How to keep a market map current without rebuilding it every month
A practical model for turning a one-off market spreadsheet into a maintained, source-backed view of companies, facts and changes.
Written by
Scraper.io
Editorial
Most market maps are accurate for one brief moment: the afternoon the analyst stops editing them. After that, companies launch products, change prices, reposition, merge and disappear while the workbook stays exactly where it was.
The usual response is another research sprint. That may refresh the cells, but it does not create a system that knows what changed. A maintainable market needs a different structure from the beginning.
Start with the decision, not the source list
A market is only useful relative to a question. “AI companies” is not a usable definition; “companies selling infrastructure for memory, retrieval or state in production AI systems” is closer. It creates an inclusion test that another researcher can apply consistently.
Write down who uses the market, which decision it supports, what belongs, what does not, which fields are required and how fresh each field needs to be. That specification becomes the stable object. Sources can change without silently changing the question.
- The market definition and explicit exclusions
- The entity type: company, product, person, claim or event
- The fields required for the decision
- The acceptable sources and expected refresh schedule
- What the reviewer will accept as evidence
Separate entities, facts, evidence and events
A spreadsheet normally compresses four different things into one cell. There is the entity being described, the current fact, the evidence supporting it and the event that changed the old value into the new one. Keeping those concepts separate is what makes history possible.
The company should have a stable identity even if its name or domain changes. A fact such as a public price should point to one or more evidence records. When the price changes, the prior fact is not erased; a new observation supersedes it and a before-and-after event records the transition.
A useful change record answers four questions: what changed, what was true before, what is true now and which evidence supports the change.
Timestamp observations, not just exports
“Last updated Tuesday” is not enough. One field may have been observed Tuesday while another has not been checked for three months. Freshness belongs at the value or evidence level, not only in the workbook title.
Record when each source was observed and keep failed checks explicit. A source that did not load is not evidence that nothing changed. This distinction prevents a broken collection process from looking like a quiet market.
Update affected facts instead of replacing the workbook
A maintainable market should make narrow updates. When one vendor changes packaging, update the supported commercial fields and create an event. Do not rebuild every row and lose the ability to explain why unrelated values moved.
Narrow updates also make review cheaper. The reviewer sees the old value, proposed value and evidence together. They can accept, reject or correct the proposal without re-reading the entire market.
Preserve uncertainty and disagreement
Public sources often disagree. A product page may call a feature generally available while release notes still call it experimental. Choosing one silently makes the table look cleaner and less trustworthy.
Keep both observations, label inference and expose the conflict. Empty is also a valid state: if the source does not support a field, leaving it blank is better than turning a plausible guess into a fact.
Deliver the change, not another dump
The maintained table is the reference surface, but the recurring product is the change brief. It should say which values changed, show the supporting evidence and explain why those changes match the customer's criteria.
That keeps the workflow bounded. People review the few rows that need attention while the full market remains available for search, filtering and export.
- A current, inspectable table
- A concise list of material changes
- Before-and-after evidence
- A clean export that does not trap the customer
- A feedback path for wrong, missing or irrelevant results
The important shift is from refreshing a document to maintaining a state. Once the definition, identities, evidence and events are durable, every future research pass improves the same market instead of producing another disconnected snapshot.