Market research with sources. And how sure the answer is.

Research is a DFIRST skill that answers a market, trend or audience question for a decision you need to make. Firsty, the DFIRST agent, writes a short plan, reads the web and your files, and gives an answer with how likely it is, how strong the proof is, three to five supporting points with sources, and what would change it.

Every number has its source
Only what was actually read gets quoted
The answer says how strong the evidence is

Works with

Examples. What you ask, what you get back.

Should ZESTA launch in Texas next spring, or wait a year? We need to tell the distributor by January.

A clear answer with both measures, four supporting points graded by strength, and the two things that would change it. Every number points to the file it came from.

ResearchShould ZESTA launch in Texas in spring 2027?

Answer

Yes, but start in Austin and Houston, not the whole state. The pilot shows demand, and the distributor can stock two cities in April. Nobody has confirmed cold shelves for a state-wide launch.

How likely: Likely, if the April shelves hold
How strong the proof: Moderate Three own files and one partner proposal

What supports it

Strong Repeat orders in the Austin pilot rose every month from June to September. ZESTA pilot sell-out, Q3
Moderate The distributor offers cold shelves in 40 stores across two cities from April. Distributor proposal, Texas
Moderate Grapefruit outsold Tangerine two to one in the pilot, so the pink can leads. ZESTA pilot sell-out, Q3
Weak 9 of 60 US shop reviews ask for a 12-pack. US shop reviews export, Oct

What would change the answer

If the distributor cannot confirm the April shelves by 15 January, wait for autumn. If Austin repeat orders fall in Q4, open one city only.

ZESTA pilot sell-out, Q3.xlsxDistributor proposal, Texas.pdfUS shop reviews export, Oct.csv

Compare what customers say about our four SOL CLUB products. What do they love, what annoys them, what do we fix first?

A comparison table built from the review export. The first three rows came back for a check before the fourth ran, and each row says how much proof sits behind it.

ResearchSOL CLUB products: what customers say

Product	Most praised	Most common complaint	Reviews read	Proof
SPF 50 tube, 100 ml	No white cast	The cap opens in the bag	212	Strong
SPF 50 spray	Easy to put on children	Runs out fast	148	Strong
SPF 30 face fluid	Works under make-up	Too little in the bottle	61	Moderate
After-sun gel	The cooling smell	Sticky on hot days	19	Weak

Fix first: the tube’s cap, named in 42 of 212 tube reviews. You saw the first three rows before the last one ran.

SOL CLUB reviews export, Oct.csvCustomer service log, Q3.xlsx

Why did sales of KINETRA trail socks drop in the Midwest last quarter? I need reasons we can test before Q1.

The likeliest reason first, then three hypotheses graded by the proof behind each, each with a simple test. What would overturn the answer is named at the end.

ResearchKINETRA trail socks in the Midwest: why Q3 sales fell

Answer

Most likely a stock gap, not lost demand. Sales fell 18% against Q2, and most of the fall sits in the six weeks when the distributor had no medium sizes. Treat each reason below as a hypothesis to test.

How likely: Likely for the stock gap, unclear for the rest
How strong the proof: Moderate Own sales and stock data, no shopper survey

Hypotheses to test

Strong Medium sizes were out of stock for six weeks. Test: compare weeks with and without stock. Distributor stock report, Sept
Moderate The 3-pack went from $24 to $28 in July. Test: two weeks at the old price in two stores. KINETRA price list, July
Weak Fewer trail races in the region after August. Test: ask three store managers. Store visit notes, Oct

What would change the answer

If sales stay low now that medium sizes are back, the stock gap is not the main reason, and the price test matters more.

KINETRA sell-out by region, Q3.xlsxDistributor stock report, Sept.pdfKINETRA price list, July.pdf

How it goes. The same steps every time.

Firsty follows the skill’s steps in order and stops where the choice is yours. Nothing paid starts without your yes.

1
Firsty pins down your decision and writes a short research plan
2
Firsty reads the web, your files and other sources
Asks before anything paid. Small quick checks just run
3
Firsty fills the first 3 rows of a comparison table first
You see 3 rows before the rest of the table runs
4
Firsty gives the answer, how likely it is and how strong the proof
5
You can ask more, strengthen a point or disagree with it

Done when

Every number has its source
Only what was actually read gets quoted
The answer says how strong the evidence is
It names what would change the answer

At a glance.

What Research needs from you and what it hands back.

You bring: A question and the decision it serves. The market and the time frame default to your brand’s market and the last 12 months.
You get: An answer with how likely it is and how strong the proof is, three to five supporting points, what would change it, and the report with its sources.
Sources: The web, your files in the Data Room and other sources. Every number has one, and only what was read is quoted.
Comparisons: For a wide question, a table. You see the first three rows before the rest is filled.
Questions first: At most three short ones, each with a default you can keep.
Paid steps: Small, quick checks just run. Anything paid waits for your yes.

Part of this skill. Each task also has its own page.

When you ask for one of these on its own, Firsty runs it with this skill.

Deep research Go from a question to an evidence-backed brief.

Trend discovery Find signals worth paying attention to.

Market validation Put an idea next to the available evidence.

Customer insights Hear what people are actually telling you.

Questions people ask.

About Research and how it fits the rest of DFIRST.

How do I do market research with AI?

Ask the question together with the decision it serves, such as whether to open a new region next spring. Firsty writes a short plan, reads the web, your files and other sources, and answers with how likely the answer is, how strong the proof is and the sources behind each point.

How do I know how much to trust the answer?

Every answer carries two measures: how likely it is, and how strong the proof is, marked strong, moderate or weak. It also names what would change the answer. When the evidence is thin, the page says so instead of sounding sure, and you can push back on any point.

Why do I see only three rows of a comparison at first?

Firsty fills the first three rows of the table and stops, so you can check the columns and the sources while changes are still easy. Then the rest of the table runs. Small, quick checks never wait for you; anything paid always does.

Can it use our own sales data?

Yes, when the files are in your Data Room, such as a sell-out spreadsheet or a stock report. Firsty reads them next to outside sources and cites them like any other. It cannot see figures that are neither in your files nor in sources it can read, and it says when that leaves a gap.

Can it find trends or test a product idea too?

Yes. The same skill handles trend questions, idea checks and customer feedback. Each of those also has its own page: trend finder, idea validation and customer insights. Whatever the question, it goes through the same steps and checks.

See all 27 skills