About · Chișinău, Moldova · working with US & UK
A product manager who learned to ship, then learned to measure
I'm Artiom Graciov, an independent AI search visibility consultant with fourteen years in product, web and technical SEO, working with B2B SaaS companies in the US and UK.
I spent most of my career on the delivery side rather than the marketing side: project manager, then product owner, on web products, SaaS platforms and digital transformation programmes. Requirements, scope, engineering trade-offs, launch, then the unglamorous years of continuous improvement afterwards. As product owner I took a SaaS platform from a concept document to a profitable business, which is where balancing business goals against technical reality stopped being a slide and became a weekly argument.
SEO came into it because the products needed it. I ran technical and on-page work as part of shipping, not as a separate service — structured data, crawlability, Core Web Vitals, GA4 and Search Console, information architecture that survives a redesign. That background is the reason my recommendations do not stop at a PDF: I can implement them in WordPress or Next.js, or hand your engineers a spec written by someone who has been on their side of the table.
In 2025 I moved my practice to AI search visibility for one reason: the measurement was obviously broken. Tools were reporting single positions for a surface that returns a different answer every time you ask, and nobody was publishing how they got their numbers. That is a product-manager problem before it is a marketing one, and it is what I now do full time.
14+ yrs
in product, web and technical SEO
1 SaaS
taken from concept to a profitable business as product owner
4–6
clients at a time, so nothing is subcontracted
0
guarantees of position in AI answers — nobody can honestly give one
Where the experience comes from
Pick a period
Fourteen years, mostly in rooms where something had to actually launch. I list it in periods rather than job titles because what matters to you is which of these skills shows up in your engagement, not which company paid for me to learn it.
What I did
Technical and on-page SEO as part of shipping products: structured data, crawl and render behaviour, information architecture, Core Web Vitals, GA4 and Search Console, WordPress and Next.js implementations.
Project manager on web builds and digital campaigns: scoping, estimates, running mixed teams of designers and developers, and shipping to dates that clients had already announced.
Product owner on a SaaS platform, taken from a concept document to a profitable business: roadmap, pricing, the arguments about which half of the feature to cut, and life after launch.
Full-time GEO and AEO work for B2B SaaS: frozen prompt sets, repeat runs across four engines, interval reporting, crawler access, and published research with the raw data attached.
What it taught me
That most SEO problems in modern stacks are rendering and access problems wearing a content costume.
That most project failures are specification failures, and that the person who writes the requirement badly is rarely the person who gets blamed for it.
That every recommendation has an implementation cost somebody has to pay, and advice that ignores that cost is not advice.
That the measurement layer is where this market is weakest, and that most of the value on offer is simply being honest about variance.
Why it matters to you
It is exactly the skill set the AI crawler layer demands, since no major AI crawler executes JavaScript and each one has its own permissions.
It is why an engagement with me starts with a written definition of what we will measure and what counts as a change, before any work begins.
When I hand your engineers a spec, it is scoped like something that has to ship — not like a list of best practices.
You get numbers with error bars and a written rule for what counts as a trend, instead of a dashboard that moves for reasons nobody can trace.
How I work
Four commitments, written down so you can hold me to them
These are not values-page decoration; each one has a cost to me, which is the only reason it is worth publishing. If I break one of these on your engagement, point at this page.
01
Intervals, never single numbers
Every figure I report carries a confidence interval, and if two intervals overlap the report says there is no trend. This costs me the ability to show you a flattering month, which is the point of publishing it.
02
The method is public
Prompt sampling, run counts, engine coverage and a reference prompt set are all published, so you can reproduce my work or hand it to someone else. Nothing about the protocol is proprietary.
03
I say no in writing
If your category has too little assistant-driven demand, or your site cannot be fixed without engineering capacity you do not have, you get that as a recommendation rather than an invoice.
04
No attribution theatre
There is no reliable way to trace revenue to an AI answer today, so I do not model one and call it attribution. Engagements are contracted on process and leading indicators, stated plainly up front.
Credentials, and what I don't have
The useful half of this section is the second half. Most consultant about-pages list only what flatters; since my whole argument is that you should judge people by what they can evidence, here is both columns.
What I can evidence
- ✓Fourteen years of delivery experience across web products, SaaS platforms and transformation programmes
- ✓A SaaS platform taken from concept to profitability as product owner
- ✓Hands-on implementation in WordPress and Next.js — I can ship the fix, not just describe it
- ✓A published measurement protocol with a reference prompt set anyone can rerun
- ✓Two research studies with methodology, sample sizes, dates and raw data attached
- ✓A degree in information technology from the Slavic University of Moldova
What I don't have — yet or ever
- ✕A wall of client logos — the practice is young and I will not borrow other people's
- ✕Case studies with revenue attribution, because that measurement does not reliably exist
- ✕Vendor certifications in AI visibility tools; none of them mean anything yet
- ✕An agency team behind me — beyond four to six clients I have to say no
- ✕A guarantee of inclusion in any AI answer, at any price
The practical details
Based in
Chișinău, Moldova · Eastern European Time, overlapping London and US East Coast
Working with
B2B SaaS, 20–200 people, selling into the US and UK
Languages
English for all client work; Russian and Romanian natively
Engagements
$1,500 audit, $750/mo monitoring, retainers from $3,000/mo, advisory at $250/hr
Stack I work in
WordPress, Next.js and React, structured data, GA4, Search Console, Cloudflare
Capacity
Four to six clients at a time, with a waiting list when it is full
Judge me on the method, not the bio.
Everything above is context. The thing worth checking is whether the measurement holds up — so read the protocol, or send me your domain and three competitors and see what comes back.