Marketing · Intermediate
Map competitor positioning from their public pages
Every competitor says they're "easy", "powerful" and "trusted by leading teams". Reading their sites one at a time, the differences blur. Side by side, they become obvious: who each one is really talking to, what they can prove, and what nobody is saying at all. This guide builds that grid from pages you've collected yourself, and keeps what competitors actually say apart from what you are guessing.
What AI is for here, and what it isn't
Positioning analysis is mostly close reading: what each competitor claims, for whom, with what evidence, and in what language. It's tedious to do well across five or six companies, which is why it's often done from memory and impressions.
AI is good at the close reading. Given the text of each competitor's pages, it will extract the same fields for each one and lay them out side by side, consistently. The patterns, such as everyone claiming the same thing or a segment nobody addresses, show up quickly in a grid.
Two cautions shape the method. First, the AI works only from pages you collect and date. Its own knowledge of a competitor may be out of date or simply wrong, and positioning changes often. Second, it will infer: one enterprise logo becomes "targets enterprise". Inferences can be useful, but they must be labelled as such. Where you'll compete, and where you won't, is your decision.
Collect and date the pages
For each competitor, copy the text of the same set of pages: home page, main product page, pricing page, and one or two customer stories. Paste each into a document with the URL and the date you captured it. Save screenshots if you can; pages change, and you may want to show what a page said in March.
Use the same pages for every competitor. Comparing one company's home page with another's pricing page tells you about page types, not positioning.
Add your own positioning statement and your own pages, collected the same way. You belong in the grid too.
Extract the positioning
Run one competitor at a time.
Extract this company's positioning from the page text below. Use only
the text provided; do not use anything you know about the company.
COMPANY: [name] PAGES: [list with URLs and capture dates]
FIELDS (quote the page wherever possible, with page name):
1. HEADLINE CLAIM: the main promise, as worded on the home page.
2. AUDIENCE: who they say it's for (role, company size, industry).
3. CATEGORY: what they call themselves ("scheduling platform",
"workforce tool").
4. PROBLEMS NAMED: the pains they say they solve.
5. DIFFERENTIATORS: what they claim makes them different.
6. PROOF POINTS: numbers, customer names, awards, certifications,
testimonials. Note what each proves.
7. PRICING LANGUAGE: model, tiers, anchors, free trial, "contact us".
8. CALL TO ACTION: what they ask visitors to do.
9. TONE: three words, with an example line.
Mark each field STATED (the page says it) or INFERRED (your reading
of the evidence), and give the reasoning for every inference.
If a field isn't covered by the pages, write NOT FOUND.
PAGE TEXT:
[paste]
"Not found" is useful information. A competitor with no pricing page and "contact us" everywhere is telling you something about who they sell to.
Build the grid and read it
Combine the extractions below into a grid: companies as columns
(including us), fields as rows. Keep STATED/INFERRED labels.
Then list:
1. CROWDED CLAIMS: claims made by three or more companies.
2. UNCLAIMED: problems, audiences or proof points none of them use.
3. CONTRADICTIONS: where a company's own pages disagree (e.g., the
home page says "enterprise", pricing says "teams of 5–50").
4. OUR OVERLAP: where our positioning is hard to tell apart from a
competitor's, quoting both.
Don't recommend a strategy.
Check every STATED cell against the page text before you rely on the grid, and treat INFERRED cells as questions. Then make your call: where you'll compete head-on, where you'll differentiate, and which segments you'll leave alone.
Keep it current
Positioning changes, sometimes monthly. Put a capture date on the grid's title, and plan to re-collect the pages quarterly, or whenever a competitor launches or rebrands. Variation B compares two captures of the same competitor and lists what changed.
Worked example: scheduling software
Fictional example. A marketer at a scheduling software company maps three competitors (A, B and C) and their own product, from pages captured on the same day.
The grid shows the pattern at once:
- Crowded: all three competitors and the marketer's own company lead with "easy". Two use nearly identical headlines ("Scheduling made simple" and "Simple scheduling for busy teams").
- Unclaimed: none mentions reliability, uptime or compliance with labor rules on shift scheduling, although two name "frontline teams" as an audience.
- Contradiction: for Competitor B, the extraction lists AUDIENCE: "enterprise (INFERRED: home page shows logos of two large retailers)". B's pricing page, STATED: "Built for teams of 5 to 50." The logos are customer stories from single store locations. The inference was wrong, and the pricing page settles it.
The marketer's decision, which the grid doesn't make: stop leading with "easy", which everyone says, and position on accuracy for frontline shift scheduling, with labor-rule checks as the proof point. Before anything public compares the product with A, B or C by name, the claims go to legal review.
Record the grid
Positioning grid — captured [date] Analyst: [name]
SOURCES
[Company] — [page] — [URL] — captured [date] — screenshot [file]
GRID
Field | Us | A | B | C
Headline claim | | | |
Audience | | | |
... (each cell: text, STATED/INFERRED, page)
FINDINGS
Crowded claims: [ ]
Unclaimed: [ ]
Contradictions: [ ]
Our overlap: [ ]
DECISION (analyst / team)
Compete on: [ ] Avoid: [ ] Leave alone: [ ]
Comparative claims sent to legal: [date]
Next capture due: [date]
What stays with you: where you will compete, and where you won't.