Which cybersecurity provider does AI recommend in Canada?

What do AI answers say about Canadian cybersecurity providers?

1. No provider owns Canadian cybersecurity in AI answers. CrowdStrike leads on 11.7 percent of provider mentions, the fifteen most-named take 64.4 percent between them, and 161 other companies split the remaining 35.6.Figure 1, share of provider mentions
2. ChatGPT and Perplexity rank the same four companies in almost opposite order. CrowdStrike is first in ChatGPT on 12.3 percent and fourth in Perplexity on 6.9. eSentire is fifth in ChatGPT on 5.3 and first in Perplexity on 9.5.Figure 2, ChatGPT versus Perplexity
3. Google AI Overview names a cybersecurity vendor a quarter as often as ChatGPT. It gave 255 provider mentions across 135 answers, covering 37 companies. ChatGPT gave 1,102, covering 111.Figure 3, mentions by engine
4. Vendor documentation is doing the recommending. Eight of the fourteen most-cited domains are provider sites, and the pages being cited are topic guides, comparisons and ranked lists rather than campaign pages.Figure 4, most-cited domains

Which cybersecurity providers does AI name most in Canada?

CrowdStrike, on 11.7 percent of all provider mentions. Then Arctic Wolf on 8.7, Sophos on 8.6 and eSentire on 7.1. Nobody else clears 4.3.

After the top four the drop is steep. The tenth provider takes 2.5 percent, the fifteenth takes 1.4, and the 161 companies below them split 35.6 percent without one of them reaching 1.4. A buyer who asks twice, or asks two engines, gets a different field each time.

Four of the fifteen are Canadian: eSentire, CDW Canada, CIRA MDR and EY Canada. They sit among the global platforms rather than in a bracket of their own, because the answers offer all of them as alternatives to one another.

Figure 1 · All 45 questions, 3 engines

No provider takes more than one mention in eight. A long tail takes one in three.

Canadian providerProvider based elsewhereEveryone below the top fifteen

1. 01CrowdStrike11.7%
2. 02Arctic Wolf8.7%
3. 03Sophos8.6%
4. 04eSentire7.1%
5. 05Rapid74.2%
6. 06SentinelOne3.8%
7. 07Expel3.5%
8. 08Huntress3.0%
9. 09Red Canary2.6%
10. 10Palo Alto Networks2.5%
11. 11CDW Canada2.1%
12. 12CIRA MDR1.9%
13. 13EY Canada1.9%
14. 14ESET1.4%
15. 15Secureworks1.4%
16. 161 other providers35.6%

0%10%20%30%40%

Data behind Figure 116 rows · CSV · JSON · PNG

Name	Value	Answers naming it	Average position	Sentiment
CrowdStrike	11.7%	24.2%	2.9	68 / 100
Arctic Wolf	8.7%	18%	2.7	69 / 100
Sophos	8.6%	17.8%	2.7	71 / 100
eSentire	7.1%	14.8%	2.3	76 / 100
Rapid7	4.2%	8.6%	4.1	69 / 100
SentinelOne	3.8%	7.9%	3.5	61 / 100
Expel	3.5%	7.2%	3.0	72 / 100
Huntress	3.0%	6.2%	2.8	70 / 100
Red Canary	2.6%	5.4%	4.2	74 / 100
Palo Alto Networks	2.5%	5.2%	3.5	56 / 100
CDW Canada	2.1%	4.4%	3.3	59 / 100
CIRA MDR	1.9%	4%	3.6	74 / 100
EY Canada	1.9%	4%	7.0	70 / 100
ESET	1.4%	3%	2.7	81 / 100
Secureworks	1.4%	3%	5.5	61 / 100
161 other providers	35.6%	under 3% each

Consumer categories do not look like this. Two brands routinely take six of every ten mentions there. Here, 176 companies compete for the same shortlist slot.

Does ChatGPT recommend the same cybersecurity providers as Perplexity?

It names the same four companies and puts them in almost the opposite order. CrowdStrike is first in ChatGPT on 12.3 percent and fourth in Perplexity on 6.9.

eSentire runs the other way, 5.3 in ChatGPT against 9.5 in Perplexity. Expel and SentinelOne are the two clearest ChatGPT specialists, at 5.5 and 5.0 against 1.6 and 2.6. EY Canada goes the opposite way, 0.8 to 4.2, which is the largest proportional swing in the set.

Figure 2 · Same 45 questions, per engine

The two engines that name providers disagree about who comes first.

ChatGPT (135 answers)Perplexity (135 answers)

1. CrowdStrikeChatGPT (135 answers)12.3%Perplexity (135 answers)6.9%
2. eSentireChatGPT (135 answers)5.3%Perplexity (135 answers)9.5%
3. Arctic WolfChatGPT (135 answers)8.8%Perplexity (135 answers)6.9%
4. SophosChatGPT (135 answers)8.6%Perplexity (135 answers)7.5%
5. ExpelChatGPT (135 answers)5.5%Perplexity (135 answers)1.6%
6. SentinelOneChatGPT (135 answers)5.0%Perplexity (135 answers)2.6%
7. Rapid7ChatGPT (135 answers)4.0%Perplexity (135 answers)2.9%
8. EY CanadaChatGPT (135 answers)0.8%Perplexity (135 answers)4.2%

0%2%4%6%8%10%12%14%

Data behind Figure 28 rows · CSV · JSON · PNG

Name	ChatGPT (135 answers)	Perplexity (135 answers)
CrowdStrike	12.3%	6.9%
eSentire	5.3%	9.5%
Arctic Wolf	8.8%	6.9%
Sophos	8.6%	7.5%
Expel	5.5%	1.6%
SentinelOne	5.0%	2.6%
Rapid7	4.0%	2.9%
EY Canada	0.8%	4.2%

They read different material to get there. ChatGPT leans on vendor documentation and comparison pages. Perplexity cites more resellers, consultancies and Canadian sites, which is where CDW Canada and EY Canada pick up their positions. Work that lifts you in one engine will not necessarily move the other.

Which AI engines name cybersecurity vendors at all?

ChatGPT most, 1,102 times across 135 answers. Perplexity 634. Google AI Overview 255.

Google AI Overview also covered far fewer companies, 37 against ChatGPT’s 111. It is the least predictable of the three: it named almost nobody in the first two runs and then named providers freely in the third, which is where most of its 255 mentions come from.

Figure 3 · 135 answers per engine

Google AI Overview names roughly one provider for every four ChatGPT names.

1. 01ChatGPT1102
2. 02Perplexity634
3. 03Google AI Overview255

04008001200

Data behind Figure 33 rows · CSV · JSON · PNG

Name	Value	Providers named	Answers
ChatGPT	1102	111	90
Perplexity	634	96	90
Google AI Overview	255	37	90

The gap decides where effort goes. ChatGPT is the engine naming the most companies and the one where an absence costs the most: 110 other providers were named in its answers. Google AI Overview is worth watching rather than optimising for, because a metric that quadruples between runs is not yet a target.

Which sources do AI answers cite about Canadian cybersecurity?

Mostly the providers themselves. Eight of the fourteen most-cited domains are provider sites, and the top two are Palo Alto Networks and Expel.

The 405 answers carried 4,742 citations to 876 domains. microsoft.com is cited from 126 separate pages, more than twice any other domain, mostly Microsoft Learn documentation. mdrproviders.io, a directory nobody in the category owns, is cited in 11.9 percent of answers.

Page counts vary far more than usage. kaseya.com reaches 11.3 percent of answers from 9 cited pages and expertinsights.com 11.9 percent from 13, while microsoft.com needs 126 to reach 16.8.

Figure 4 · 405 answers, 3 engines

Eight of the fourteen most-cited domains belong to providers themselves.

A provider’s own siteCanadian resellerAnalyst, review, directory or platform

1. 01paloaltonetworks.comPROVIDER20.6%
2. 02expel.comPROVIDER18.1%
3. 03microsoft.comPLATFORM16.8%
4. 04rapid7.comPROVIDER16.8%
5. 05sophos.comPROVIDER16.1%
6. 06esentire.comPROVIDER14.2%
7. 07gartner.comANALYST13.5%
8. 08huntress.comPROVIDER13.2%
9. 09underdefense.comPROVIDER12.9%
10. 10cdw.caRESELLER12.6%
11. 11expertinsights.comEDITORIAL11.9%
12. 12mdrproviders.ioDIRECTORY11.9%
13. 13g2.comREVIEW11.6%
14. 14kaseya.comPROVIDER11.3%

0%5%10%15%20%25%

Data behind Figure 414 rows · CSV · JSON · PNG

Domain	Value	Citations	Pages cited
paloaltonetworks.com	20.6%	104	50
expel.com	18.1%	109	49
microsoft.com	16.8%	173	126
rapid7.com	16.8%	100	49
sophos.com	16.1%	79	38
esentire.com	14.2%	81	47
gartner.com	13.5%	75	37
huntress.com	13.2%	66	23
underdefense.com	12.9%	58	30
cdw.ca	12.6%	49	19
expertinsights.com	11.9%	49	13
mdrproviders.io	11.9%	68	40
g2.com	11.6%	62	30
kaseya.com	11.3%	37	9

What should a Canadian cybersecurity provider do about this?

Stop reporting one visibility score. CrowdStrike and eSentire both sit in the top four overall and get there from opposite directions: CrowdStrike runs 12.3 in ChatGPT and 6.9 in Perplexity, eSentire 5.3 and 9.5. One number reads as a close race. Split by engine, they are two different problems with two different fixes.

The pages getting cited are documentation. Provider sites hold eight of the fourteen most-cited positions, and what the engines quote from them is topic guides, comparisons and ranked lists rather than campaign pages.

Nine pages can beat a hundred and twenty-six. kaseya.com is cited in 11.3 percent of answers from 9 pages. microsoft.com reaches 16.8 percent from 126. Coverage of the buying question decides this, not the size of the site.

ChatGPT is where an absence costs you. ChatGPT named 111 providers across its 135 answers, so being missing there means a hundred other companies were named instead. Google AI Overview named 37, and quadrupled its output between runs, so it is worth watching before it is worth spending on.

How was this cybersecurity AI visibility study measured?

Panel

50 tracked questions written to match what a Canadian security buyer types on the way to a shortlist. The 45 that name no company are the panel for this report; the other 5 are excluded.

What the panel covers

Eight topics, weighted towards detection and response: managed detection and response 24 questions, vendor evaluation 9, implementation and onboarding 7, incident response 7, threat hunting 6, security operations 5, compliance reporting 4, security assessment 4. Questions can carry more than one topic. The panel does not cover identity, cloud posture, email security or penetration testing pricing, so it describes the detection and response shortlist rather than every security budget line.

Unbranded filter

Every figure on this page is drawn from questions that name no provider. A question that names a company will name it back, so mixing branded and unbranded questions inflates whoever the questions mention. Applying the filter moves the numbers: esentire.com falls from the most-cited domain in the full panel to sixth here.

Collection

Three runs. Two on 4 September 2026 five minutes apart, and one on 5 September 2026. Each question asked once per engine per run from a Canadian location. 405 unbranded answers in total, 135 per engine.

Share of provider mentions

A provider’s mentions divided by all provider mentions in the set being reported. A provider named twice in one answer counts once. Sums to 100 percent within a set.

Answers naming a provider

Answers naming the provider at least once, divided by all 405 answers. Two providers can share a share of mentions and differ here.

Average position and sentiment

Position is where the provider appears in the list an answer gives, averaged across the answers naming it. Sentiment is scored from the language around each mention on a 0 to 100 scale, where 50 is neutral.

Answers citing a domain

Answers containing at least one citation to a page on the domain, divided by all 405 answers.

Entity extraction

Providers were identified by automated extraction from each answer. Resellers, consultancies and accounting firms appear alongside product vendors because the answers offer them as alternatives to one another.

Provider marks

A provider shows its favicon only where we confirmed the site belongs to it by opening the site. A cited page is never treated as a provider’s website.

Limitations

Three runs across two days, and two of those runs were five minutes apart, so this is closer to two collection moments than three. Answer engines vary run to run and this panel shows how much: Google AI Overview named almost nobody in the first two runs and named providers freely in the third, which is where most of its 255 mentions come from. Read the order of providers below 3 percent as unstable. The panel is 45 questions weighted towards detection and response, so it describes that shortlist rather than the whole security market. The unbranded classification is automated; we checked all 50 questions by hand against the flag and found no disagreement, but a larger panel might.

More questions about AI and Canadian cybersecurity

Written by

Anton Sopov

Founder, The Prompt Group

Anton Sopov is the founder of The Prompt Group, an AI brand strategy and agentic discovery firm. He works with brands on how they are ranked, cited and described in ChatGPT, Claude, Gemini and Perplexity answers, and is the author of The Delegation Curve.

LinkedInXAll reports