How to Gather Market Intelligence: Start With the Decision, Not the Data

Why adding sources never fixes it

Most teams answer "how do we gather market intelligence" by adding inputs. Another newsletter, another keyword alert, another competitor page on the watch list. Six months later the feeds have tripled and the decisions have not improved. Volume was never the constraint.

Gathering is a method with a shape. You name a pending decision, specify what evidence would move it, collect against that specification, then hand the finding to the person who owns the call. Collection without that shape produces a reading list that nobody reads.

The test takes ten seconds. Name one decision your company made differently last quarter because of something you gathered. If nobody in the room can answer, you are running a clipping service. What follows is the method that fixes it, in the order you should run it.

Start with the decision, not the sources

Every useful gathering effort begins with a decision that is pending and a date it has to be made. Do we match the competitor tier that launched last week before renewals open in November. Should the roadmap fund the integration mid-market prospects keep asking about. Which objection does next quarter’s messaging have to answer first.

Write the decision down. Then write down what evidence would change it. That second sentence does most of the work in this entire method. If you cannot say what would convince you, no volume of collection will, and you will keep gathering forever because nothing ever feels like enough.

The specification also bounds the job, which is the point. A pricing decision does not need brand sentiment or share of voice. It needs competitor list prices, the discounts your reps are actually granting and what buyers say about value when no one from your company is in the room. Three targets beat thirty feeds.

Give the decision an owner before you collect anything. Intelligence that arrives without a named recipient dies in a channel. If the head of product is not expecting this by the fifteenth, you are writing into a void.

Where market intelligence actually comes from

Sources fall into three tiers. Most teams work the first two and skip the third, which is where the timing advantage lives.

The first tier is what you already own. Win and loss notes, support tickets, call recordings, churn interviews, the pricing exceptions your finance team approves. This is the cheapest intelligence available and the most neglected, because it arrives as operational exhaust rather than as a report. Read thirty lost-deal notes in one sitting and patterns appear that no dashboard surfaces.

The second tier is published and public. Competitor pricing pages, release notes, changelogs, job postings, filings, review sites, conference agendas, analyst coverage. Job postings are the cheapest forecast in the set. A competitor hiring six field engineers in Germany has told you its next market months before any announcement does.

The third tier is unprompted public discussion. Forums, subreddits, Slack and Discord communities, review threads, comment sections where your category gets argued about. Nobody wrote those posts for your benefit, which is exactly why they are worth reading. The method for studying that discussion deserves its own discipline.

The layer most gathering plans skip

Tier one tells you what already happened inside your funnel. Tier two tells you what companies decided to publish about themselves. Neither is dishonest, but both are curated. The third tier is not curated by anyone, and that is its whole value.

This is community intelligence, and it behaves differently from the social listening most tools sell. Listening counts mentions of your brand. Community intelligence reads the conversation your buyers have with each other, most of which never mentions you at all. Someone posting at 11pm about the workaround they built for a competitor’s export limit is describing a market gap in public.

Watch four things specifically. The question that gets asked repeatedly, because repeated questions are unmet demand. The workaround post, because it maps the exact edge of what the category currently does. The comparison request, because it names your real competitive set rather than the one in your deck. And the vocabulary, because the words buyers use rarely match the words your website uses.

Repetition is the signal here, not volume. Forty mentions from one angry customer is noise. Six independent people describing the same friction in three different communities over two weeks is a finding. That distinction separates intelligence from research in practice.

The five-step gathering loop

Step one, frame. State the decision, its date and its owner in three lines. Then state what evidence would change it. Do not skip to collection because framing feels like overhead. Framing is the only reason the collection will be usable.

Step two, scope the sources. Pick the five or six specific places most likely to hold that evidence, drawn from all three tiers. Name the venues, the time window and the vocabulary you will search on. A scope you can write down is a scope someone else can repeat next quarter.

Step three, collect on a schedule you can actually hold. Ninety minutes a week beats an eight-hour sprint every March. Capture the source link, the date and the raw quote every time. Paraphrase at collection time and you will spend the analysis step arguing about what somebody meant.

Step four, corroborate. Nothing becomes a finding on one source. Require two independent tiers before you call it real. A competitor price change on a pricing page plus three community posts about the change equals a finding. Either one alone stays a hypothesis with a date on it.

Step five, write the finding and route it. Five lines: the claim, the evidence behind it, your confidence, the recommended action and the owner. That format forces a position. "Mentions of onboarding difficulty rose" is reporting. "Three of our last eight losses cited setup time and two competitors now advertise guided setup, so we should fund it this quarter" is intelligence.

How to make it continuous instead of heroic

Most gathering programs die because someone built them as a project. Run yours on a cadence instead. A weekly pass of sixty to ninety minutes against your live decisions. A monthly synthesis that consolidates findings into themes. A quarterly review that kills the sources which produced nothing and adds the ones you kept wishing you had.

Store findings in one place, tagged by competitor, theme and the decision they support. Tools matter far less than retrievability. A finding you cannot pull up in ninety seconds during a pricing argument is not an asset, it is a document. Feed the strong ones into your recurring market intelligence report rather than starting that report from scratch each month.

Automate the collection, not the judgment. Alerts, scrapes and saved searches are good at watching things change. They are bad at deciding what a change means for your roadmap. Keep the boring monitoring on a machine and keep the interpretation on a person who understands your buyer.

Two habits keep the whole loop honest. Log what you predicted and check it a quarter later, because calibration is the only way to learn whether your sources are any good. And record the findings that went nowhere. Knowing which competitor moves turned out to be noise is worth as much as knowing which ones mattered.

What to do on Monday

Pick one decision your team has to make in the next sixty days. Write the evidence that would change it. Choose five sources across the three tiers, block ninety minutes on Thursday and collect against that list. Corroborate, write one five-line finding and send it to the person who owns the call.

That single pass will teach you more about your gathering gaps than a quarter of planning. You will discover which tier you have been ignoring, usually the third. Pair the result with structured competitive research once the rhythm holds.

Gathering market intelligence is not a collection problem. It is a specification problem wearing a collection problem’s clothes. Decide what you need to know and who will act on it, and the sources narrow themselves.