Skip to content
Colt Agar

The work,
with the evidence.

Results from SEO and AI search, and the decisions behind them.

Consulting engagements, in-house programs, a publication I own, and independent AEO research.

See selected signed agreements

In-house · Company name withheldHistorical results · May–June 2026

Visible in AI answers.

I built and ran the SEO and AEO program behind this site. These reports show its visibility across Google, Microsoft and the AI tools tracked by PromptWatch.

Part of the in-house program behind $150M+ in deal pipeline since October 2023, measured as projected annual revenue across opportunities sourced through organic SEO and AEO. See the full program

Microsoft

Bing Webmaster Tools

139.9K

Citations across Copilots and partners

May 18–June 18, 2026

Microsoft AI Performance report showing 139.9 thousand citations for May 18 through June 18, 2026.

Google

Google Search Console

1.77M

Impressions in generative AI features

May 18–June 18, 2026

Google Generative AI features report showing 1.77 million impressions, with daily chart labels May 18 through June 18, 2026.

PromptWatch

ChatGPT, Claude, Perplexity & Mistral

228,654

Page fetches for AI answers

May 18–June 18, 2026 · API dates, UTC

PromptWatch crawler report showing 227,535 ChatGPT, 461 Perplexity, 658 Claude and zero Mistral citation-agent fetches. Date picker reads May 17 to June 17; plotted daily series starts May 18.

Each report measures a different part of AI visibility. PromptWatch counts citation-agent page fetches in response to users’ questions. A fetched page is not necessarily cited in the final answer.

Building and running
a B2B search program.

In-house · Company name withheldHead of SEO / AEO

I owned the full SEO and AEO program, from strategy through implementation.

What I changed

I built the AI search program, rebuilt the website backend, and brought Core Web Vitals from failing to passing.

How I changed the operation

I replaced content and link agencies with documented processes and automation. That saved over $100K per month while improving results.

Program results · Since October 2023

in deal pipeline via organic search and AEO
$150M+
average monthly MQL growth
+300%
average monthly SQL growth
+500%

Pipeline is projected annual revenue across opportunities I sourced through organic SEO and AEO, not closed deals.

Lead growth compares average monthly marketing-qualified leads (MQLs) and sales-qualified leads (SQLs) after I joined in October 2023 with the monthly averages before I joined.

Selected recent agreements from my search program.

A few notable agreements I brought in through organic and AI search. These are a sample of the program’s results; other signed agreements have higher and lower values.

Figures below are projected annual revenue from signed agreements, not revenue already collected.

Signed agreement · June 4, 2026

$461,960

Projected annual revenue
from a signed agreement

A ChatGPT visit. A signed agreement.

The attribution notes record a first visit through ChatGPT; the prospect also credited Google.

Redacted agreement notification dated June 4, 2026, recording $461,960 in total annual revenue. This is projected annual revenue for a signed agreement. The attribution notes record a first visit through ChatGPT; the prospect also credited Google.
Original agreement notification · Identifying details redacted.

Signed agreement · May 20, 2026

$2,386,323

Projected annual revenue
from a signed agreement

Organic search led to a new agreement.

The attribution notes record an organic-search visit, followed by a return visit and form submission the next day.

Redacted agreement notification dated May 20, 2026, recording $2,386,323 in total annual revenue. This is projected annual revenue for a signed agreement. The attribution notes record an organic-search visit, followed by a return visit and form submission the next day.
Original agreement notification · Identifying details redacted.

Signed agreement · September 3, 2026

$210,745

Projected annual revenue
from a signed agreement

The prospect credited Claude.

The prospect named Claude when asked how they heard about the business, then submitted an inquiry through a branded search ad.

Redacted agreement notification dated September 3, 2026, recording $210,745 in total annual revenue. This is projected annual revenue for a signed agreement. The prospect named Claude when asked how they heard about the business, then submitted an inquiry through a branded search ad.
Original agreement notification · Identifying details redacted.
Discuss your search program

From 6 to 501 fetches a day
for ChatGPT answers.

Owned publicationPromptWatch · ChatGPT CitationsMay–August 2026 · 7-day daily averages

A publication I built and run.

I built the site, researched the keywords, wrote and published the content, and optimized it for AI search.

PromptWatch labels this activity “ChatGPT Citations”: pages fetched in response to a user’s question. These are answer-time fetches, separate from routine indexing and model-training crawls.

Publication name withheld.

Page fetches for ChatGPT answersPromptWatch · ChatGPT Citations · daily fetches
6 / day7-day avg · May 7–13501 / day7-day avg · Aug 13–19
Daily ChatGPT citation-agent fetches, May 7 to August 19, 2026Selected growth period ending at the August 19 daily peak of 678 fetches. Seven-day averages increase from 5.6 in May 7–13 to 501.3 in August 13–19. Historical results through August 19, 2026.0200400600May 7JuneJulyAug 19
Daily fetches7-day average

Historical results · May 7–August 19, 2026 UTC.
Rounded 7-day averages. Fetches do not necessarily result in a visible citation in the final answer.

What an AEO audit can uncover.

I trace AI answers back to their sources and test what the pages actually communicate. These findings show how pricing, comparisons and product details can go wrong.

Selected independent research on NVIDIA, ElevenLabs and ClickUp. These were not client engagements.

One product in the answer.
Another in the pricing source.

NVIDIAAnswer accuracyJuly 20, 2026

ChatGPT answered a question about one NVIDIA rack with a price sourced from a different model. Checking the cited article exposed the mismatch.

What I’d addressMake the product generation and pricing scope explicit in first-party buying information, then retest whether answers cite evidence for the correct product.

From the captured answer to its pricing source

“How much does an NVIDIA GB200 NVL72 rack cost and what are its specs?”

“Some reports have cited figures around $3 million per rack.”
Product in the question
GB200 NVL72 rack
Product in the cited price article
GB300 Barron’s coverage

ChatGPT via DataForSEO · Two same-day captures. A source mismatch, not a verified GB200 price or an estimate of how often this happens.

Recommended by AI.
Competitors in the citations.

ElevenLabsCompetitive sourcesJuly 29, 2026

Perplexity recommended ElevenLabs, but relied on other companies’ pages to describe the market.

The same answer recommended WellSaid for enterprise and compliance, and Inworld for real-time agents, citing those companies’ own pages.

What I’d addressPublish first-party comparisons that answer those buying questions directly. Repeat the tests to check both the cited sources and how each product is described.

“Which AI voice generator is best?”
“The best all-around AI voice generator for most people in 2026 is ElevenLabs”
Cited sources
15
ElevenLabs pages
0
Competitor roundups
4

Competitor-owned sources in this answer

  • Inworld
  • Visme
  • WellSaid
  • Jotform

Selected Perplexity Sonar Pro capture. Three same-day captures varied model and phrasing; all recommended ElevenLabs with zero owned citations. No traffic loss was measured.

Can AI tell what
the product includes?

ClickUpPurchasing informationSeptember 4, 2026

ClickUp’s comparison page used checkmarks and crosses to show which product offered a feature. In my text-extraction test, those answers disappeared.

This affected 53 of 56 product values in the comparison. The row shown here kept the feature name but lost both yes/no answers.

Its pages also gave conflicting free-plan storage limits. An AI answer could change depending on which page it used.

What I’d addressConfirm the current features and storage limits, make the comparison’s yes/no answers readable as text, and correct conflicting pages. Then retest the extraction and AI answers.

Which product includes
AI meeting summaries?One row from ClickUp’s comparison with Monday
Published comparison claims versus text extracted for AI
VersionClickUpMonday
On the published pageYes (checkmark)No (cross)
In text extracted for AIMissingMissing

The feature name remained. The yes/no answers were missing.

Simplified audited row, not verified product availability. Monday also documents AI meeting summaries, so the comparison claims need checking too.

How much storage does ClickUp’s free plan include?

Pricing, FAQ & Help pages
60MB
Marketing-team page
100MB

Two published storage limits for the same free plan, recorded in the September 4 audit.

Organic search results.

SEO campaignsContent optimization & site growth

Client engagement · SEO consulting

Growth after a flat stretch.

After months of fairly flat traffic, this client’s estimated organic traffic climbed in the months after I started working with them.

17,071estimated organic visits per month

Historical organic traffic chart showing a relatively flat trend followed by growth after early June, with a reported estimate of 17,071 visits per month.
Original traffic report
Tool-estimated traffic, not measured sessions.

Client engagement · Content optimization

More clicks from an existing article.

I audited and re-optimized a client’s affiliate article. Here’s the change in average daily clicks, comparing the two weeks before with the two weeks after.

6.57 → 25.43average article clicks per day

+287% average daily clicks

SEOTesting click chart: daily clicks rise after the August 2023 test-start marker.
Original SEOTesting chart

Separate in-house role · Company name withheld

A B2B site finding its audience.

In a different in-house role from the pipeline program above, I handled the full SEO strategy and execution for this B2B site. This Ahrefs report shows its organic growth across foundational industry terms.

40,148estimated organic visits per month

Historical Ahrefs report: organic traffic estimate reaches 40,148 per month, with 18,644 organic keywords and a traffic-value estimate of 22,429 dollars. Curves show growth from 2022 into 2023.
Original Ahrefs screenshot
Traffic and traffic value are tool estimates, not measured sessions or revenue.

Search traffic that led to sales.

Results from my affiliate publishing work.

Revenue referred to an electronics retailer.

My SEO work sent 5,258 clicks to this retailer through its affiliate program. The report records 1,762 products shipped.

$385,424retailer revenue from shipped orders

Historical affiliate dashboard: 5,258 clicks, 385,424 dollars and 33 cents in shipped orders, and 1,762 products shipped. Commission earnings and dates were obscured in the original image.
Original affiliate report
Revenue generated for the retailer, not my commission earnings.

Organic traffic that led to Amazon orders.

One of my sites sent 51,282 clicks to Amazon through organic search traffic. The affiliate report shows a 12.25% ordered-item conversion rate.

6,282items ordered through Amazon Associates

Amazon Associates report, January 1 to September 5, 2023: 51,282 clicks, 6,282 ordered items, 6,157 shipped items, 399 returned items and 12.25 percent conversion.
Original Amazon Associates report
Ordered items, not unique customers or net sales after returns.

Let’s talk about
your site.

Tell me what you want to improve.

Discuss your project