MetaCTO

Mobile app design and development that helps business owners create new revenue streams, increase customer lifetime value, and improve profit margins.

Partner

olive

Head of Engineering

Round

Round is a curated community of technology executives and senior leaders who support, grow, mentor, and learn from each other. The Round experience supports leaders who are at the forefront of technology, working on products and services that have an outsized impact on the world.

Founding Member

Power Digital Marketing

Team and product leadership for the Nova platform at Power Digital Marketing. The machine learning platform that can analyze any business’s growth opportunities, build a marketing strategy to execute on them, and track progress across 13+ channels in custom-tailored dashboards.

VP Engineering & Product

KlientBoost

Founded, launched and sold a marketing tech SaaS product.

- Grew team of 6 engineers plus product design and support staff
- Managed feature roadmap, technical architecture, development, delivery, and support
- Produced 50% trial-to-paid customer conversions
- Created education path for junior engineers
- Managed agile process of reviews, retros, standup, backlog

The flagship product for Kite Software integrated with Google Ads API to analyze accounts with custom algorithms and generate ROI focused recommendations. Scaled to process millions of records at a time with AWS serverless and queue services. Built with TypeScript, Node.js, Express, Python, Serverless, React, Redux, CircleCI, Docker, Gatsby, GraphQL.

VP Engineering, Kite Software

Maverick

Lead server side API development and web application team to support iOS startup venture creating a new social community for empowering young women. genmaverick.com

Develop microservices with NodeJS and MongoDB.  React and redux front end development.  Machine learning for moderation and transcriptions.  Rapid feature development including comments service, trending algorithms, and gamification progression engine.  Support PHP Zend MySQL API framework.  Mentor and train team members.  Manage technical contractors and support the agile development process.

Software Engineering Manager

JNR Incorporated

- Designed and developed a Node.js and MongoDB microservice-based API with a custom Express Gateway on an AWS Lambda-hosted serverless environment (api.paymentrise.com).
- Built a React and Redux front-end web application for customers to redeem prepaid reward cards with unique offer codes for BoostMobile clients.
- Led the technical team for new product development, defined software engineering processes, and implemented Agile Scrum methodologies.
- Created a custom Python script for serverless S3 deployment integrated with GitLab CI/CD.
- Introduced Git, Continuous Integration (CI), and Agile project methodologies to the team.
- Led the migration of a data center to an AWS cloud-hosted environment.
- Developed and supported large-scale Prepaid Card Industry applications for incentive programs, card activation, and card ordering.
- Built a REST API to securely interface new applications with legacy cardholder data systems, ensuring PCI DSS compliance at all levels.

Senior Software Architect, Lead

Blink Development, Inc.

Founded Blink Development with one partner and grew to a peak staff over 10 employees servicing website development and marketing.  Launched and serviced over 40 projects in 5 years, generating over $1.5M annual gross revenue.  Served as lead developer on all major projects.

Founder & Lead Developer

Fletcher Jones Motorcars

Started as a PC installation technician in 1999.  Advanced through many positions over 11 years with responsibilities including database administration, web development, help desk, project management, vendor management, and ultimately data warehousing and BI analysis with C#, .NET MVC, SharePoint, MSSQL Server, and SSRS.

Business Information Analyst, Developer, System Admin

At MetaCTO, we build custom mobile apps that help businesses grow new revenue streams, create online communities, increasing customer LTV with digital products, and make client communication more personal.

My first “exit” wasn’t a startup.

It was a Jeep.

Back in 1999, I hand-coded a car auction website for a local dealership.

Payment for my hard work = a black 1990 Jeep Cherokee, 4x4 
(valued ~$2,000).

At the time, I didn’t think of it as equity, or ROI, or “value capture.” 
My brain didn’t think about those things.

I just thought: I need a car to go hang out with my friends. They need a website. Deal done.

That dealership sold more cars. 

I drove home in one. 

Win-win.

Christopher Fitkin

Partner at MetaCTO · Custom Mobile Apps 📱

If you’re looking for a software dev team…

Make sure you hire a developer… NOT a prompt refiner.

There’s a lot of noise right now about devs focusing on AI “prompting” skills > coding skills.

Their mantra: ”Break problems into smaller chunks. Iterate on your prompts. Keep refining until AI gets it right.”

But that is not engineering.

Yes, engineers should decompose problems. That’s a given. But if your team is spending hours playing prompt roulette with ChatGPT… they’re not building your product. They’re wasting your time & money.

Here’s the bare minimum you should require:

- They write real code & use AI to speed up the boilerplate
- They use AI to review, test, & improve code (not “replace” it)
- They lean on AI for repetitive tasks… not for core business logic.

You don’t need a part-time coder, part-time prompt refiner. 
You need a full-time software engineer. 

AI is a great assistant… but still a terrible engineer.

I’m a CTO with 100+ successful software launches.

This is what I care about. And devs should too…

Do I care what text editor you use? No.
Do I care if you prefer React over Vue? Definitely not.

Here’s what I do care about when I’m hiring or leading devs on high-stakes products:

-

1. Can you ship production-ready code fast?

Speed matters. Especially in startups. But speed without stability is just a disaster waiting to happen. I want engineers who move quickly and leave a trail of stable, testable code behind them. Getting it out the door is great. Keeping it running at scale is the real job.

-

2. Are your commits small and useful?

Every commit tells a story. If yours say “misc fixes” or “final final changes,” we’ve got a problem. Clean commits are easier to review, easier to roll back, and show you’re thinking like an engineer, not a cowboy.

-

3. Does your code help the next person succeed?

Readable, well-structured code is one of the most underrated team accelerators. It speeds up onboarding, cuts down bug hunts, and lets your team scale without silly bottlenecks. If you’ve ever spent 2+ days trying to reverse-engineer someone’s logic, you know exactly why this matters.

-

4. Are you thinking ahead? Or just taking orders?

I’ll take an engineer who raises concerns, suggests improvements, or flags edge cases over someone who just ticks off tickets ANY DAY OF THE WEEK! Great teams don’t need micro-managing. Surface problems before they become blockers.

-

5. Are you growing? Or just clocking in?

The best engineers I’ve worked with are always leveling up. Whether it’s digging into app performance, learning a new AI model, or just asking better questions. Growth-minded devs push the whole product forward and are worth their weight in gold! Can the right mindset scale? — yes, yes it can.

-

Tech stacks change. 
Teams change. 

But these 5 traits hold up across every high-functioning build I've led.

This is how great products get shipped.

Here’s an idea…

Move fast WITHOUT breaking things.

Before I ship anything (especially something customer-facing or tied to revenue) I try to answer 3 questions. 

Doesn’t take long. But it keeps me and my team moving fast + safe (& clients happy).


Question 1: 
What happens if I’m wrong?

Not all mistakes are equal. A harmless UI hiccup isn’t the same as breaking data integrity or disrupting a payment flow. When the stakes are high, it’s okay to slow down, review the assumptions, and involve a second set of eyes (that’s where @Garrett Fritz comes in). But if the impact is minor, I’ll take the risk and fix forward. It’s simply understanding the cost of being wrong.


Question 2: 
Can I reverse this easily?

A deploy is a lot less stressful when you’ve got a kill switch, a clear rollback plan, and alarms that actually catch issues (ideally in real-time). If I know I can roll something back in 10 mins, I bias toward action. But if reversal means days of damage control or downtime, I won’t push until we’ve rehearsed the failure modes and have a clean exit strategy. Worth it.


Question 3: 
How big is the blast radius?

This is the big one. Think: Can I test this behind a feature flag? Ship it to 5% of users first? Trigger it only under specific conditions? If I can contain the impact, I’m totally fine experimenting. But if it touches everything and we’re not confident in the boundaries, that’s a flag to scope things smaller. Or even hold off entirely.


3 quick questions… and now speed and safety aren’t at odds. 

I think of it like ‘CMD + Z’… but for projects. You just assess risk BEFORE you move. (Crazy, i know)

Devs: “AI makes me faster.”

Data: “AI makes devs 19% slower.”

I’ll be honest… I thought AI made me faster too.

I use AI tools daily. I’ve shipped entire internal tools with AI scaffolding. And like most devs, I feel faster. 

But this new RCT study has me rethinking that belief. The researchers looked at 16 experienced OSS contributors fixing real bugs and features in their own repos. 

→ Developers expected AI to make them 24% faster. 
→ But in reality, they were 19% slower when using AI.

Even after seeing the slowdown, they still believed it made them faster. That gap between perception and performance is wild… but honestly, I’ve felt it too.

I’ve had my share of trust issues with AI code. I still spend time auditing logic, fixing subtle errors, and double-checking edge cases. But I assumed that was outweighed by the time saved on implementation.

This study makes me wonder: how much of the "speed" is just shifted effort?

More code isn’t always more progress. 
Especially if we’re just creating messes faster.

So I’m really curious to hear from other devs: Do you think AI make you faster? Or just… busier?

Not all “senior” devs are created equal.

Here’s how I separate the pros from the pretenders:

After working with (and hiring) hundreds of engineers, I’ve learned that resumes rarely tell the full story. 

Years of experience? Is that 10 years of actual high-level dev work? Or 10 years with a computer and YouTube certificate? 

Let me save you some trouble…

These are 5 traits I look for before a dev gets anywhere near our team:

_____

01/ Do they think in tradeoffs? Or rules?

Mid-level answer: 
“We never touch prod.”

Senior:
“I’d touch prod only when rollback is safe, logs are clean, and metrics are monitored.”

_____

02/ Can they debug under pressure?

Mid-level answer: 
“It’s probably the database... or the last PR... or maybe the moon cycle.”

Senior:
“I’d check logs, compare graphs, and ask for a peer review before touching anything.”

_____

03/ Portfolio code? Or team-first code?

Mid-level answer: 
"I thought it would be cool to try this new pattern I saw on Reddit."

Senior:
"Readable, maintainable, and boring code is the job."

_____

04/ “Yes” person? Or know when/how to push back?

Mid-level answer: 
“Sure, I can add that to the API.”

Senior:
“What’s the actual user need? Do we really need to own this complexity?”

_____

05/ Order taker? Or problem solver?

Mid-level answer: 
“I’d just build what the ticket said.” 

Senior:
“This spec has edge cases — here’s what I’d adjust, and why.”

_____

If you’re hiring a dev… Judgment > credentials.

Back in January I wrote a “viral” post (~34k views). The topic = execs at Meta, Salesforce, and others saying 2025 was the year AI would replace developers.

We’re about to enter October — month 10 of 2025. And it’s not even close.

Here’s the reality I’m seeing on the ground:

- Eval systems = still slow, inconsistent, & unreliable.
- Developers still spend more time fixing AI’s “help” than shipping production code.
- AI tools generating impressive snippets… but choking on new frameworks, unique business logic, + non-linear systems.

Meanwhile, the incentives for big-name execs haven’t changed:

- Promise massive productivity gains to shareholders
- Sell more AI tools
- Justify cost-cutting engineering headcount

I’ve seen AI make teams faster in some areas… i.e. boilerplate code, prototypes, automation.

But the idea that we’d see fully autonomous AI agents replacing senior devs this year was never realistic.

As we close out 2025, this is my take for the future:

👉 AI will continue to shape software development, but it won’t fully replace developers. Not even close.

Will AI take some coding jobs? = YES

Will AI learn how to build products that combine AI’s speed with human judgment? = NOPE

“Bad estimates kill more software projects than bad code.”

Couldn’t agree more… (pay attention PMs!)

Props to Raul Junco who recently wrote a post on this idea. Like Raul, I’ve seen this cycle too many times:

1. A project gets ballparked at 6 weeks.
2. Management pushes for 3.
3. Developers explain that unknowns always show up — it’s part of building.
4. PMs insist: “Just give us 3. Update if it slips.”
5. Team delivers in 6. Exactly as predicted.
6. Retro: “How did this 3-week project turn into 6?”

And the cycle repeats.

The problem = pretending estimates are promises instead of guesses
NOT the problem = engineering discipline

Good engineers surface risks early & adapt as unknowns come up. That’s the job.

But if (when) management demands certainty… everyone loses (the product slips, the team burns out, and trust erodes on both sides). Not a winning combo.

Better idea PMs =

Treat estimates as ranges (not deadlines). 
Accept you’re planning for unknowns. 
And measure the success by outcomes… not an engineers ability to predict the future.

LTV:CAC is a terrible metric for subscription apps.

(bc retention isn’t linear in 2025)


Myth 1: 
“LTV tells us how much each customer is worth.”

Reality: 
People believe it because it feels like a clean, simple number investors love. But most apps can’t calculate LTV accurately. Subscription lifecycles are messy: users cancel, ghost, and re-subscribe months later. Seasonal churn, paywalls, pricing changes… all of it warps the “lifetime” view. You’re making bets on averages that don’t reflect how people actually behave.


Myth 2: 
“CAC is easy to measure, so we just need to divide it by LTV.”

Reality: 
CAC varies wildly by channel, campaign, and creative. And if you’re not segmenting cohorts, you’ve got no clue. Pair that with an overestimated LTV and suddenly your “healthy” ratio is leading you straight into negative unit economics.


Myth 3: 
“If LTV:CAC is >3:1, we’re good.”

Reality: 
I think someone said this on a panel in like 2015 and people just never questioned it. But it’s wrong. That rule of thumb might work for B2B SaaS with high retention and predictable contracts… but for mobile consumer apps it’s borderline meaningless! What actually matters: payback period, gross contribution after CAC, and cohort-level ARPPU snapshots. They tell you what your users are worth right now, not in theory.


If you’re building or scaling a subscription app, the smartest thing you can do isn’t tweaking LTV:CAC… it’s throwing it out.

Need help figuring out what to track instead? 
Shoot me a DM!

This one made me laugh harder than I’d like to admit. 😂

Happy Friday everyone!

Duolingo’s paywall strategy is brilliant.

(even if you hate it)

I’ve been doing this for awhile. (20+ years)

I see companies treat their paywall like just another a popup. But there’s another way… 

👉 Treat your paywall like a product.

Duolingo is one of the best examples of this. 

And the data backs it up: their top-tier plan, Duolingo Max, has seen increased ARPU and broader tier adoption, while their user base keeps growing.

So what are they doing differently?

1. Paywalls are timed, not just shown

Duolingo doesn’t throw up a wall when new user starts clicking around. It builds momentum. Shows progress. Gets users emotionally invested in their learning BEFORE asking them to commit. Patience is a virtue.

2. Paywall = part of the experience

Personalized feedback. AI coaching. Even video calls. These are positioned as improvements to your experience, not barriers to it. Paywalls will always add some friction. But Duolingo turns a little friction into a value opportunity.

3. Users are reminded WHY they paid

Post-purchase experience matters! After you subscribe, Duolingo doesn’t leave you wondering what you bought. It reinforces the benefits with visuals, icons, and subtle celebration mechanics. Users told them that the “aftercare” is a big reason they stick around. Good idea Owl!

4. It’s all driven by data

This seems almost unnecessary to say. But it’s just too important not to mentions. Yes, some users find the push toward premium a little aggressive. But… Duolingo has 100+ million free users to test with. They iterate fast, optimize hard (WHILE ALSO tracking retention, refund rates, and churn). So yeah, you can balance monetization with trust.

And Duolingo isn’t alone.

Apps like Blinkist and Strava have tweaked their full subscription funnel. Not just the moment of conversion…. but the journey around it. 

The pre-frame. 
The context. 
The moment after.

So what do I want you to takeaway from this post? 
(Besides… don’t mess with The Owl)

A paywall is NOT just another screen. 

It’s a product strategy opportunity that you should give a HOOT about.


PS
Their paywall strategy is great, but… I still can’t speak French. 

So…

Coinbase. Citi Bank. Stripe. Lemonade. 

Big companies are starting to mandate AI coding assistants for their developers.

Velocity is up. ☝️
But what is also up?… Security risks. Up big time. 10x up. ☝️

Apiiro’s data shows what I’ve been seeing:

AI performing pretty well at surface-level fixes (fewer typos, cleaner syntax - the boring, but necessary stuff). But when it touches multiple services or architectural flows, the blast radius is enormous! I.e. one AI-driven pull request could quietly break authentication across an entire system.

Bad code is one thing.

But the real danger is the structural flaws you don’t catch until it’s too late.

And if Fortune 50 companies with armies of reviewers and security engineers are struggling to keep up… imagine what happens inside a 5-person dev team or an agency running fast on your dime.

If you’re a founder or exec, here’s what you should require from anyone writing code for your business:

- DON’T use AI as a replacement for judgment
- DON’T sacrifice security for speed
- DO use smaller, isolated pull requests (with clear rollback plan)
- DO rely on human review (that focuses on architecture + business logic)

It should be pretty obvious this point…

More speed is only a win IF you can control the blast radius.

Taplio

Christopher Fitkin's Linkedin Analytics

Who is engaging with Christopher

's Best Posts (last 30 days)

Want to drive more opportunities from LinkedIn?

Taplio

.css-1jgreyo{background-image:linear-gradient(284deg, rgba(39,43,72,1) 0%, rgba(33,47,183,1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Christopher Fitkin's Linkedin Analytics

Who is engaging with Christopher

's Best Posts .css-qqfgvy{font-size:var(--chakra-fontSizes-md);}(last 30 days)

Want to drive more opportunities from LinkedIn?

Christopher Fitkin's Linkedin Analytics

's Best Posts (last 30 days)