Every competitor you have publishes a running account of its strategy. Prices and how they move. Product range. Open roles. Delivery promises. Payment terms. Which markets they entered this year and which they quietly left. None of it is confidential. All of it is on their website.

Most companies read it once a quarter, usually the week somebody has to prepare a board deck.

What the public record actually contains

Pricing, and more importantly the pattern of pricing. A competitor discounting every third week has told you something about its cash position and its inventory that will not appear in any filing for a year.

Assortment. What they add and what they drop is a product roadmap, read in retrospect. A category quietly disappearing is a decision somebody made and lost an argument over.

Hiring. Fifteen open roles in logistics and none in sales is a company building capacity ahead of demand. The reverse is a company trying to sell its way out of a problem.

Terms and policies. Changes to returns, delivery windows or payment terms are usually the first externally visible sign of a margin problem, and they show up months before anyone talks about it.

Customer reviews, though not the star rating. The complaints. The same complaint appearing forty times is a product gap you can price against.

Why it does not happen

Three reasons, and none of them is technical.

It has no owner. Competitive intelligence sits between marketing, product and finance, which means it belongs to nobody and gets done by whoever has a quiet week.

It is run as a project instead of a process. A consultant delivers sixty slides, the slides get read once, and eighteen months later every number in them is wrong.

The manual version does not scale. One analyst can check ten competitor pages a week and will hate every minute of it. The frequency that produces decisions is daily across a few hundred pages, and the moment you say that out loud it stops being a staffing question and becomes a software one.

What it costs to run

This is where most conversations end, generally because somebody has quoted an enterprise platform at five figures a month and the room has concluded that the whole idea is expensive.

The honest picture is different. The expensive parts of competitive intelligence are deciding what to track and acting on what you find. Both are human. The collection itself has been a commodity for years.

A company monitoring a few hundred competitor pages daily is looking at a modest piece of engineering and infrastructure that runs to tens of dollars a month. The main technical line item is the connection layer, because a single server making the same requests day after day from one address gets rate limited within a week. Spreading those requests across datacenter proxy servers resolves it at a cost most finance teams would not notice in a stationery budget.

Whether you build or buy is a question about your team, not your budget. With engineers in house this is a week of work and a small recurring cost. Without them, buy the platform, but negotiate knowing that you are paying for the interface and the support rather than for the data.

The part to settle before anyone starts

Two boundaries, both worth writing down before a line of code exists.

Collect what any visitor can see without logging in. Creating accounts to reach material behind a login changes the character of the activity entirely, and no commercial advantage justifies that exposure.

Stay away from personal data. Competitor prices are not personal data. The names and contact details of their customers are, and gathering those without a lawful basis creates a liability that will outlive whatever edge you thought you were buying.

Beyond that, respect the technical signals sites publish about automated access. It costs nothing and it keeps the activity boring, which is the objective.

Where to start

Pick ten competitor pages that would change a decision if they changed. Ten, not a hundred.

Track them daily for a month and route every change to one named person. At the end of the month ask that person which changes actually led to a decision. Usually two or three did, and they are never the ones anybody predicted.

Then expand only in the direction those two or three point. Companies that begin with a full intelligence programme end up with a dashboard nobody opens. Companies that begin with ten pages and one owner end up with something the executive team reads on a Monday morning.