Every product and marketing team wants clean, actionable data—but actually getting there is tough. Shifting browser policies, consent regulations, and fragmented tooling are the new normal. Even well-resourced teams can spend months building systems where all the components play nicely together.
I’ve spent years solving tracking challenges—making sure events fire reliably across platforms and that the data tells a consistent story. This piece is a reflection of those lessons—hard-won through mistakes, rewrites, and the occasional fire drill.
Whether you're fixing a brittle setup or starting fresh, I hope my experience will help you to avoid the common pitfalls and get measurement working.
Pick Right Analytics Foundation
First of all you need to pick the analytics tool that matches your product structure and monetisation model. It’s fair that most of the tools are very advanced, however, they still have different pros and cons. And considering the overall cost, make sure you pay for the right service.
There are plenty of tools on the market: GA4, Amplitude, and Mixpanel to name a few. In my experience you should focus on what your business focus is. If your company is heavily invested in Google Stack, Google Ads and Google-based solutions, you are likely to benefit more from using Google Analytics.
At the same time, if you are looking for a more mature tool oriented on product development, Amplitude might be a great choice. It has one of the most user-friendly interfaces and is incredibly agile when it comes to in-platform analytics. Identify core products that you're going to measure and pick a tool that was purpose-built to solve that problem.
Lastly, make sure you consider the cost. Amplitude and GA4 both offer free tiers, however, the limits are different. Paid versions also start on various breakpoints. Make sure the tool you are picking will fit long term. It’s very unlikely that you will consider changing tools once the tagging and tracking are in place.
Design a Robust Data Layer
A DataLayer is a structured JavaScript object (usually window.dataLayer) embedded on a website to store and pass information about a user’s interaction with the site. It acts as a central place for your data, where it gets distributed to digital marketing platforms and analytics tools. dataLayer is populated by the developers team and it unlocks tracking and measurement if done right.
The main issues with dataLayer implementations are a lack of structure and a lack of consistency. To resolve this issue you need to define a website map and group core areas. Start with defining common properties relevant to those areas, and establish a consistent format for the data.
Surfacing user information is one of the common examples.
“user_properties”: {
“user_loggedin”: true,
“user_id”:”22211134”,
“hashed_email_sha256”:””
}
Here is a quick list of questions you want to answer:
- What data needs to exists on all pages. User information is a common example.
- What areas have information sepecific only to them. Such places can be blog, checkout, etc.
- What is agreed structure to present each type of data. Define common names for events and context data about user and page content.
If you are facing an existing data structure that is not consistent, avoid rebuilding everything from scratch. Instead, pick a business vertical or area and start there. Keep backward compatibility until you transition all components to a new format. It might take a very long time, but it will reduce stress on the business and boost your chances of success.
Ensure this data exists on every page in a consistent manner. This will allow you to pass it as context to Analytics and Marketing tools and improve user ID recognition. This exercise has to be done with every major part of the website. For example, if you have a content or blog section, you need to define metadata that needs to be passed alongside events. The same goes to properties of the main funnel, checkout or user registration.
dataLayer can be used to store vital information about various user interactions. It's widely used across the web. Here is a quick example how Hackernoon.com uses DataLayer to store events of an ad view.
It's especially helpful with ecommerce and lead-gen business as it allows to structure and store your transaction data for analytics tools:
window.dataLayer = window.dataLayer || [];
window.dataLayer.push({
event: "purchase",
ecommerce: {
transaction_id: "1234",
value: 250,
currency: "GBP"
}
});
Privacy Gate: Consent Management Platform (CMP + IAB/TCF)
This article is informational and not legal advice. Always consult qualified counsel before processing personal data.
Consent Management has to be present in all regulated countries and it will require a lot of attention. Complying with consent regulations is crucial as otherwise, your business can face significant fines. Thankfully there are great tools on the market that make it easier. I will focus on EU/UK implementation as I’m mostly familiar with those.
First of all, you need to familiarise yourself with consent regulations in the regions you are present. Regulation in the EU can differ country by country and EU and UK regulations can differ as well. Another common mistake is people assume GDPR and cookie consent regulation are the same - they are not.
In a lot of countries, users have to give consent for the use of all other cookies that are not required for a core website functionality. Consent has to be explicitly given (you cannot assume it), and it has to be as easy to give consent as it is to withdraw it. Thankfully most Consent Management Platforms come with a lot of guidance and support on these measures.
When a user consents to cookies, CMP can build a consent string which later you can add to the dataLayer to manage the deployment of the tracking scripts.
Every script requires a different set of consent options. To simplify this process, at least in Europe, you need to familiarise yourself with the IAB Framework and TCF v2.
TCF contains a predefined list of vendors and the exact settings those vendors need for their tracking scripts. Using CMP which is integrated with TCF will save you hours in settings and management of consent.
Deploy Tag Management
With a working Consent Management Platform and chosen analytics tool, the next major step is Tag Management. Tag management allows you to bypass long Product integration of scripts in every part of the website and instead have a dedicated version-controlled environment.
For example, if a user makes a purchase, you want to log that data in your analytics tool, as well as send it to Google Ads, Facebook Ads, etc. To make sure you stay consistent, solutions such as Google Tag Manager become very handy.
My favourite tool on the market is Google Tag Manager, free and very easy to use. It's compatible with any analytics tool, so it's a great pick regardless of your stack.
With all the limitations on tracking and cookies, I find it very important to look at both online and offline solutions. Online Tag Manager orchestrates online tags running from a user’s browser. Server Side Tag Manager runs scripts from your cloud container. It adds to your cloud cost and infrastructure, however, it increases the volume of the events as it’s not constrained by the time the user stays active on the page or ad blockers.
Here is a view of Server-side event flow from Google Developers website:
Not every business needs a server-side Tag Manager. However it allows significantly more flexibility in data governance and control, and it can help you increase the number of events that are tracked, since they are running outside user browser.
Finally, consider the implementation of the first-party mode. It allows routing JS scripts running on your website, including Tag Manager, to go through your CDN and significantly reduces the chances for them to be blocked. The direct impact is likely to be an increase in the number of events tracked in your analytics tool.
It’s crucial to ensure that at least your analytics platform and Tag Manager are loaded as 1st party scripts.
Search for “Your Tool First Party Mode” or “Your Tool Proxy load” to learn how can it be set up for your analytics tool.
Build Identity Resolution
Identity and cross-device resolution is another very big problem when it comes to tracking events that you will need to tackle next. There is no simple way to solve it and it would depend heavily on the product that you are working on. It’s important your business considers that issue when designing a product.
One of the core pieces of puzzle is to ensure you have user_id sent to the analytics platform whenever a user is logged in. That also means you want to create flows that will lead users to stay signed in. Otherwise, the majority of the traffic will likely be marked as “new” users and it would be challenging to move towards more user-centric metrics such as LTV and ARPU over time.
Another big challenge is the definition of device_id. In a lot of cases, it’s worthwhile to create a persistent device_id on the platform side and pass it over to the analytics tool. First of all, it extends your options to connect more offline/backend data to online tracking. It also allows you to control the logic behind device_id and ensure persistence over time.
Finally, event matching based on user Identifier and persistent IDs is another crucial piece. The biggest advertising platforms already need persistent identifiers such as email to enhanced campaign performance. For example, both Google Ads Enhanced Conversions and Facebook CAPI heavily rely on those for advanced user matching.
To enable advanced bidding and ID matching, you can pass hashed PII into your Tag Manager—just ensure it aligns with local privacy regulations.
Implement Server-to-Server / Warehouse Streaming
Server-to-Server event streaming allows you to send data directly from your backend services bypassing the front end and all tracking limitations coming with it. It’s very helpful to build complete user journeys and map events on a single timeline. It's not a replacement to online and Server Side tagging, but another technique that you should take advantage of.
I find it crucial for the business that have very compelx funnels or long user engagements. It's also widely used in apps.
Some of the events, such as refunds or calls cannot be recorded online. To make sure you still have them in your analytics tool & your database you can stream those events through native analytics SDKs.
Usually, those events would lack session_id as it's always a context of the online browsing experience. However, you can attribute sessions manually based on the event data in your data warehouse. Even without attribution, it still can be handy to model user engagement status and build marketing audiences.
For example, users that have unsubscribed can be automatically excluded from your Ads via the following route:
CRM Tool <S-to-S> Google Analytics / Amplitude /… <S-to-S> Google Ads
Your Takeaways
Crawl → walk → run: start with clean events + consent. Invest in the correct setup of your analytical and tagging tools, as they will define your data capabilities. Use first-party mode to maximize tracking and build products around services that would encourage users to sign up.
Your culture of measurement beats any single tool. You can achieve great measurement with almost any tool on the market as long as you apply these principles consistently. Also, avoid measuring everything, instead focus on figuring out what to measure & recording data consistently.
Make yourself familiar with advanced identity matching in ad platforms such as Enhanced Conversions in Google Ads and CAPI in Facebook. Stay compliant utilising the CMP framework and make sure you review laws in all regions you operate in.
Final tip from me: once you have the core up and running, look into modelling missing events. Google Analytics and Google Ads, as well as lots of other tools, provide Advanced Consent Mode.
Good luck in setting things up and don't hesitate to ask quesitons.
