Evaluate Steve when competitive intelligence repeats or needs shared review

The three scenarios are illustrative. None describes a customer, and none carries a measured result. They exist so you can put a label on your own work.

Handle a simple one-off public-page check in chat

Open the page. If it is long, paste its text into a Claude chat and ask the narrow question the seller asked: which plan includes the feature, what the stated limit is, whether the page contradicts what the seller remembers. Then open the passage Claude quoted and confirm it says that before you repeat it.

The check is good when you can point at the sentence on the competitor's page that answers the question. This two-copy method cannot compare the current page with a prior version you cannot retrieve. Supply a saved copy or a reliable dated archive; if neither exists, do not claim what changed. If you expect the question again, save the page text with today's date; the competitor page changes lesson shows the two-copies method.

Recurring checks need explicit coverage

A weekly manual check can stop when the assigned person is away and no backup is named. If a plan change lands during that gap, it may be discovered only later. Two things make the manual version dependable: a saved copy of the last version to compare against, and a note of every check that did not happen.

Run it in a Claude Project so the competitor list, the pages and the last copies sit in one place; the Project setup lesson covers that once. Each Monday, give Claude the saved copy and the current text of each page and ask for the passages that differ. Check the differences against the two files yourself. A short list of changed passages is the useful result; a summary of the whole page means the question was too broad. A page that would not load is recorded as unchecked, never as no change.

For a cloud recurring check, use a Cowork scheduled task on a paid plan and give it accessible sources. A task that needs local files or apps runs locally and requires the relevant computer access. Anthropic's Cowork getting-started guide, checked October 12, 2026, documents cloud scheduling on paid Pro, Max, Team and Enterprise plans, with availability varying by platform. On a free plan or in plain chat, this weekly check remains manual.

For a competitor set whose pages you can keep up with, our recommendation is to start with that manual routine. The solo competitive intelligence lesson describes the weekly note that keeps it going and treats repeated misses as the signal to evaluate a tool. Competitor count on its own is not that signal.

Team work needs one version and a record of the decision

One team failure mode is parallel editing: when two people edit separate copies, the seller may receive the most recently sent copy rather than an agreed current version. The battlecard lesson covers what belongs on the card, and the battlecard feedback lesson covers how to check a seller's objection to a claim.

Anthropic's project visibility guide, checked October 12, 2026, documents Project sharing on Team and Enterprise plans, with "Can view" and "Can edit" permissions. On Pro, Max or Free, use a separate shared document when teammates need to edit the card. A small team with one named editor can work that way. Confirm whether the chosen document's history and approval features provide the record your team needs. If they do not, keep a short change log beside it.

Sort your own work before you evaluate a tool

A product marketer at a small company can carry all three patterns at once. One illustrative week holds a pre-call check on Tuesday, a Monday sweep of the same pages, and two sellers who want the card current. The pattern to evaluate a tool against is the one that keeps failing.

• List every competitive check from last month and mark each one as one-off, recurring or team
• For each recurring check, count the weeks it was skipped and note what changed in those weeks
• For each shared card or document, write who decides a disputed claim and where that decision is recorded
• Name the one pattern that costs the most time, and evaluate a tool for that pattern only

If the one-off checks are fine and the Monday sweep is what lapses, evaluate monitoring and leave the one-off work manual. The build versus buy guide compares the ongoing cost of each pattern, and the evaluation checklist says what to test before committing.

What to confirm with Steve during evaluation

Steve's product page describes automatically deploying and analyzing win/loss surveys. Confirm with Steve how competitor tracking cadence, scheduled reports and battlecard updates work for your team, and whether they fit your plan. The competitor digest lesson and the win/loss lesson show the corresponding DIY outputs.

For the team pattern, ask how a shared battlecard is reviewed before it is published and how the record shows who approved a change.

The pricing page is a quote-request form with no listed dollar price. If recurring collection or shared review is the burden, Steve's competitive intelligence product page is the next place to confirm current availability. Your team owns competitor scope and wording approval and should inspect the evidence, even where Steve proposes competitors or verifies claims.