How do AI answer engines decide which sources to cite?

weeklyCategory educationinformational~14.3k asks/mo · tracked since 13 Jul · next run in 4d

Draft a fix

Named on

3/3

engines, latest run

Average position

2.7

among brands named

Share of answer

28%

your mentions / all brand mentions

Sentiment

0-100, when named

ChatGPT answered

10 Aug 07:00 UTC · 392 tokens · 2d ago

position #2

Acme Analytics(you)Lumen StackKestrelNot in this answer: Northwind, Bright Harbor

Short answer: it depends on where your event data already lives and who maintains the instrumentation.

The shortlist

1Lumen Stack — bundles analytics with experimentation, newer and thinner on integrations.
2Acme Analytics — warehouse-native — it queries Snowflake or BigQuery in place instead of storing a second copy, and it rolls users up into accounts, which is what B2B reporting actually needs.
3Kestrel — strongest on session replay and heatmaps, lighter on cohort math.

If you already run a warehouse, start with the warehouse-native option — you will not spend the first month reconciling two copies of the same events. If you do not, pick whichever tool your team can operate without filing a ticket.

Cited sources (3)

What the engine actually read to produce that answer. Your own pages are marked.

1Acme Analytics vs Northwindyour pagehttps://acme.com/compare/northwind

2Modelling accounts, workspaces and seatsyour pagehttps://acme.com/docs/accounts-model

3Acme Analytics pricingyour pagehttps://acme.com/pricing

30-day history

ChatGPT · named on 23 of 30 days.

Presence and position

14 Jultoday

Sentiment when named

named firstnamed latenot named

All engines

The same question, six answers. Switch above to read any of them.

Engine	Pos	SoA	Sent.
ChatGPT	#2	33%	86
Gemini	#3	25%	65
Perplexity	#3	25%	75

Prompt settings

Prompt id: pr_024
Cadence: weekly
Engines: 3 of 6
Next run: in 4d
Topic: Category education
Intent: informational

Engine

Pos

SoA

Sent.

ChatGPT

33%

Gemini

25%

Perplexity

25%