After a client approached me about improving their Google Ads campaign, I could tell they were doing a lot of things right—but they weren’t getting the results they wanted. Their ads were active, their click-through rate was decent, and they were spending about $650 per month. On the surface, everything seemed fine.
But as I dug deeper, I saw an opportunity to transform their campaign from “good enough” into something that could truly drive new clients to their business.
Google Ad Account Audit
The first thing I noticed was that their ads setup was missing some key pieces. For starters, there was no way to measure what happened after someone clicked on an ad. Conversion tracking hadn’t been set up, so there was no way to tell if those clicks led to sign-ups, contact form submissions, or any other meaningful action. It’s like running a sales campaign and never finding out how many people bought something.
Another issue was the structure of the ads themselves. Everything was running through a single ad and ad group, all of which pointed to their homepage. While this kind of setup can work as a starting point, it’s not ideal for targeting specific audiences. Imagine searching for “self-defense classes” and landing on a generic homepage with no mention of self-defense—it’s a missed opportunity to connect with a potential client.
Finally, their campaign wasn’t linked to Google Analytics, which meant there was no insight into what visitors did once they arrived on the site. Did they spend time exploring? Did they leave immediately? Did they look at multiple pages? These questions were left unanswered.
Google Ad Campaign Improvements Plan
After reviewing their campaign, I proposed a complete overhaul. First, I explained the importance of defining what success looks like—what actions do we want people to take when they click on an ad? For this client, it made sense to track things like contact form submissions and clicks to their scheduling platform. By setting up conversion tracking, we could finally measure the real impact of their ad spend.
Next, I suggested restructuring their campaign into three distinct ad groups, each with its own audience and purpose. The first ad group would target people new to their services, highlighting beginner-friendly options. The second would focus on fitness enthusiasts looking for a fun and effective workout. The third would spotlight programs for women, emphasizing empowerment and inclusion.
Each ad group would feature two different ads to test different messages and see what resonated most with each audience. And instead of pointing all traffic to the homepage, I recommended creating dedicated landing pages tailored to the ad group. If someone clicked on an ad for fitness classes, they’d land on a page showcasing the physical and mental benefits of their programs. If they clicked on an ad for women’s classes, they’d be greeted with content specifically designed for that audience.
Finally, I emphasized the need to connect Google Analytics to their campaign. This would give us a clearer picture of how people interacted with the site after clicking on an ad, providing valuable insights for ongoing optimization.
Marketing & Development Implementation
The client loved the plan, and here’s what I did next. I spent a couple of hours implementing these changes, setting up the ad groups, writing the ads, building the landing pages, and integrating analytics and conversion tracking. Once that was done, we were ready to relaunch the campaign with a much stronger foundation. I used the Google Site Kit plugin for WordPress to manage analytics.
I built the landing pages as part of their WordPress website. I drafted the copy for each page, added relevant photos, and applied proven design principles that would guide potential customers through the sales funnel. I used the same copy outline and content layout for each.
Landing Page Content Outline
Hero Section
Headline
Subheadline
CTA Button
Benefits Section
Headline: Why this offer is valuable
Highlight key benefit 1
Highlight key benefit 2
Highlight key benefit 3
What to Expect / Program Details
Headline: Set expectations
Short paragraph explaining details or structure
CTA button
Testimonials
Headline (Social Proof): “What are customers are saying”
Testimonial 1
Testimonial 2
FAQ Section
Headline: Answer common questions
Q1 & Answer
Q2 & Answer
Q3 & Answer
Final CTA Section
Headline (Closing statement)
CTA button
Getting Results with Google Ads
Of course, ads don’t run themselves. The client will need to decide when they want to turn the campaign back on and how much they’re comfortable spending each month. But with these optimizations in place, they’ll be able to make data-driven decisions and see exactly how their investment is paying off.
Every business wants to get the most out of their marketing budget. For this client, that means attracting new customers who are truly interested in their services. By taking the time to set up conversion tracking, target specific audiences, and create tailored landing pages, we’re turning their Google Ads campaign into a powerful tool for growth.
If you’re running Google Ads and feel like you’re not getting the results you should, you’re not alone. With the right strategy and a little bit of work, you can transform your campaign into something that delivers real, measurable value.
Want to learn more about how I can help your business grow? Let’s talk.
An operation to rotate an array will shift each element to the left. We can perform this function X number of times depending on the input parameters. Since the lowest index element moves to the highest index upon rotation, this is called a circular array.
All we need to do is remove the first element of the array, save it in a variable, and then attach it to the end of the array. This can be done inside of a loop that iterates the number of times that we want to rotate the data. The function can take in the original array and the number of rotations. It returns the newly rotated array:
Although this code will work, and gets the job done, it is slow. When I tried this as a solution on HackRank (a code challenge website) it passed 9/10 test cases. It failed once, citing “time limit exceeded”.
The problem is with array_shift. It’s a PHP function that removes “the first value of an array off and returns it.” This shortens the array by a single element, moving everything else down one index. The algorithmic efficiency (expressed in Big-O notation) of array_shift() is O(n). That means, the larger the input array, the more time it will take to complete.
Next, I tried using array_reverse and array_pop. I figured that since array_pop() has a constant algorithmic efficiency, noted as O(1), it would help. Regardless of the size of the input, it would always take the same amount of time. But, due to the use of array_reverse (twice!) it was even slower:
This code does not need to use any kind of loop, which helps to speed things up. First, array_slice() grabs the elements up to the point that the data should be rotated, and saves them to the variable $remaining. Next, array_splice() removes those same elements. Lastly, array_merge() glues the elements back together in the expected, rotated order.
This sort of computer science programming challenge can commonly be found during software engineering job interviews. When coming up with an answer, it is important to consider speed and performance. Understanding computing cost-effectiveness and Big-O can be a deciding factor in how your coding skills are judged.
The look-and-say sequence is a series of integers. It can grow indefinitely. It is generated by reciting a number phonetically, and writing what you spoke numerically. Its popularity is attributed to famed cryptographer Robert Morris. It was introduced by mathematician John Conway. It looks like this:
1
11
21
1211
111221
312211
13112221
The first line would be pronounced as “one 1”, and then written as “11” on the second line. That record would be spoken as “two 1’s”, giving us the third line “21”. The greatest individual symbol you’ll ever find in this consecution is a 3.
This topic has lots of trivia, variations, and history that could be dug up and expounded upon. Here, I’ll explain a solution written in PHP to produce this chain of numerals. The input will be the count of how many lines, or iterations, in the series to generate. Below is the code:
<?php
echo "Count And Say: \n";
function countAndSay($count=0){
$value = 1; // initial seed
for($i=1;$i<=$count;$i++){
echo $value . "\n";
$value = calcOutput($value);
}
}
function calcOutput($value){
$value = "$value"; // change it into a string, so we can iterate over each character
$current = $value[0]; // first character
$count = 1;
$return = '';
for ($i = 1; $i <= strlen($value); $i++) { // keep going until we get through the whole string
if ($current != $value[$i] || $i == strlen($value)) { // found a different character, or end of the input string
$return .= "$count$current";
$count = 1; // reset count
$current = $value[$i]; // set new current character
} else {
$count++;
}
}
return $return;
}
countAndSay(7);
echo "\n\n";
?>
I separated my code into two functions. I think this is the best approach. As an exercise, see if you can figure out how to refactor it into one. This could help you to internalize the logic as you write it out for yourself.
The initial seed value is “1”, and that is hard-coded at the top. The for-loop iterates based on the count input parameter. That means the code circles back and re-runs, with updated values, until its internal count (represented by the variable $i ) matches the $count variable passed into countAndSay($count).
The code that we loop over outputs the current sequence value (starting with 1) as its own line (“\n” will output a new line in most programming languages) , and then calculates the next. The function that determines the next line of output, calcOutput($value), takes the current value as an argument.
The first thing we do is cast the integer value passed along into a string. This lets us refer to each character by index – starting at zero – and save it to a variable $current. We start a new $count, to keep track of how many times we see the same digit.
The next for-loop executes for the length of the $value string. On each loop, we check if the $current character we saved matches the subsequent one in that $value string. It is again referenced by index, this time based on the for-loop’s iteration count represented by the variable $i.
If it does match, one is added to the $count variable that is keeping track of how many times we see the same character is a row. If it doesn’t match (or we’ve reached the end of the input), the $count and $current number are concatenated to the $return element. At that point, the $count is reset to 1, and the $current value is updated.
Writing an algorithm to generate the look-and-say (also known as, count-and-say) sequence is a common coding puzzle. You might run into it during a job interview as a software engineer. As practice, see if you can simplify my example code, or even write it in a different programming language than PHP.
How to add an underline to website text should be covered in any intro to web development course. The old-fashioned HTML way uses a now-deprecated tag:
<u>This will appear underlined!</u>
Modern approaches use CSS to define such a style:
<p style="text-decoration: underline;">This will appear underlined!</p>
Even better, properly written code will separate the inline styles, like so:
But, suppose I want to have that underline to become thicker instead of disappearing. That will require an advanced, super-secret CSS technique. To make it work, we will utilize box-shadow.
In the world of cascading style sheets, the box-shadow property adds a shadow effect around any HTML element. The shadow is described by its offsets, relative to the element. Leveraging this, we can create a shadow that looks like an underline. On hover, we can adjust the settings to change its appearance:
Writing is the best way to develop new ideas. It exercises our ability to explain what we mean, what we want, and why. Clear writing is clear thinking.
Putting words down is good for our mind, and is calisthenics for the brain. Forcing ourselves to build coherent, connecting streams of thought flexes mental muscles that we usually ignore. Expression and mutation are two sides of the same coin.
It’s easy to state your opinions at a high level. But once you dig into it, you might realize you haven’t thought it through. At least, not recently. Writing adds flesh to otherwise gaunt stances. It turns shape-shifting clouds into crystals with obvious foundations.
Make it a habit. Turn it into a daily practice. Here are ten tips to develop a writing routine:
Spend time writing fiction. Short stories, character profiles, fantasy, and fan-fiction each count. Glue yourself to the keyboard, until you can say “That’s my story, and I’m sticking to it.”
Journal daily. Type out the things for which you have gratitude. Document your feelings. Spell out your hopes. Cast a silent incantation. Craft an ancient curse. The universe is listening. I wrote a blog post about journaling, you can read it here.
Don’t be a perfectionist. Put sentences down, even if you don’t like them. Keep them around, even if you hate them. But please remember, that doesn’t apply to people in your life.
Everything is a work-in-progress. Revisit your posts, essays and articles. Add that literary polish and be a wordsmith. This will help keep you from trying to be perfect.
Start a blog. Build in public. Let the world see how you think. Most people aren’t paying attention anyway. And if you can’t come up with something new, update existing pieces.
Editing counts. Delete unnecessary sentences. Remove unneeded words. Clarify things, find synonyms, and avoid using the word “really”.
When I edit, I never actually throw anything away. I borrow the “soft delete” methodology from software engineering. I flag writing as unusable, without actually getting rid of it. That way I can revisit it, use it for subsequent ideas, and remember where my mind was.
Hoard notes. Keep a physical notebook. Use a note-taking app. Bookmark stuff. Markup books while you’re reading them. Keep a stack of index cards near by for when inspiration strikes
Write more than you promised yourself you would, but only if you feel like it.
The Great Resignation is an early 2020s buzz phrase. It describes a large portion of the workforce quitting. This Big Quit has been precipitated by numerous factors, but most obviously the CoVid-19 pandemic. Truthfully, the snowball started rolling prior to the worldwide shutdown. Online freelancing and digital nomadism have been picking up steam for nearly a decade. Ubiquitous Wi-Fi, inexpensive devices, and on-demand cloud computing paved the road of the Extraordinary Exodus. The ability to work remotely, I believe, is one of the strongest drivers of this cultural shift.
Employment as we know it is a raw deal. It exploits workers and business owners. Only since the advent of the industrial revolution has the modern work schedule and environment taken this shape.
Employees are tasked to toil in serfdom. The employer is expected to be a parent. Those costs, risks, and responsibilities weigh heavily on executive shoulders. And that gravity ultimately crushes workers as exploitation.
Employer sponsored medicine took hold after World War 2. To fight inflation, the 1942 Stabilization Act was passed. That law prevented employers’ from raising wages. As a work-around to compete for high-demand workers, companies began offering health benefits as a competitive incentive. Nearly a century later, healthcare and employment are intertwined in the bleakest of scenarios.
Now can be the epoch of freedom. It’s the original promise of all the technological advances we now enjoy. Greater prosperity, with fewer shackles to hold us down. That also means businesses can enjoy larger output with less overhead.
When writing this, I was careful not to refer to employers as bosses. Ultimately, they are the clients. Workers should be expected to “manage up”, as they control labor, and in turn, production output and capital income.
What about health insurance?
It’s an absurd burden that businesses are expected to provide employees with healthcare. With that type of expectation, employers feel justified squeezing out every last drop of labor juice – be it sweat or tears.
Even worse, as workers, it’s crazy that access to medicine is tied to employment. Public health should be a public service, paid for by taxes. And, I’m not suggesting that taxes should be raised, but instead reallocated from bureaucratic waste, corruption, and cruelty.
Healthcare is among numerous benefits subsidized and off-loaded to the private sector. It has a taste similar to the Western tipping scheme found in the restaurant service industry.
Freedom
Nine hours in an office each day doesn’t leave time for much else. It was a routine I faced for years, along with many other New York straphangers. A long daily commute was the little time I could find to read, meditate, or relax. My only exercise came from pushing deadlines, jumping through hoops, and squeezing in meetings.
An age of abundance is possible within our lifetime. Laborers are being replaced by technology. Remember, most jobs are bullshit anyway. Humanity can feed itself, and anything more is just pollution.
Freelance
A freelance side hustle is the first step in buying your freedom. Owning, and running, your own remote business comes pretty close to what I imagine as the Digital Dream. When working as a freelance web developer I follow a process to ensure success. I am able to pitch potential clients on improvements, audits, and maintenance agreements.
Recently, one of my websites went down. After noticing, I checked my EC2 dashboard and saw the instance stopped. AWS had emailed me to say that due to a physical hardware issue, it was terminated. When an instance is terminated, all of its data is lost. Luckily, all of my data is backed up automatically every night.
Since I don’t use RDS, I have to manually manage data redundancy. After a few disasters, I came up with a solution to handle it. I trigger a nightly cron-job to run a shell script. That script takes a MySQL dump and uploads it to S3.
As long as I have the user generated data, everything else is replaceable. The website that went down is a fitness tracking app. Every day users record their martial arts progress. Below are the ten steps taken to bring everything back up.
After writing my last post about How to create a Shopify app, I decided to build a new one as a side project. Taking myself through the entire process helped me to tighten up the details I mentioned. This one adds a sticky banner to a store’s front-end, prompting users to “click to call”.
It’s built on top of the code I used for the SplitWit Shopify app. I adjusted some of the methods to accept configuration parameters to differentiate between the two. Code had to be added to support new functionality. SplitWit already had a feature to add a sticky banner to a site’s existing UI. I used the same workflow to inject the merchant’s settings as JavaScript.
The app’s admin view is a simple input form with settings to control the sticky bar UI that is injected into the merchant’s store-front.
In addition to updating and refactoring my code, I wrote copy and drafted design for this digital product. I used SplitWit branding guidelines (fonts, colors, etc.) to establish an adjacent feel.
Although it’s optional, I wanted to include a promo video in the listing. Having had previously created videos for SplitWit, I was able to quickly spin one together. I already created background music files in Garage Band for other projects. Here’s the one I chose to use – feel free to borrow it for what ever you like. The text animations were exported from Keynote. I added screenshots, included stock animation from VideoPlasty, and recorded voice-over lines using a Yeti microphone.
I drafted other graphic assets that were required in the app listing using the GIMP – software I’ve used for over twenty years
A few days after submission, I received an email with required changes.
They were mostly minor issues. Things like the app’s name, a screenshot used in the listing, and an OAuth redirect bug.
One of the requests said, “Update your app to theme app extensions to ensure compatibility with Online Store 2.0 themes.”
2.0 themes? What does that mean?
Shopify recently announced Online Store 2.0 (OS 2.0). It’s essentially a set of improvements to the platform that makes themes and apps more flexible and maintainable. This benefits both merchants and developers. Enhanced app support means app functionality can be leveraged anywhere in a theme by using app blocks in the theme customizer.
SplitWit Click-to-Call injects HTML to manipulate a store’s user interface. That code comes from a JavaScript file that’s referenced in the page’s source code. That reference is added upon installation using the ScriptTag API. The JS file itself is generated & updated whenever a merchant clicks “save” in the app’s admin view. This required change is requesting that we provide an app block option as an alternative for compatible themes.
Shopify is encouraging OS 2.0 apps to instead use “theme app extensions” because they don’t edit theme code. It allows merchants to add your app’s UI elements, along with any settings, through its theme editor.
The documentation mentions that it “reduce[s] the effort required to integrate apps in themes”. In my particular case, it actually seems to add a step.
Theme App Extensions
App blocks are a type of “theme app extension” supported by Shopify’s Online Store 2.0. To create an app block available in the theme editor, I added a directory to my app using the below file structure. I was able to auto-generate it with the Shopify CLI by using the command shopify extension create.
From the command line, I registered this folder as an extension, and pushed my code.
In my app block .liquid file I used the same HTML template from my original PHP snippet, swapping my database variables for Shopify block settings.
The schema JSON explicates the settings inputs exposed to the merchant. I set them to match the original settings view from my app’s interface.
This approach lets Shopify maintain the app’s configurations, instead of my SplitWit server database that’s hosted on AWS EC2. That’s less data developers can capture, but also less of a hosting burden.
Any time you update the theme app extension you’ll need to re-push the code from the Shopify CLI with the command shopify extension push. The extension code is not hosted on your own server. It lives solely in the Shopify infrastructure ecosystem.
Verify theme support
Not all themes support theme app extensions. Theme support needs to be verified at the time of installation.
The original settings view is still needed, just in case the merchant’s published theme does not support app blocks. If app blocks are supported, I don’t install the script tag snippet at all. Instead, the settings view is replaced with integration instructions telling the merchant how to activate the sticky banner from the theme editor.
If app blocks are not supported by the active theme, the snippet is installed and the settings input form displayed. Determining if the merchant’s theme supports app blocks requires adding the read_themes Shopify API scope access to the oAuth request.
I tested my code by switching from the basic “Minimalist” theme to a OS2.0 theme called “Dawn”. When the app is being installed, I do a few things to check if app blocks are supported:
Get a list of the merchant’s installed themes, and check which is currently published
Retrieve a list of assets in the published theme
Check if JSON template files exist for at least one of the desired templates
$params = [];
$json_string_params = json_encode($params);
$headers = array(
'X-Shopify-Access-Token:' . $access_token,
'content-type: application/json'
);
// $install_ctc_curl_response_json = $this->curlApiUrl("https://www.splitwit.com/service-layer/click-to-call-service.php?method=installShopifyApp&installation_complete_id=".$installation_complete_id, $params);
// check if this merchant's published theme supports app blocks
// https://shopify.dev/api/admin-rest/2021-10/resources/theme
$read_themes_url = "https://" . $this->api_key . ":" . $this->secret . "@" . $shop . "/admin/api/2021-10/themes.json"; // list of all installed themes
$read_themes_curl_response_json = $this->curlApiUrl($read_themes_url, $json_string_params, $headers, false);
$themes = $read_themes_curl_response_json['themes'];
$published_theme_id = 0;
foreach ($themes as $theme) {
// live theme has a role of main
if($theme['role'] == "main"){
$published_theme_id = $theme['id'];
// echo "The main theme is " . $theme['name'] . "<br /><br /><br />";
}
}
// Retrieve a list of assets in the published theme
$get_theme_assets_url = "https://" . $this->api_key . ":" . $this->secret . "@" . $shop . "/admin/api/2021-10/themes/".$published_theme_id."/assets.json";
$get_theme_assets_curl_response_json = $this->curlApiUrl($get_theme_assets_url, $json_string_params, $headers, false);
// Check if JSON template files exist for at least one of the desired templates
// For other applications, you might want to check that they exist for ALL desired templates
$assets = $get_theme_assets_curl_response_json['assets'];
$probably_block_support = false;
$templates = ['index', 'cart', 'page.contact', 'product', 'collection'];
foreach ($assets as $asset) {
foreach ($templates as $template) {
if($asset['key'] == "templates/".$template.".json" ){
$probably_block_support = true;
break; // this checks that JSON template files exist for at least one of the desired templates. If you want to check that they exist for ALL desired templates, you can move this break to the 'else' condition
}else{
$probably_block_support = false;
// break;
}
}
if($probably_block_support){
break;
}
}
Shopify recommends additionally checking:
The body of JSON templates to determine what section is set as `main`
The content of each `main` section and if it has a schema that contains a block of type ‘@app’
I check for that value from the front-end to display steps for integrating the app block from the theme customizer. Shopify guidelines require that we provide merchants with post-installation onboarding instructions. Those directions replace the settings input form. Configurations will be managed through the block itself.
These updates satisfied the app review team’s request. I responded to their email, and a day later received their reply. It complained that the “app doesn’t have a functional user interface (UI)” when app blocks are enabled. That was because all of the settings were being managed by the app block data. To solve this issue, I moved the phone number and message fields back to the app’s settings view. I saved those values as metafields using the Shopify REST API.
Update: The Shopify metafield POST API would only create/update a single metafield per request. I had to break it out into two calls. From what I’ve read it seems like a PUT request might do that trick for multiple fields, but for my use-case this approach is fine. Here you can see how I do it when setting default metadata on installation:
After another response, they commented that I should be using “App Embed Blocks” instead of just “App Blocks”. That was because my UI component is a “floating or overlaid element”. It exists outside of the normal DOM flow and was not inline with other HTML nodes. This only required a small update to the liquid file’s schema, changing the “target” from “section” to “body”.
Although only a small difference, it does affect how merchants add Click To Call in the theme customizer. They must navigate to the Theme Settings area, and add it as an “App Embed”.
Luckily, I’m able to deep link merchants directly to that view from my onboarding instructions. The link also automatically activates my app embed. All I needed to do was get the extension’s UUID by running shopify extension info from my command line and I was able to build the URL.
Add the Click To Call App Block from <a href="https://<?php echo $clickToCallRecord['shop']; ?>/admin/themes/current/editor?context=apps&activateAppId=b52ccd8e-54b1-4b6d-a76f-abaed45dea97/click-to-call" target="_blank">the theme editor</a>
I updated my app’s home screen onboarding instructions to reflect this new flow. Everything appeared to be working when I tested, yet the app review team complained that the above issues were still unresolved. It turns out, I was able to immediately see changes to the extension that I pushed from the CLI because “development store preview” was enabled. The review team could not until I published a new version:
After that fix, the app was accepted to the Shopify App Market. If you are a Shopify merchant, check it out and let me know what you think.
cURL is a PHP library that lets you make HTTP requests. To use it, you need to install the libcurl package. If you’re using PHP on web server, it’s probably already there. The cURL functions have a number of options, depending on what request you’d like to make. You can read all about it in the PHP official documentation.
I had a number of projects that was using it all over the place, to make a variety of requests. I was repeating a lot of code, and my mix-matching began to get confusing. Finally, I wrote a function that takes request options as arguments and does all the work:
cURL’s default request type is always GET. If I want to use a different HTTP verb, I can specify it as an argument. Adding this function as an abstraction layer over existing methods helps me get things done more quickly, and kept my code clear, clean, and under control.
