Measurement and attribution are two of the most important components of digital marketing. Without measurement, you can’t determine how well your campaigns are performing or if they’re even working at all. And without attribution, you can’t know which channels are responsible for driving the most value to your business.

Measurement is the process of analyzing data and interpreting the results. It’s also used to determine if your marketing efforts are working and how effective they are.

For example, if you want to know if someone clicked on your Google AdWords ad, you’d need to set up a tracking pixel on your website so you can track any visitors who click on that ad back to your site and see if they convert into customers or not.

Attribution is what happens after measurement — assigning credit (or blame) for specific actions on your website or social media accounts based on the actions taken by visitors before those actions happened. For example, if you have a new customer who purchased your product after clicking on an ad for it on Facebook, then the attribution model would determine that Facebook was responsible for that sale.

Measurement and Attribution Post-iOS 14

After iOS 14, measurement and attribution have become significantly more difficult.

Imagine you own a sailboat with all of the state of the art navigation tools you need.

This dashboard tells you exactly where you’re going, where you are, how long it’s going to take to get there, etc.

It even has the ability to let you autopilot for perfect wind momentum.
Then, someone comes along with a sledgehammer and knocks out the entire system.
While people have been sailing for thousands of years, not having modern tools can increase the difficulty to to succeed.

Unless you’re Magellan or have a large ship with lot’s of resources/momentum, sailing used to be really hard!

We’re seeing a similar trend in the advertising world.

For many advertisers, not having the state of the art measurement tools we’re used to left them paralyzed.

More critically, smaller advertisers no longer have the proper wind in their sales to keep momentum going.

Unfortunately, the all in-one-silver bullet solution (or close to it) we had is gone and now we have to pull out a bunch of old school tools to triangulate success like a map, compass, quadrant, etc.

In sailing, there was (and still is) a solution to seeing which way the wind is blowing by sticking a wet finger up in the air and seeing which side dries faster to understand which way the wind is blowing.

Without adapting to this new normal, it can kind of feel like that with measuring ad performance.

Thankfully, measurement doesn’t have to be that bad even if we don’t have our ideal solution in place.

Unfortunately, the all in-one-silver bullet solution (or close to it) we had is gone and now we have to pull out a bunch of old school tools to triangulate success like a map, compass, quadrant, etc.

Attribution and Measurement Roadmap

Phase 1: Last-Click Attribution and Post-Purchase Surveys

Source: AppsFlyer

The challenge with this type of attribution is that it doesn’t consider other channels that played a role in converting the customer. For instance, if you had a customer who saw your ad on Facebook and then later visited your website before making a purchase, Facebook would receive 100% credit for driving the sale — even though they only contributed 50%.

The post-purchase survey is a good gut-check to helps you understand the accuracy of your last click model and what influenced them along the way.

Post-purchase survey are also a great way to gather data on why customers made their purchase decisions.

It’s best used when you have multiple products or services that are similar but not exactly alike (e.g., different prices). For example, one option might be more expensive than another but offer better features; understanding which one customer prefers will help you better understand which features drive sales and whether that justifies charging more for it.

Last click measurement and attribution alongside a post-purchase survey answers the following questions:

• What is the last touch point someone interacted with before converting?
• Do how people say they heard about us align with their last click to convert for our brand?

Recommended Last Click Attribution Tools

Measurement tools help you gauge the effectiveness of your campaigns. They also let you see how much traffic is coming from each channel and which channels drive conversions. The most common measurement tools include:

• Google Analytics (GA4): The latest version of Google Analytics allows partners to combine a web + app view of their data and understand their full-funnel customer journey. This is one of the most popular tools for measuring website performance. It provides detailed information about traffic sources, user behavior, and much more. The downside is that it’s not free – if you want to use the full functionality of GA4, you’ll need an enterprise plan or higher. You can also use a free version of GA4 but with limited data reporting capabilities.

Recommended Post-Purchase Survey Tools

• Iterate: Iterate is a suite of research tools that help you learn directly from your visitors, customers, and users. With Iterate you can easily send email surveys, share links to surveys, display surveys directly within your website, or in your mobile app product.
• KnoCommerce: KnoCommerce is a customer survey platform built for eCommerce brands on Shopify. They let you collect the customer insights you want, turn them into the marketing strategy you need, and take action that changes everything.
• Enquire: Enquire is another customer survey platform built for Shopify Merchants similar to KnoCommerce.

Estimated Technical Length to Implement

Last-Click Attribution takes under a week to implement and you may get it up and running in one day.

Post Purchase Survey takes under a week to implement with many companies implementing these solutions in a few days.

Phase 2: Multi-touch attribution (MTA) and Omni-Channel Attribution

Multi-touch attribution (MTA) is an extension of last-click attribution. MTA takes into account multiple touchpoints, or interactions with a brand across various channels and media, to determine the effectiveness of each channel.

Additionally, layering in omnichannel measurement allows these models to understand the relationship between your online media (web and app) and offline media.

You can use MTA with any media, including paid search and display ads, social media marketing, and even direct mail campaigns. By looking at the entire path to purchase and not just one interaction, you can better allocate your budget and measure the impact of your marketing efforts.

You can use MTA to answer the following questions:

• Am I over-crediting specific channels in a last-click model?
• Am I under crediting impression-driven channels in a last-click model?
• What is the path to purchase from a media touch point perspective?
• Do I have the optimal channel mix?
• Should I shift the budget from brand to DR?
• Should I shift the budget from web to app?

Types of MTA

Multi-touch attribution (MTA) can be divided into rule-based and data-driven.

Rule-Based Multi-Touch Attribution

Rule-based MTA is typically used by large enterprises with access to internal data sources such as CRM and eCommerce systems. It uses rules to attribute a conversion to a previous touch point. For example, if a customer converts after visiting your website and then later makes an in-store purchase, the rule would attribute the conversion partially to the website visit and the in-store purchase.

Rule-based systems work by assigning points based on predetermined rules—for example, if someone comes through Facebook Messenger, they get three points; if they come through Google search, they get five points; if they come through an email newsletter link in your blog post, they get 10 points. This system works well when you have limited data and need to make decisions quickly, but it’s not very accurate because it doesn’t consider all the available data.

Data-Driven MTA

Data-driven MTA is open source or self-hosted software that uses web analytics data to perform multi-touch attribution analysis. It uses statistical models trained using historical data from your marketing efforts and website interactions. The model then determines how much weight each interaction should receive regarding contribution towards the final conversion event.

Starting to look at the path to purchase instead of focusing on the last click will allow you to invest more on the channels that are driving sales vs the channels that are taking credit.

Google Analytics Data-Driven Attribution (GA4) is an excellent example of this type of model. It uses advanced algorithms to analyze your website traffic, sessions, and events to determine which actions are most likely responsible for each conversion. You can use GA4 as an alternative to “last touch” attribution models because it can help you identify potential sources of traffic that traditional last touch models would otherwise overlook.

Other data-driven tools include NorthBeam and RockerBox. These use machine learning to analyze all available data and determine what actions result in positive outcomes. For example, if we know that someone who came through a Facebook Messenger ad converted at a rate of 10%, we can infer that sending more Facebook Messenger ads will increase conversion rates. Data-driven tools work better than rule-based ones because they allow us to make more precise predictions about what will happen in the future based on past performance.

The advent of iOS 14.5 and the loss of cookies on the web have made this model harder to use.

This was the gold standard for measuring what people did on your website for a long time. It provided a great way to see where people were coming from, how they interacted with your site, how many people converted, and what happened after they converted.

However, as more people started using ad blockers and privacy tools like Ghostery and uBlock Origin, it became harder for companies to track these events. There are still ways to do it — ecommerce companies can use third-party integrations such as Google Analytics or Shopify — but it’s not as easy as it once was.

Recommended Multi-Touch Attribution Tools

• Google Analytics Data Driven Attribution (GA4): This is the latest version of Google analytics that allows partners to combine a web + app view of their data and understand their full funnel customer journey.
• RockerBox: A tool for measuring user behavior, tracking online marketing campaigns, and optimizing websites for conversions. Rockerbox also works with companies that want to track their offline sales.
• NorthBeam: NorthBeam provides both a web-based and app-based solution to help businesses measure their digital performance and ROI and improve their future strategies. NorthBeam helps with all aspects of digital marketing, including website design, search engine optimization, paid search advertising, social media optimization, and more.
• TripleWhale: TripleWhale is an attribution platform that helps eCommerce businesses grow their sales by connecting with customers across multiple channels, including email, social media, display advertising, and mobile advertising in one place. The platform lets you analyze every single touchpoint between your brand and its customers — so you can optimize every dollar spent on marketing initiatives (and beyond).

Expected Return on Investment

The expected return on investment (ROI) of MTA depends on several factors, including:

• How much you are spending on advertising and marketing
• The number of channels you are using
• How many touch points your customer has with your brand before they convert

Implementation Requirements

Rule-Based MTA:

• Defining rules and looking at multiple model types
• Ingesting data from online and offline sources in a way that will allow you to better understand the path to purchase
• Understanding that rule-based models are limited in scope compared to machine learning models, you should look at the path to purchase in multiple ways vs. with one key solution

Phase 3: Media Mix Modeling (MMM)

Media Mix Modeling (MMM) is a method for forecasting sales using customer data and objective information about the product or service, such as pricing and promotion. It is also known as media mix optimization, and it’s a useful tool for marketers because it shows them how various combinations of different marketing channels can affect sales.

MMM is based on the idea that customers respond to different types of advertising in different ways, and marketers should use this insight to maximize their return on investment in each channel. The best way to do this is by trying different combinations of media types until you find the one that gives you the highest return on investment (ROI).

Questions MMM can answer:

• Is my MTA model calibrated correctly?
• Do I Have The Optimal Channel Mix?
• Should I shift The budget from brand dollars to DR dollars?
• Should I shift the budget from web to app?
• Should I shift the budget from offline to online media?

Potential Tools

• Recast– Recast is a new self-serve MMM tool if you’re looking for something simple. It’s an easy-to-use tool that will give you everything you need without requiring technical knowledge or effort.
• Outpoint– Another self-serve MMM tool that offers several different products, including software and training materials that will help keep your campaigns running smoothly no matter how complex they get!
• Analytic Edge- This tool offers simple, user-friendly interfaces to more advanced features like automated optimization, which allows you to save time while still getting the same results.
• Analytic Partners- The tool offers a wide variety of services designed specifically for marketers and advertisers who want to ensure their campaigns are running smoothly. Their platform gives users access to robust analytics tools that will help them understand where their budget is going to better allocate it in the future.

Expected Return On Investment

Media Mix Modeling Tools can save you up to 64% in CPA on average! (Source: Facebook, “Unlock Business Growth with Incrementality Measurement,” 2018)

It’s easy to see why: Media Mix Modeling Tools are designed to help you find the right media channels for your campaign to deliver the most relevant message to the right audience at the right time.

The result? You get more bang for your buck, which means more sales and conversions.

Estimated Technical Length To Implement

The Estimated Technical Length to Implement (ETLI) for Media Mix Modeling Tools is between four and six weeks for SaaS platforms like Recast and Outpoint. For more consultative engagements, it can take up to a year.
Implementation Requirements

You must meet the following requirements to implement a media max modeling solution effectively:

• Aggregate offline and online media inputs
• Individual/log level user journeys piped into MMM software

Phase 4: Conversion Lift and Geo-Based Measurement

Geo-based measurement is where you measure sales based on location. For example, if your business sells products in New York City and San Francisco but not Chicago, you can use geo-based measurement to see how many customers from Chicago have purchased from you.

Questions lift and geo-based measurement can answer

• Are my MMM models and MTA models calibrated correctly?
• Is the media I’m running incremental to my business?
• Am I paying for users I would have acquired organically without spending on my remarketing campaign?

Tools

Conversion lift: Using conversion lift, you can see at a glance which advertisements are driving conversions and which aren’t. You can also see if your ads are bringing in customers who have already made a purchase or if they’re bringing in new customers from other channels.

Geo-based measurement: The geo-based measurement tool allows you to see how many people have been reached by your ads and where those people are located. If you want to see if your ads are attracting people from different cities or states, this tool lets you do that quickly and easily.

Expected Return on Investment

Conversion lift: The expected return on investment (ROI) for conversion lift is the amount of money you expect to receive back from your efforts in a given period. This is calculated by taking the cost of your campaign and dividing it by the number of conversions.

For example, let’s say that you run a campaign that costs $100 and generates 10 new customers. Your ROI would be $10 (10 customers at $100 each).
Geo-based measurement: The expected return on investment for geo-based measurement varies based on the type of business you are in and your goals. If you’re a retail store, it’s probably best to find ways to increase foot traffic and sales. If you’re a restaurant, it may be more important to have a happy staff that consistently turns customers away.

Estimated Technical Length to Implement

• Conversion lift: None
• Geo-based measurement: None

Implementation Requirements

• Conversion lift (on Facebook): Requires CRM uploads or server-to-server integrations like the conversions API

Attribution vs MMM vs Lift

If you’re still confused about the differences between attribution, MMM and lift, the chart below is a great comparison of how these differ.

Like the sailing analogy above, there’s no silver bullet for measurement and you need all of these tools to succeed in a post-iOS 14 world.

Other: A/B Testing for Channel Specific Attribution and Measurement

Outside of measuring across channels, there’s still a lot of measurement solutions available within the ad platforms themselves.

Split testing is the best example of how to measure within a platform.

Split tests are a great way to answer the following questions:

• Should I shift the budget from brand to DR?
• Should I shift the budget from web to app?
• Does tactic A work better than tactic B?
• Does creative A work better than creative B?

While split testing might be daunting, many platforms have these abilities built in natively to their platform to make this capability easy to implement.

Expected Return on Investment

Split testing ROI can vary depending on the test with some leading to sub-10% improvements to while their are instances of tests driving 100% improvements for in-platform efficiency!

Estimated Technical Length to Implement and Implementation Requirements

Because split-testing is an in-platform solution, there’s no requirement technical implementation.

Final Thoughts

Measurement helps you understand how well your marketing efforts are working. With measurement and attribution, you can figure out which channels are bringing in customers, what they’re doing before they buy, and whether or not they’re converting into sales.