not established”.
Is there a finished example study I can open?
Not yet. No completed landscape, matrix or ledger is published on this page, so the structures above are specifications to hold a run to, not results. You can scope a study of your own and review the output against the checks listed here.
Is this market-research workflow free?
This page does not describe a free run. The workflow runs as Work on a paid plan or with credits; Work is listed for Plus and above on the pricing page. There is a Free plan with no card required, but it does not include Work. Check pricing before you start.
Explore related workflows
- Compare market research tools
Choosing between tools rather than scoping a study. Organized by research job.
- Market-sizing research
For when estimating size is the decision, with the assumptions written down.
- Ongoing competitor monitoring
A study is a snapshot. Monitoring is the follow-on once the one-time research is done.
- Source-led research features
The broader Deep research feature. The plan comparison lists a limited amount on Free, 150 per month on Plus and more on higher plans.
- Market research tool guide
ZeroTwo's separate guide to structuring market research. This page covers the product workflow.