Build a Competitor Report That Emails You Every Monday

The Monday morning edge

Pricing and ordering decisions get made every week against what your competitors are doing — who dropped prices, who's suddenly hiring five techs in a market you share, whose lot shows inventory piling up. You can't price well blind. And the problem in most businesses is that watching competitors is nobody's actual job, so it happens when someone has spare time, which is never.

The fully-built version of a system like this uses scrapers and change detection. You don't need any of that to start. In this guide you'll build the starter version in about an hour: a sources list, one analyst prompt, and a 15-minute Monday ritual that produces a 5-bullet brief. No code. No scrapers. Then you'll see the honest path to automating it — and the test that tells you whether to bother.

Step 1: Build your sources list

Open a Google Sheet (or a notes doc — anything you'll actually keep). You're going to list 5–10 URLs. Not fifty. Ten is the ceiling, because you'll be checking these by hand at first, and a list you won't check is worthless.

Good source types:

1. Competitor pricing or product pages — the 2–3 competitors who actually take deals from you. Their pricing page, their inventory page, their "specials" page.
2. Ad libraries — Meta Ad Library (facebook.com/ads/library) and Google Ads Transparency Center (adstransparency.google.com) show you every ad a competitor is currently running, free, no login. This is the single most underused source on this list.
3. Review pages — their Google Business profile or Yelp page. New 1-star reviews tell you where they're fumbling. A run of 5-stars naming an employee tells you who their best people are.
4. Job boards — their Indeed or LinkedIn jobs page. Hiring is strategy leaking in public. Five service tech openings means a service push. A "second location manager" posting means expansion.
5. One industry news source — one trade publication or association newsletter. Just one.

Now the part most people skip, and it's the part that makes this work. For every URL, fill in one more column: "What decision does this feed?"

If you can't fill in the decision column for a URL, delete the URL. A source that doesn't feed a decision is entertainment.

Step 2: The analyst prompt

This is the engine. It turns raw pasted content into a brief you'll actually read. It works in Claude (claude.ai) or ChatGPT (chatgpt.com) — free plans handle it, though paid plans give you longer pastes and better output.

Copy this into a note somewhere permanent. You'll reuse it every week. In Claude, save it as the instructions in a Project; in ChatGPT, save it as a custom GPT or just keep it in a doc. Fill in the bracketed sections once.

You are my competitive intelligence analyst. I run [YOUR BUSINESS —
e.g., "a 4-location equipment dealership in the Midwest"]. My main
competitors are [COMPETITOR NAMES].

Below I'm pasting this week's content from my tracked sources
(pricing pages, ad libraries, reviews, job postings, industry news).
Last week's brief is also pasted at the bottom so you can compare.

Produce a Monday analyst brief. Rules:

1. ONLY report CHANGES. If a price, ad, posting, or review situation
   is the same as last week's brief, do not mention it. If nothing
   changed at a source, skip it silently.
2. Maximum 5 bullets. If there are more than 5 changes, keep the 5
   most important and add one line: "Also noted: ..." with the rest
   in a few words each.
3. Every bullet gets a SO-WHAT. Format each bullet exactly:
   - **[Source]:** what changed → why it matters to my business →
     suggested action or "no action, keep watching."
4. Flag anything URGENT (a price undercut on a product I sell, a
   hiring move in my territory) with the word URGENT at the start of
   that bullet.
5. No filler, no throat-clearing, no summary paragraph. Bullets only.
6. If the pasted content is too thin to judge a source, say
   "insufficient data" for that source rather than guessing.

=== THIS WEEK'S CONTENT ===
[PASTE CONTENT HERE — label each chunk with its source, e.g.
"SOURCE: Competitor A pricing page" then the pasted text]

=== LAST WEEK'S BRIEF ===
[PASTE LAST WEEK'S 5 BULLETS HERE — first week, write "This is
week 1, no prior brief. Establish the baseline: summarize the
current state of each source in 5 bullets so we can track changes
from here."]

The "only report changes" rule is what separates a brief you read from a report you archive. Week one is a baseline. Every week after, you only hear what's new. That's how a real analyst works.

Step 3: The Monday ritual (15 minutes)

Here's the whole workflow. Monday morning, coffee in hand:

1. Open your sources sheet. Open each URL.
2. On each page, select the relevant content and copy it. Pricing tables, the ad library results, the newest reviews, the job listings. Ugly formatting is fine — the model doesn't care.
3. Paste each chunk into the prompt under its source label.
4. Paste last week's brief at the bottom.
5. Send. Read your 5 bullets. Paste them into an email to yourself (or your team) with the subject "Competitor Brief — [date]." That email thread becomes your archive, and next week's "last week's brief."

To be straight about it: this is the training-wheels version, and that's the point. Doing it by hand for a month proves the thing automation can't — whether you'd actually read this report and act on it. Plenty of people automate a report first and discover they ignore it. That's backwards. Prove the habit with 15 minutes a week, then automate the habit.

Step 4: The automated step-up

After a month of Mondays, if you're reading the brief and it's feeding decisions, automate the boring half. Honest options, easiest first:

• ChatGPT Tasks (paid plan): you can schedule ChatGPT to run a prompt on a schedule — including "browse these URLs and produce my brief" — and it notifies you when it's done. Setup takes minutes. The catch: its browsing gets blocked by some sites, and it can't log into anything.
• Claude with scheduled runs: Claude can fetch and read public web pages in a conversation, and paired with a Project holding your analyst prompt, your Monday ritual shrinks to "open the project, say go." Scheduling options depend on your plan and setup; the paste-free version alone saves most of the 15 minutes.
• Zapier or Make + a spreadsheet: keep your URLs in a Google Sheet, have Zapier or Make fetch each page weekly, push the content to Claude or ChatGPT through their API step, and email you the result via Gmail. This is the most reliable DIY automation, but it's a few hours of fiddly setup and both tools' useful tiers are paid (roughly $20–$30/month).

The honest limits, in one paragraph: plenty of sites block automated fetching — dealer platforms, some ad libraries, anything behind a login — and no legitimate no-code tool gets past a login wall. When a source blocks automation, keep that one manual and automate the rest. A half-automated report you receive beats a fully-automated one that silently breaks.

Step 5: The kill test

Put a note in your calendar one month out: "Did the competitor brief change a decision this month?"

A price move you countered. An ad angle you copied or avoided. An order you sized differently. If yes — keep it, tune the sources, maybe automate more. If no — kill it. Don't tweak it, don't feel bad, kill it. A report nobody acts on isn't intelligence; it's homework. The kill test is what keeps this from becoming one more thing you maintain out of guilt.

Step 6: What the fully-built version looks like

For most owners, the ritual-plus-partial-automation version is genuinely enough. But so you know what's up the road: the fully-built version runs real scrapers on a schedule against every source, does change detection (it stores what each page said last week and diffs it — so nothing depends on anyone's memory), pushes only the changes through the analyst prompt, and lands a formatted brief in your inbox every Monday at 7 a.m. whether anyone remembers or not. It handles the sites that block casual tools, and it keeps a searchable history of every competitor move going back months. How automated reports actually work covers the architecture, what to monitor by function, and a fill-in spec.

Built properly, a system like that is typically a $3,000–$8,000 one-time project depending on how many sources and how hostile they are to scraping, plus a small monthly cost for the tools that run it. Compare that to enterprise competitive-intel platforms that charge that much every year, forever. But — kill test first. The $0 version you just built is how you find out whether the $5,000 version would pay off.

Tool names, plans and prices in this guide were checked in August 2026. AI products change fast — verify anything before you buy it.

Where Hoven fits

If you want help deciding whether this — or something else entirely — is where AI pays off in your business, that's the job of my audit. It's $999/day, most take 1–5 days, and you get a prioritized plan plus a "don't build this" list, and if it isn't worth more than you paid, you don't pay. Details on the pricing page, or get in touch.