Fractional MarTech Engineer.
Fractional MarTech Engineer.When your website drives the revenue,
the martech layer can't break.
For the last 12 years, I owned the part of neilpatel.com that could not go down:
lead capture, routing, tracking, GDPR/CCPA compliance, and performance.
Marketing shipped weekly changes to over 5 million monthly visitors.
I made sure they never landed in a broken funnel.
The hidden cost
Your marketing site is breaking silently.
Every tag and "quick integration" is a hidden failure point. When a form breaks at 2 AM, it doesn't throw an error - it just quietly stops sending you leads.
The true cost of a fragile martech stack:
- Delayed discovery: You find out from the sales team three weeks later.
- Corrupted decisions: Budget meetings run on wrong attribution data.
- Scaled losses: At 5M visits a month, small mistakes leak real money.
That gap, between marketing's need for speed and engineering's need for stability, is exactly where I work. For 12 years, it's been the only thing I've done.
What I own
What stops being your problem when I own it.
You focus on growth and strategy. I make sure the data and leads actually get there.
-
Leads that never silently drop.
A dropped lead is burned ad spend.
Every submission validated, enriched, and routed to the right place in your CRM, automatically:
- real-time validation, so junk data never reaches your sales team
- routing and booking flows that put each lead with the right owner
- built to fit your funnel, however complex it gets
If a lead comes in, it lands where it should. Zero manual handling. Zero lost leads.
-
Attribution you can defend in a budget meeting.
If you can't trust the numbers, every budget call is a guess.
GTM and GA4, implemented and QA'd properly, not just dropped in:
- full click-to-CRM path, respecting privacy rules
- events firing correctly, not double-counted or silently dropping
- tracking that survives your next site change instead of breaking with it
Clean source data. No setup tracks 100% in a post-cookie world. I capture everything that legally and technically can be, and I tell you where the blind spots are.
-
A site that stays fast as the stack grows.
A sluggish site kills conversions before your tracking even fires.
Performance is mostly universal rules, whatever CMS you run. I profile the site, find what's slowing it down, and act on it:
- the martech-side drag is mine to fix: tags, 3rd-party scripts, bloated consent logic
- structural issues go to your dev team as exact, prioritized fixes, not vague tickets
- the tag stack keeps growing without pushing your load times into the red
Marketing gets their 47+ tracking tags, and the Core Web Vitals stay green.
-
Compliance that does not cost you data.
A blanket consent banner cannibalizes your analytics.
I build consent that adapts to where each visitor actually is, instead of forcing the strictest rules on everyone:
- opt-in where the law demands it, opt-out where it allows it
- switched automatically by geolocation, region by region
- ready to extend as your traffic grows into new markets
GDPR and CCPA are where I've built this. The same logic works for whatever footprint you have.
-
Build or buy, decided on the merits.
Martech bloat drains budgets and slows down campaigns.
For every piece of your marketing stack, I weigh reliability against maintenance cost, so you're never:
- maintaining custom code that should have been a simple plugin
- fighting a bloated plugin that should have been fifty lines of code
- paying for tools you barely use
Custom code where it earns its keep, proven tools everywhere else. Always your call, with a clear recommendation from me.
Production systems
Built once, ran in production for years.
A few examples of what my team and I built and ran in production. I led the small technical team inside marketing, 2 people I managed directly, and I owned the architecture and the hard calls. We built and ran the rest together.
-
One website. 28 countries. Four CRMs.
One contact form, routed to the right regional HubSpot account automatically, before the page even finished loading.
How it workedThe company runs a single lead-generating website, but sales is split across four separate HubSpot accounts - US, Brazil, India and International - covering 28 countries. Some regional offices run almost as their own company under one brand. One form, four destinations, and the visitor must never know there was a decision to make.
So we built one endpoint that every form on the site submits to.
Validation and geolocation happen in the browser, where the form lives: it reads the visitor's location (served by Cloudflare) and checks the lead before anything is sent.
The endpoint is the single place all that clean data lands, and from there it routes each submission to the correct regional HubSpot account and Kit, and logs everything for debugging. No individual page carries the CRM integration code, it just hands off to the endpoint on the same server.
One place to maintain, one place to debug.
It became a platform, not a feature.
The endpoint takes data from any form that knows its address, secret key and field names, not just the Contact Form 7 setup on the marketing site.
Once we documented it, product teams with no connection to marketing could pipe their own leads into the right CRM through it.
Built once, reused across the company for years.
Geolocation, decided on the merits.
There are 4 ways to get a visitor's location, and 3 are wrong for a contact form:
- browser permission prompt is fine for "find my nearest store," but creepy when someone is contacting a company
- self-hosted IP database means constant updates, shaky accuracy and server load
- server-side or JS lookup means passing the IP around and waiting on a response
- edge network services (like Cloudflare)
Cloudflare hands us geolocation data right at the edge. It's the fastest and most accurate approach, resolving before the form even thinks about submitting.
That's why we went with it.
Validation that protects four things at once.
Email, phone and website are checked in the browser before anything reaches the endpoint: phone via an open-source library, email through a deliverability API, website with a live existence check plus a blacklist of known bot domains.
A typo gets caught and shown to the visitor on the spot.
The payoff is not just clean data, it is a stable server, a CRM not paying to store junk, and a sales team that never dials a number that doesn't exist.
Attribution that survives the long way round.
Every visit captures its UTMs, referral domain and landing page, into the form if there's one on the page, otherwise into a consent-gated cookie that works across subdomains.
Come back days later with no tags in the URL and the form still fills its hidden fields from that cookie.
The lead reaches the CRM carrying where it actually came from, not "direct / none."
Then real data complicated it.
Digging through the CRM, we found a pattern: a lot of visitors browse on a VPN. Plenty of India-based prospects showed up with their location set to the US, so the geo read "US" while the phone number was clearly +91.
The tempting fix is to override the routing on the phone number.
But a mismatch settles nothing: you can live in India and work for a US company, or own it, or just be visiting family and leave a local number. So instead of letting a script guess, these contacts were flagged in the CRM reporting, where a human could weigh the conflict.
Routing you can't be sure of is worse than routing a person can review.
The right calendar, automatically.
On a valid submission, the form already has the location and the selected budget range, so it serves the matching booking calendar right there, no round-trip needed.
Different countries and budget tiers map to different Chili Piper, Calendly and Outreach accounts, so the right prospect lands on the right rep's calendar without anyone sorting anything.
Every successful submission and booking fires a custom event to GA4 and Meta Pixel, so the whole pipeline stays measurable end to end.
-
The leads that vanish before they exist.
We caught a silent HubSpot failure that drops re-submitting unsubscribed leads, and built the form to recover them.
How it workedA failure most teams never even see: if someone who previously unsubscribed in HubSpot fills out your form again, the lead goes in, but HubSpot will never email them.
Their status is still "unsubscribed." No error, no warning. The prospect thinks they reached out, you think nothing's wrong, the lead just evaporates.
And HubSpot's API won't let you flip that status back - by design, only the contact can re-subscribe themselves.
So the form checks subscription status before it submits. If a visitor is unsubscribed, they get a modal explaining what happened and a HubSpot-generated link to fix it: they click, HubSpot emails them the steps, and once it's done they come back and submit.
The lead that would have silently disappeared now closes the loop itself.
-
"Call me now," but only when someone can pick up.
Instant callback for high-intent visitors, gated so the promise is only made when sales can actually keep it.
How it workedMost instant-callback widgets make a promise the business can't keep: click at 2am on a Sunday, nobody calls, and you've just burned trust on your highest-intent lead.
So we gated it.
The "Call me now" button only appears next to the usual "Book a call" when the conditions line up: the visitor is in a region a rep covers, it's a working day, and it's inside calling hours.
The timing check runs in a Cloudflare Worker, not the browser, so it reads real server time instead of whatever clock or timezone the visitor's device is set to. All the logic in one place, fast, and it can't be thrown off by a device clock that's wrong.
On click, the lead reaches the CRM like any other, plus a CRM automation fires a priority alert to sales. The rep calls within minutes, on a 5-minute target.
Meanwhile the visitor watches a countdown, so they know a call is coming. When it ends, the form asks one thing: did someone actually call? The system captures real delivery data instead of assuming the callback happened.
Working together
How I work with your team.
One engineer who speaks both marketing and engineering, owns a clear slice of your stack, and leaves no lock-in behind.
-
Marketing and dev that finally speak the same language.
Marketing wants it shipped yesterday. Dev wants it not to break. Usually they argue, and you wait.
Twelve years on both sides means I translate in real time:
- I turn what marketing wants into something dev can actually build
- I explain to marketing what something costs, what it breaks, and why it takes as long as it does
- no standoffs, no week-long waits for feedback, no requests lost in translation
Most developers push back on marketing requests. I find the way to make them work.
-
I own the martech layer, not your whole stack.
I am not a full-stack product engineer. I don't touch your core app.
I isolate and own:
- the marketing site
- tracking and analytics
- the lead pipeline
Marketing moves fast without touching the product team's deployment cycle. They build the product. I make sure the layer that feeds your pipeline never becomes the thing that breaks.
That job is the same whatever your site runs on.
And if it happens to be WordPress and Cloudflare, even better: I've built on WordPress since 2008 and work in Cloudflare daily, so less has to bounce to your dev team and there's nothing for me to learn on your time.
-
No lock-in. Built for handoff.
No black boxes. Nothing proprietary that only I understand, no tool that holds your data hostage, nothing your own team couldn't take over. When I write custom code, it's because it beats the off-the-shelf option, never to make you depend on me.
Every custom setup is documented for your internal team, step by step, with the how and the why. If I step away, any competent engineer picks up the stack.
I build to outlive my own involvement.
-
Your timezone is an asset, either way.
I'm based in the EU (Poland).
- if our hours overlap: you get same-day execution, not next-day delays
- if they don't: my timezone becomes your night shift coverage
I ship risky changes while your traffic is low. A problem that happens overnight is often fixed before your team even logs on.
I've run this across teams in the US, Europe, Brazil, and India.
-
The judgment AI doesn't have.
AI writes code fast and cheap, but the technical debt quietly piles up:
- half-understood scripts
- broken integrations
- silently drifting attribution
On a revenue-driving site, you need the judgment to catch errors before they ship.
I use AI daily to move faster, and rely on deep infrastructure experience to know exactly when not to trust it.
Testimonials
What people say.
The founders and leaders who trusted me with their revenue infrastructure.
-
I managed Szymon directly for over 5 years and he is one of the most reliable and technically skilled people I've ever worked with.
As VP of Marketing, I constantly needed things shipped fast, like new landing pages, GTM setups, CRM integrations, A/B tests... All on a site with millions of monthly visitors. Szymon always delivered without breaking site performance or stability.
What sets him apart is that he truly understands both sides, marketing and engineering. Most developers push back on marketing requests. Szymon found ways to make them work.
Over the years, he grew from a strong developer into a technical leader who built and mentored his own team, managed the technical roadmap, and became the person everyone relied on when things got complex.
One thing I want to highlight: his consistency. Over all those years, through countless projects and high-pressure moments, I never once doubted whether something would get done if Szymon was on it.
I'd love to work with Szymon again and recommend him without hesitation.
-
Szymon has worked with Neil Patel and I for over 10 years managing all things related to the development of our website.
He's been with us from when we were only driving a few thousand visitors a month to upwards of five million visitors per month.
He is one of the best developers that we have ever worked with. Highly recommend him for anybody operating a major website with a high volume of traffic!
-
I worked with Szymon in a high-velocity environment where WordPress changes could directly affect business outcomes.
While we weren't in constant collaboration, whenever something needed to be implemented on the WordPress side - especially changes impacting Ubersuggest landing pages hosted on neilpatel.com - Szymon was the person we trusted to jump in, unblock teams, and ship.
He's not limited to themes and UI tweaks; he can build real functionality from requirements, and he does it with excellent engineering hygiene (clean code standards, Git/versioning discipline).
When timelines were tight, he stayed calm, moved fast, and got it done.
I'd recommend him without hesitation.
Engagement
One engineer, no ramp-up.
Skip the hiring process. Here is how we secure your stack starting this week.
-
Start with an audit, not a rebuild.
Before touching anything live, I precisely map:
- what you already have
- where it leaks
- what to fix first
Low risk, zero disruption, and a clear picture even if we go no further.
-
If it's a fit, I take the layer.
I own the martech layer from there.
No hand-holding. You focus on strategy, I keep the revenue pipeline unbroken.
-
On a fractional retainer.
A few days a month, ongoing, not a one-off project.
I can start this week, with no headcount to justify and no months of onboarding.
When your marketing site brings in revenue, the cost of it breaking quietly is far higher than the cost of someone owning it so it doesn't.
or email me at szymon@szymonskulimowski.com