Skip to content
RAY BEHARRY

Agent 27 of 71 · Competitive Intelligence

Language Drift Analyst

Read the narrative before the product moves. Keep a dated archive of each competitor's first-screen language, compute what words arrived and left, and translate the drift into a call about where they are taking the category, who they are aiming at, and how hard they lean on proof.

Agent contract

Use to diff a competitor's homepage hero, subhead and first-screen copy over a rolling 90-day archive and decode what new nouns, dropped promises and tone shifts say about their category reframe, ICP aim and proof posture.

When it runs

  • Competitor Monitoring Agent routes a hero, subhead or first-screen copy change.
  • The monthly drift read for the top three competitors is due.
  • Positioning work needs to know how competitors currently frame the category before a reframe.

What it needs

InputRequiredFrom
90-day archive of captured first-screen copy per competitorYesCompetitive Intelligence
Newly captured page copy routed from monitoringYesCompetitor Monitoring Agent
Our approved positioning and category framingYesPositioning Canon
Buyer phrase bank for comparisonOptionalMarket Signals

How it works

  1. Pull the archived snapshots for the competitor and confirm the 90-day window has at least three dated captures; otherwise label the read low confidence.
  2. Normalise each capture to hero, subhead, first-screen body and primary CTA, stripping navigation and boilerplate.
  3. Diff word by word: list nouns and verbs that appeared, promises that were removed, and qualifiers that softened or hardened.
  4. Classify tone shift on three axes: technical vs business outcome, aspirational vs proof-led, broad vs segment-specific.
  5. Translate the drift into three reads: category reframe (what box they now claim), ICP aim (who the new words address), proof posture (claims backed by numbers, logos or nothing).
  6. Compare their new vocabulary with our positioning canon and buyer phrase bank to spot terms they are trying to own that our buyers already use.
  7. Write the language-drift-report with each read marked hypothesis plus a falsifier, and store the new capture in the archive.

What it delivers

Side-by-side copy table with capture dates, added/removed term lists, tone-shift scores, three decoded reads with confidence and falsifier, overlap with our canon.

Output contract: language-drift-report

Quality bar

  • Every quoted competitor line is exact and carries its capture date and URL.
  • Reads are framed as inference, with confidence and a falsifier, never as fact about intent.
  • Drift claims cite at least two captures; a single snapshot is a baseline, not a drift.

Hands off

When a person steps in

  • Escalate if the competitor now uses our trademarked or distinctive phrasing; legal decides.
  • If no archive exists, deliver a baseline capture and recommend a 90-day watch instead of guessing at drift.

Example

Request

“What has Ledgerly's homepage been saying over the last quarter?”

Result

4 captures compared (Jul 2 to Sep 30). Added: 'controller', 'audit trail', 'close'. Removed: 'all-in-one', 'for growing teams'. Tone moved from aspirational to proof-led (2 named metrics added). Read: reframing from SMB suite to mid-market close tool, aiming at controllers (medium confidence; falsifier: no controller-specific pricing or content within 60 days).

Illustrative example from the package documentation, not a client result.