You will set up an AI-driven competitor analysis and turn it into messaging, creative angles and tests. You can refresh it monthly without starting from zero.
Decide what this analysis must change
Start with the decision you need to make. AI becomes useful when you point it at a choice, not a vague scan.
Pick one priority:
Ad angles to try next month
Product page messaging to rewrite this week
Content themes for the next four weeks
Note the metric you’ll judge with (e.g. CTR for ads, add‑to‑cart rate for PDPs, post saves for content). This keeps the AI’s output practical, not abstract. If you need a refresher on automation limits and how to validate, read What automated market research can and cannot tell you.
Build a focused competitor set
You don’t need a long list. Aim for 5–8:
Direct: same product and price band
Indirect: same job-to-be-done, different product format
Adjacent: upsell or downshift options your buyer compares against
Write your list with URLs for the home page, two bestsellers, and one landing page per competitor. Keep a date next to each link so you know what period the copy belongs to.
Collect real samples (don’t rely on AI to “know” them)
Give the AI primary material to analyse. Collect for each competitor:
Product page copy: headline, bullets, FAQs, guarantees, returns
Landing page/offer page: hero, value props, social proof snippets
3–5 recent social posts or ad creatives: headline, first line, CTA text
Email welcome or post‑purchase copy if publicly visible (e.g. via a test signup)
Save these in a doc with clear labels (brand, URL, date). Paste the plain text, then below it paste short quotes exactly as written. You’ll ask the AI to build tables and grids from these, not to invent them.
Use structured prompts, not open questions
Skip “what are their strengths/weaknesses?” Give the AI formats to fill. Here are prompts you can paste, one at a time.
1) Value proposition grid “From the quoted copy below, extract value propositions as a table: Pillar | Evidence/Proof | Where found (URL) | Quote. Do not invent claims. Only use text that appears verbatim in quotes. If uncertain, leave the cell blank.”
2) Feature → benefit → objection “From the product page quotes, create rows: Feature | Customer benefit | Likely objection | Copy snippet addressing the objection (verbatim). If no snippet exists, mark ‘No copy provided’.”
3) Brand voice summary “Summarise tone-of-voice in bullets: sentence length, formality, use of slang, use of numbers, risk-taking, emotion level. Include one short example quote for each.”
4) Angle-by-segment table “Using the samples, list up to 5 buyer segments (behaviour- or need-based). For each, propose 2 creative angles competitor copy already targets. Output: Segment | Angle | Supporting quote | URL.”
5) Claim–proof matrix “List all hard claims. Output: Claim | Type of proof used (review, certification, before/after, guarantee) | Verbatim proof line | URL. If a claim lacks proof, note ‘No proof shown’.”
6) Content themes with examples “Group the social/ad samples into 5–7 repeatable themes. Output: Theme | Purpose | Example post/ad headline (verbatim) | CTA type.”
These formats produce testable hypotheses, message pillars, an angle-by-segment table, competitor claims & proof, and content themes with examples — outputs you can act on.
Clean, compare, and de-duplicate
Once the AI returns tables, do three fast passes:
Trace to source: check each quote has a working URL and date.
Cross‑check claims: if the AI labelled a claim as ‘guarantee’, confirm the guarantee exists in the sample.
De-duplicate: combine near‑identical value props across brands and note who uses them strongest.
Keep all originals. You’ll reuse them when you refresh the analysis next month.
Turn insights into concrete tests
Translate the analysis into actions for the next sprint.
Ads
Pick 3 angles from the angle-by-segment table that your current ads do not cover.
Write one headline and one primary line per angle, staying in your own voice.
Define the CTA per theme from the themes table.
Plan a simple A/B rotation for a week.
Product pages
Take 3 value propositions with strong proof in the market but missing on your page.
Add one section per prop: headline, 1–2 lines, and a proof element (review quote, certification, guarantee line).
Use the feature → benefit → objection rows to tighten bullets.
Content
Choose 3 themes your channel mix can publish next week.
Draft one post per theme, and plan a light series.
If you need a push from analysis to execution, store your pillars and forbidden phrases in your Brand book and plan publish dates in Planning or your Content calendar. For video angles, see AI video is built scene by scene and One idea, nine aspect ratios.
Keep your own voice
Do not copy competitor phrasing. Strip brand names from any text you reuse as a working note. If you need to define tone and guardrails before drafting, set them once in the Brand book and have them open while you write. For how to set voice beyond adjectives, read Your brand voice is not a brief.
Set a refresh cadence and a light archive
A competitor analysis decays fast. Make it a light, recurring task.
Frequency: monthly for fast-moving categories; quarterly otherwise.
Scope: update only bestsellers and new landers; skip anything unchanged.
Archive: a folder per competitor with dated subfolders (YYYY‑MM) holding the pasted samples and your latest tables.
Cadence matters more than size. See Cadence beats volume for a sustainable rhythm.
What to automate (and what not to)
Automate
The tables: value props, feature→benefit→objection, angle-by-segment, claims & proof, content themes.
The comparison: which props appear across brands; who pairs claims with proof.
Summaries: tone notes and CTA patterns.
Human check
Pricing nuance, promotions, and legal fine print.
Images and layout impact on message strength.
Whether an idea fits your positioning.
If you want this packaged into a repeatable workflow with storage for sources, try the AI competitor research tool and save your watchlist for monthly refreshes.
From findings to dashboards
Turn the analysis into a one-page view you can revisit:
Top 5 shared value props, ranked by how many competitors use them
Your chosen 3–5 message pillars (distinct from competitors)
Angle-by-segment table (which angles you cover vs. plan to test)
Claim–proof coverage (where to add proof on your PDPs)
Keep this in your workspace, or mirror the highlights in Marketing dashboards so you can tie next month’s CTR or add‑to‑cart lifts back to the angles you shipped.
Quick prompts you can paste today
“Summarise tone-of-voice from these quotes in 5 bullets: sentence length, formality, slang, numbers, emotion. Include one example quote per bullet.”
“From these product page quotes, output: Feature | Benefit | Likely objection | Verbatim copy handling the objection or ‘No copy provided’.”
“Build a claims–proof table: Claim | Proof type | Verbatim proof line | URL. Leave blank if no proof.”
“Create an angle-by-segment table (max 5 segments) from these ads/posts: Segment | Angle | Supporting quote | URL.”
“Group these posts/ads into 5–7 content themes: Theme | Purpose | Example headline (verbatim) | CTA type.”
When you’re ready to turn angles into assets, generate variations in your own tone and deliver them to channels via Platform integrations. If you prefer a single flow from research to draft copy and video, Reachype can take your chosen pillars and angles from Research straight into creative production.
Guardrails and ethics
Don’t copy or imply endorsement. Avoid trademarked phrases in your copy.
Cite sources in your internal doc. Keep dates on all samples.
Treat AI outputs as hypotheses. Validate with small tests before rolling out.
If you only do five things
Define the decision and metric before you start.
Pick 5–8 competitors and collect fresh, dated samples.
Use structured prompts to produce tables, not prose.
Turn outputs into three ads, three PDP edits, three posts.
Refresh monthly with just new bestsellers and landers.