Northbound

Customer-interview repository for product teams

https://northbound.example

/ 100

Needs work

Executive read

Northbound is a plausible product with a thin citation surface. Models can describe what it is in generic terms, but they do not treat the site as a source. Competitors with named methods, original numbers, and comparison pages win the citation. The ten fixes below are ordered by expected lift for ChatGPT, Perplexity, and Google AI Overviews — not by how impressive they look on a marketing site.

Engine snapshot

Engine	Status	Note
ChatGPT	Mentioned	Appears as a category example in some answers, almost never as the cited source. Positioning is restated in generic SaaS language.
Perplexity	Competitor cited	Dovetail and Grain take the footnote. Northbound’s homepage is marketing copy; those sites offer definitions, tables, and dated posts.
Google AI Overviews	Absent	No Overview inclusion for head terms such as “customer interview analysis tool.” Thin FAQ and missing SoftwareApplication markup.

Appears as a category example in some answers, almost never as the cited source. Positioning is restated in generic SaaS language.
Dovetail and Grain take the footnote. Northbound’s homepage is marketing copy; those sites offer definitions, tables, and dated posts.
No Overview inclusion for head terms such as “customer interview analysis tool.” Thin FAQ and missing SoftwareApplication markup.

CiteScore does not query ChatGPT, Perplexity, or Google AI Overviews live. Engine snapshots estimate citation likelihood from on-page GEO signals (definitions, entities, evidence, schema, freshness, crawler access) plus, when configured, a language-model review of the public HTML we fetched. Treat them as a diagnostic, not a screenshot of an AI product’s UI.

Dimension breakdown

Answer fitness
Pages sell; they rarely answer the question a buyer would type into ChatGPT. Weight High.
41
Entity & brand clarity
Name and category are clear. Category membership vs. adjacent tools is fuzzy. Weight High.
58
Citable evidence
No original statistic, named protocol, or dated finding a model can quote. Weight High.
33
Structured markup
Almost no JSON-LD. Title tags are brand-first, not question-first. Weight Medium.
19
Source-worthiness
Clean product, thin third-party corroboration. Authors are unnamed. Weight Medium.
52
Freshness
Blog last updated 11 months ago. No visible changelog or “updated” stamps. Weight Medium.
48
AI-crawler access
No llms.txt. robots.txt is silent on GPTBot / PerplexityBot / Google-Extended. Weight Medium.
36
Mention share
Competitors occupy the comparison set. Northbound is an afterthought, if present. Weight Low.
44

Ten fixes that would get this site cited

Ranked by expected citation lift, not by how good they look on a launch tweet.

1.Lead the homepage with a 90-word definitional answer
High impact · Half day
northbound.example/
Why it matters. ChatGPT and Overviews extract the first self-contained answer. The current hero (“Interview intelligence for modern product teams”) is a slogan, not a citation.
Do this. Open with: what Northbound is, who it is for, what it replaces, and one concrete outcome. Write it as a paragraph a model can lift verbatim. Keep the slogan below the fold.
2.Publish a named, dated methodology page
High impact · 1–2 days
northbound.example/method/interview-protocol
Why it matters. Perplexity prefers pages that look like sources — a named thing with a date, an author, and steps. Feature pages do not qualify.
Do this. Ship “The Northbound Interview Protocol” (2026). Include steps, a short glossary, and a last-reviewed date. Link it from the homepage and docs.
3.Add Organization, SoftwareApplication, and FAQPage JSON-LD
High impact · Half day
Homepage, /pricing, /faq
Why it matters. Google AI Overviews lean on explicit entities. Northbound currently looks like an untyped marketing site.
Do this. Mark the product name, category, URL, pricing model, and 6–8 real questions customers ask. Mirror the visible FAQ; do not invent markup that is not on the page.
4.Create two factual comparison pages
High impact · This week
/compare/dovetail and /compare/grain
Why it matters. “X vs Y” is how both buyers and models assemble a shortlist. Those queries currently resolve to the competitors’ own pages.
Do this. Tables, not adjectives: export formats, seat model, interview sources, on-device vs cloud, SOC 2. State where Northbound is the worse fit.
5.Release one original, quotable number
High impact · This week
/research/interview-to-ticket-2026
Why it matters. Models cite statistics. Northbound has none of its own, so answers borrow Dovetail posts and generic “70% of insights go unused” claims.
Do this. Survey 80+ PMs or analyze anonymized in-product events. Publish n, date, and method. Put the headline number in the first 100 words.
6.Add llms.txt and a plain-language /ai page
Medium impact · Half day
/llms.txt and /ai
Why it matters. Crawlers that honor machine-readable summaries currently get a JavaScript marketing shell.
Do this. llms.txt should list canonical URLs for product, method, compare pages, and research. /ai restates positioning without animation or cookie banners.
7.Put named authors and credentials on every article
Medium impact · 1–2 days
Blog and method pages
Why it matters. Unattributed posts are treated as marketing. Bylines with a role and a prior company are treated as expertise.
Do this. Add person pages (not a team grid). One paragraph of relevant experience, a portrait, and the same name in JSON-LD.
8.Rewrite feature pages as “when to use / how it works”
Medium impact · This week
/product/highlights, /product/themes, /product/clips
Why it matters. Adjective stacks (“beautiful, collaborative, AI-native”) give models nothing to cite and nothing to rank against a competitor.
Do this. Each feature page: the job to be done, the workflow in 5 steps, limits, and a short example transcript or screenshot with alt text that states the fact.
9.Earn three third-party mentions this quarter
Medium impact · This week
Off-site
Why it matters. Citation systems triangulate. A product that only exists on its own domain is easy to skip.
Do this. One founder interview, one changelog/directory listing, one honest G2 or AlternativeTo profile. Link back to the method and research pages, not the homepage hero.
10.Show that the product is current
Medium impact · Half day
/changelog and article bylines
Why it matters. Stale lastmod dates push models toward last year’s roundup posts.
Do this. Public changelog with month-level entries. “Updated 12 Sep 2026” on method and compare pages. Keep sitemap lastmod honest.