Competitive White Space

Before you write anything

Read these first. They are the source of truth and this skill does not invent around them.

1. brand-kit/brand.md: name, palette, fonts, CTAs
2. brand-kit/voice.md: house style and banned constructions
3. brand-kit/positioning.md: the frame, pillars, traps, stats, vocabulary
4. brand-kit/icp.md: who you sell to, and your competitor set
5. brand-kit/capabilities.md: read before making any product claim

If brand-kit/ is missing or still has its template placeholders, stop and tell the user to run brand-kit-setup. Do not proceed with invented positioning.

One exception, and only one. Sections 1 through 4 of this analysis describe the market, not you. If the kit is empty and the user wants to proceed anyway, you may run the market-level version: competitor profiles, comparison matrix, saturation map, and white space. Say clearly which sections you are skipping and why. Never fill Sections 5 through 8 from assumption. Those sections make claims about a company, and a claim without the kit behind it is a guess with a letterhead.

What this produces

A competitive analysis that goes past "here is who else is out there." It maps the crowded messaging territory, isolates the empty territory, and converts the empty territory into a positioning direction and usable campaign language.

The pipeline is the point, and each stage feeds the next:

Competitor research → saturation map → white space → UVP direction → stage hooks

Skip a stage and the output degrades into a vendor list. The hooks are only credible because they came from a gap, and the gap is only real because the saturation map proved everyone else is standing somewhere else.

When to use it

• Refining positioning against a known competitor set
• A list of competitor URLs or names arrives with a request for a gap analysis
• "Run the competitors" after a company recon
• Pre-work before building a campaign, a content plan, or a messaging refresh
• Sales enablement input for battlecards

When not to use it

• A single company from a single URL. Use company-recon.
• Paid channel performance. Use google-ads-audit or linkedin-ads-audit.
• Writing the assets themselves. This skill produces the angle; content-writer and social-generator produce the copy.

Where the competitor set comes from

In priority order:

1. URLs the user provided. Best input. Use them and skip discovery.
2. Names the user provided. Search, confirm you have the right company, then research. A one-line "confirming these are the right companies" is fine.
3. The Competitors table in brand-kit/icp.md. This is the default set when the user asks for a competitive analysis without naming anyone. Respect the tiers: tier one is who you lose deals to, tier two is who shows up in search and analyst coverage. They need different treatment, so do not blend them into one undifferentiated list.
4. Discovery by search, when that table is empty. Run the queries below, pick three to five, and then tell the user what you found and why you picked it before going deep. Something like: "Your Competitors table is empty, so I searched and found these five. I am analyzing the first three because they target the same buyer. Say the word and I will swap any of them." Do not invent a competitor set silently, and do not proceed with a set the user has not seen.

Discovery queries. Seed these with the Category terms and the Competitor and alternative terms from Search Terms in brand-kit/icp.md:

• [subject] competitors
• [subject] vs
• [subject] alternatives
• best [category term] platforms OR tools OR vendors
• [category term] vendor comparison OR shortlist

Selection criteria:

• Direct ICP overlap. Same buyer, same problem, per The Fit Matrix in icp.md.
• Market proximity. Same category, or an adjacent one listed under Categories We Are Not in positioning.md that buyers confuse you with.
• Competitive reality. Not aspirational comparisons against companies a hundred times your size, unless the analysis is specifically about incumbents.

Three to five competitors. More than five dilutes the pattern. Fewer than three means you cannot tell a saturated theme from a coincidence.

Workflow

Step 1: Establish subject context

If company-recon ran earlier in this conversation, use its findings as the subject context. Do not re-research what you already have.

Otherwise gather what you need directly: read the subject's homepage, positioning page, and pricing page, and read The Frame, The Pillars, and Naming and Usage Rules in brand-kit/positioning.md.

If there is no subject at all, that is a valid mode. Frame the whole analysis at market level and say so in the document's opening line.

Step 2: Research each competitor

Per competitor, run these in parallel, and run competitors in parallel with each other.

Fetch: homepage, about, product or services, pricing if public, case studies or customers.

Search:

• [competitor] G2 reviews OR [competitor] Capterra: outside perception, named strengths, named complaints
• [competitor] funding employees revenue: scale signals
• [competitor] competitors: how they position against others, which is often more honest than how they position themselves
• [competitor] [your named trap]: for each trap in Named Traps in positioning.md, check whether the competitor has anything to say about it. Silence is the finding. A trap nobody answers is white space.

Extract per competitor:

• Self-description in their own words, from the hero or H1
• Primary hook and CTA
• Claimed differentiators
• Audience signals: firmographics, verticals, size, named buyer roles
• What they actually sell
• Pricing structure if visible
• Proof assets: case studies, logos, review scores, awards
• Visible limitations and gaps
• Support and service model signals
• Category posture: how they talk about the terms in Category terms, and whether they take a stance on any of your named traps

Step 3: Apply the evaluation lens

Judge every competitor on the same dimensions, so the comparison holds.

Four dimensions are constant, regardless of company:

• Positioning versus execution. Do they deliver what they claim? Is the claim backed by proof, or is it a sentence?
• Audience fit. Where their ICP actually is, versus where they claim to play. Their case studies tell the truth their homepage does not.
• Proof strength. Named customers with numbers, or logos with adjectives.
• Entanglement. How hard are they to replace, and what creates the lock-in?

The remaining dimensions come from your own pillars. Read The Pillars in brand-kit/positioning.md. Turn each pillar into a question you can ask of a competitor's website, using that pillar's Vocabulary that signals it as the search vocabulary and its The failure it prevents as the test.

For example, a pillar whose failure mode is "work stalls when it crosses a system boundary" becomes the question: does this competitor address what happens at the handoff, or does their story stop at their own product edge?

Critical rule, from positioning.md itself: pillars are internal scaffolding. They can name your columns in the internal analysis. They never appear by name in customer-facing copy, and Section 8's hooks must not contain them.

If pillars are not yet defined, fall back to these four and say in the document that they are generic defaults:

• Scope. Point solution or platform? How much of the buyer's problem do they actually cover?
• Depth. Is the capability real and demonstrated, or a feature list?
• Trust. Governance, security, compliance, reliability, and what happens when the product is wrong.
• Time to value. How long from purchase to the first outcome, and who has to do the work.

Document sections

Section 1: Competitor profiles

One card per competitor, roughly a page each, laid out as a two-column table (Field, Detail). Page break between cards.

The twelve fields:

Formatting: field label column around 2400 DXA, detail column around 6960 DXA. Alternating row shading. One to three bullet lines per cell, prioritizing what changes a buyer's mind. Cite evidence inline and briefly: (site), (G2), (case study), (press). Unknown is "Unknown." Never invent a field.

Section 2: Comparison matrix

Every competitor side by side. The at-a-glance page.

Columns: one per competitor, plus a label column.

Rows:

1. Positioning, one line
2. Primary audience
3. Pricing model
4. Key differentiator
5. Proof strength: Strong / Moderate / Weak, with the basis in four words
6. Entanglement: High / Medium / Low, with what creates the lock-in
7. Pillar coverage: which of your pillars they can credibly claim, using the pillar names from The Pillars as the row's sub-labels, or the four fallback dimensions if pillars are undefined

Rubrics for rows 5 and 6:

One to two lines per cell. This table fits on one page. Use landscape or abbreviated text if it will not.

Section 3: Messaging saturation map

What everyone is already saying. This is the territory where a buyer cannot tell you apart, so competing in it means competing on budget.

Discover saturation empirically. Do not seed it from a list. The method:

1. Pull the exact hero line, subhead, and top three navigation labels from every competitor site you fetched. Add each one's primary hook and top claimed differentiator. That is your raw corpus.
2. Cluster the corpus by claim, not by wording. "Cut resolution time by half" and "faster resolution, less effort" are one theme wearing two outfits.
3. Count competitors per theme, not mentions per theme. A vendor saying the same thing on six pages is one competitor, not six.
4. Keep every theme claimed by two or more competitors. Discard the rest into Section 4's raw material.
5. Rate saturation: High is most of the set, Medium is roughly half, and anything under two competitors is not saturated, it is a data point.

Present four to seven themes:

Worked example, in a neutral category. Five field service scheduling vendors were fetched. Their hero lines clustered into four themes: "cut windshield time" (four of five, High), "AI-optimized routing" (four of five, High), "delight your customers" (three of five, Medium), and "one app for every technician" (two of five, Medium). One vendor claimed something nobody else did: guaranteed parts availability before dispatch. Two vendors is not a theme, one vendor is not saturation, so that last claim moved to Section 4 as candidate white space. The useful output of this map was not the four crowded themes. It was learning that every vendor competes on routing math and none of them competes on whether the technician arrives with the right part.

After the table, write two to three sentences on what the saturation means: these claims are table stakes now, and using one as primary positioning guarantees the buyer hears you as interchangeable.

Section 4: White space analysis

What nobody is saying. The most valuable section in the document. Spend the most analytical effort here.

Sources for candidate gaps, in order of reliability:

• Claims made by exactly one competitor, from step 4 of the saturation method
• Complaints that appear in review sites across multiple vendors, which nobody answers in their marketing
• Your Named Traps from positioning.md that no competitor addresses
• The distance between what a category's buyers ask about and what its vendors lead with

Three to five gaps. For each:

• The gap: what is missing from the conversation, in one sentence
• Why it matters: the real buyer need or fear going unaddressed
• Evidence: what signals this is real: review complaints, search behavior, a competitor's conspicuous silence, an analyst note
• Opportunity size: rated by the rubric below
• Pillar mapped: which of your pillars this gap sits under, if any. A gap under no pillar is still worth naming; it may be telling you a pillar is missing.

A gap you cannot back with evidence is a hypothesis. Label it as one or cut it.

Section 5: UVP direction

Requires subject context. Without it, title this "Positioning opportunity" and frame everything as market-level recommendation.

• UVP statement. One sentence, thirty-five words maximum, built from the white space rather than from existing copy. Check it against Naming and Usage Rules in positioning.md before writing it down: the always-say list, the never-say list, and the capitalization rules are binding. If an existing tagline in that section already says this, use the tagline verbatim rather than paraphrasing it. Taglines are used exactly or not at all.
• Messaging to dial back. Which of the subject's current claims land inside Section 3's saturated themes. Name the phrase and the theme it collides with.
• True differentiators. What is genuinely defensible, checked against What We Do in brand-kit/capabilities.md. A differentiator with no capability behind it is a promise the product will break. Check What We Do Not Do before writing any of them.

Land the direction on the pillars from The Pillars in positioning.md. Say which pillar the white space strengthens and which one it exposes as thin. Use pillar names in this internal section; strip them from anything customer-facing.

Section 6: Pain points uniquely solvable

Requires subject context. Without it, title this "Underserved pain points in market" and drop the ownership claim.

Draw the pains from The Fit Matrix and Buying Committee in brand-kit/icp.md, filtered to the ones the white space actually covers.

• Functional pains: what breaks in the buyer's week
• Strategic pains: what the economic buyer is exposed to, taken from the What kills the deal for them column of the Buying Committee table
• For each pain, tie it back to the white space and explain why competitors are not solving it. Cannot, will not, or have not noticed. Those are three very different competitive situations and only one of them is durable.

Section 7: Targeting priorities

Requires subject context. Without it, title this "Recommended targeting based on white space."

Build the priority tiers from The Fit Matrix in brand-kit/icp.md, then re-rank by white-space resonance. Fit tells you who can buy. White space tells you who will listen.

• Primary. ICP-column matches who feel the Section 4 gaps most acutely. Name the seniority, function, size band, and industry using the platform taxonomy values in the Fit Matrix so the list is directly usable in a campaign.
• Secondary. Mixed-column matches, or ICP matches with lower urgency.
• De-prioritize. Non-ICP, plus any segment where a competitor's proof is genuinely stronger than yours. Say which competitor and why.

Section 8: Messaging hooks by buying stage

Turn the UVP and the white space into language someone can put in an ad tomorrow.

Generate these. Do not recite stored messaging. Every hook must trace to a specific gap in Section 4 and must avoid every theme in Section 3. A hook that could appear on a competitor's homepage has failed the only test that matters.

• Unaware (problem education). Two to three hooks that name the problem in language the market is not using. Your Named Traps are the strongest raw material here, because a named trap is a problem statement the buyer can repeat back to you.
• Aware (solution differentiation). Two to three hooks that separate your approach from the saturated themes. Use the separating lines in Categories We Are Not when the confusion is with an adjacent category.
• Engaged (conversion). Two to three hooks tied to a specific offer, proof point, or removed risk. Use the exact CTA wording from Calls to Action in brand-kit/brand.md. Cite proof only from Headline Stats in positioning.md, and only rows marked verified.

Each hook is a crisp phrase for an ad, a headline, or a sales conversation. Not a paragraph of copy. Run every hook through brand-kit/voice.md before it goes in the document, and strip any pillar name that survived.

Section 9: Key takeaways

Five to eight executive bullets:

• Where the messaging is most crowded, and therefore least effective
• The one or two biggest white space opportunities
• The clearest UVP angle
• Which competitors are most vulnerable, and on which dimension
• The single most important strategic move this analysis implies

Quotable in a strategy meeting. Written in the voice brand-kit/voice.md describes.

Writing rules

• Tone: executive, neutral, confident. No hype, no filler.
• No em-dashes. Commas, periods, colons.
• No emojis.
• No links or full URLs in the body. Cite briefly: (site), (G2), (case study), (press), (LinkedIn).
• No code blocks.
• Paragraphs of two to three sentences, or bullets.
• Bold sparingly.
• Tables for anything comparative.
• Do not invent data. Unknown is "Unknown."
• Quote competitors sparingly. Short quotes only, paraphrase by default, and never characterize a competitor's product beyond what you can evidence.
• Every line clarifies focus or difference. Who to target, or why to choose the subject. A line that does neither is cut.
• No hedging. Note the basis for an inference once, then move on.
• Apply the banned words and AI tells from brand-kit/voice.md before delivery.

Output

Produce a Word document.

Use the bundled docx skill and let the host resolve it. Follow that skill's own implementation rules rather than restating them here: explicit page size, real bullet numbering rather than unicode characters, table widths declared on both the table and its cells, cell margins for readability, heading styles overridden by their built-in IDs, page breaks between competitor cards, and separate paragraphs instead of newline characters inside a run. Validate the file with whatever validation step that skill provides before presenting it.

If no docx skill is available: output the same nine sections as clean markdown in the conversation, keep every table, and tell the user plainly that no document skill was found so they got markdown instead. Do not silently downgrade. If brand-visuals is installed, offer to turn the markdown into a branded PDF instead. Nothing in the analysis changes; only the format does.

Styling comes from the brand kit, not from this file. Read the Palette table in brand-kit/brand.md and map it:

Use the fonts from Typography in the same file, falling back to a common system font if they are not guaranteed on a recipient's machine, and note the substitution once. The footer string is the Document Footer value from brand.md, right-aligned in the muted text color.

Name the file after the subject and the analysis. Where it goes depends on the host, so do not assume a path. If the session has a designated output or deliverables folder, write it there. If it does not, which is the normal case in a plain command line session on a laptop, write it to the current working directory. Either way, say in one line where you put it. A file written to a folder that does not exist on this machine is a silent failure, and the user finds out by not finding it.

Speed and quality notes

• Do not ask clarifying questions when URLs are provided. Go.
• If only names are given, confirm you have the right companies before deep research. One line, then proceed.
• Fetch and search in parallel, across competitors and within each one.
• Sections 3 and 4 are the product. Everything else is supporting evidence. Budget your effort accordingly.
• After the document exists, give a two to three sentence summary in chat: the top white space finding and what it implies. Do not reproduce the document.

What comes next

Optional handoffs. Each runs on its own if the user goes straight to it.

• google-ads-audit: test whether paid search terms are buying saturated territory. The Section 3 themes are the ones to check for waste.
• linkedin-ads-audit: test whether paid social is reaching the Section 7 primary targets. This analysis sharpens the ICP Fit Score that audit computes.
• content-writer: turn a Section 4 gap into a long-form asset.
• social-generator: turn Section 8 hooks into posts.

Each of those skills will use this analysis if it is in the conversation, and gather what it needs directly if it is not.