Skip to Content

Smarter Data = Better Outcomes

August 7, 2023

Over the past 12-18 months, we’ve seen platforms like Google and Meta progressively push advertisers to integrate “enhanced” and “offline” conversions into their ad accounts, all with the promise of better measurement and more accurate data. 

After reviewing dozens of ad accounts (with and without enhanced/offline conversions) I firmly believe this is one of the most impactful, high-leverage actions marketers can take to both future-proof their ad accounts over the long-run AND improve their account performance in the short-run. In fact, I’d go one step further: maintaining a high-performing ad account without enhanced / offline conversions is a never-ending high-wire act, which – sooner or later – will end in disaster.

Before we dive into the “why” behind that bold statement, I want to answer a few common questions I’ve heard from brands and fellow advertisers about enhanced conversions: 

What are enhanced & offline conversions? These are two separate features, each of which work slightly differently, but toward the same end: 

  • Enhanced Conversions (EC) – ECs are a privacy-friendly feature that supplements existing tag-based conversion tracking with hashed zero-party customer data, in an effort to improve matching and attribution. When a user submits a form or takes an action on your site involving zero-party data (name, email, phone, address, etc.), enhanced conversions hash that data using SHA256 (a secure one-way algorithm), then send it to Google. Google matches the hashed data against their database of first-party data, and then sends it back to Google ads using the same secure, anonymous form. This has the practical benefit of (a) ensuring that you’re accurately counting real conversions, not accidental loads or erroneous events and (b) giving the ad platform access to incremental data based on the matched user’s profile.
  • Offline Conversions (OC) – related to Enhanced Conversions, offline conversions provide a secure method for advertisers to share incremental data about conversions with Google. This could include lead qualification (i.e. MQL/SQL/Closed-Won), actual sales data (net of returns + upsells), etc. Where Enhanced Conversions increase the data available to match the conversion to a user, Offline Conversions supplement existing, website/app-based conversion data with your business data (i.e. customer qualification, total customer value, etc.). 
  • Meta Ads (Facebook) has unified these two features via their Conversions API (“CAPI”), which directly sends customer data to Meta, while enabling advertisers to configure offline conversion imports.

Do I need just enhanced conversions, or should I use offline conversions, too? EC + OC work better together – EC provides incremental platform/engine data, along with better matching of conversion events to people; OC supplements your lead data with relevant, smart business data – such as lead qualification stage, customer value, sales status (booked demo / attended demo / etc.). 

Should I be worried about the platforms stealing my data? Honestly, no. If Meta or Google wanted to destroy your business, they (candidly) have the resources to do it outright; they don’t need to dramatically expand their risk profile by circumventing their own Terms of Service. 

My advice is this: provide the platforms with as much data as you can sustainably + reliably generate and maintain, within their attribution windows (typically 30 to 90 days). If the best you can consistently + sustainably generate is a MQL, then use that. 

Is it better to stack conversion actions, or just bid on the one I want? 

My general rule is this: always bid for the thing you want. Don’t pollute your dataset. Once you are using offline conversions, you should switch your primary optimization action to the one closest to business impact (for instance, MQL is closer to business impact than website lead; SQL is closer to business impact than MQL) with the velocity necessary to power a ML algorithm (typically 30+ actions per month for Google, and 50 per week for Meta). 

Finally, The “Why” Behind Enhanced + Offline Conversions: 

Smarter Data Wins

It’s no secret that machine learning (ML) and artificial intelligence (AI) are transforming how digital advertising platforms work, as well as how we (marketers) work in them. In-platform automation will only increase as “smart” bidding goes from an option to “the” option, and progressively more of the traditional targeting levers (match types, audience data, etc.) is removed. Both Google and Meta have telegraphed this pending revolution for 2+ years.

In-platform automation works by optimizing toward a specific target within the constraints set by the advertiser (for example: a target CPA, a bid cap, or an audience exclusion). However, unlike a human, automation doesn’t have a broader context to consider when making bidding decisions within your account; it assumes the data it has (either platform data, engine data, user data, business data, or some combination thereof) is sufficient, complete, and accurate. Unfortunately, those assumptions are rarely, if ever, true.

This means that the optimization levers marketers have moves from the in-platform level (like match types, granular audience data, etc.) to the data level. If you, as a marketer/business owner, can provide your ad platforms with better, more complete, or more accurate data (which is what EC + OC do), you gain an advantage over advertisers who do not. 

In short: the advertiser with the best data will win over the long-term. The better your ability to pass relevant data to platforms, the higher the probability is that you’ll be the one with the best data. That isn’t simply more conversions or more accurate reporting, but faster model training and more relevant audience targeting.

ML-Driven Ad Targeting Is Sensitive To Initial Conditions

There’s a principle in chaos theory (yeah, we’re going there on a Sunday morning) that is highly relevant to ML-driven digital advertising: sensitive dependence on initial conditions. 

To distill a lot of science into a single sentence: miniscule changes in the initial state of a system can cause the behavior of that system to diverge rapidly, because those changes are persistently magnified through the dynamics of the system. 

Practically, this explains why “spam” leads or low-value customers tend to cascade into accounts, especially those without EC/OC: the automation doesn’t “see” that the leads are being filled by people trying to sell you $19 Jordan jerseys or Nigerian princes – it just “sees” that a certain sub-segment of eligible universe (the “initial condition” for automation) tends to complete forms at a high rate, and that results in a rapid divergence: 

The same is true for negative/low-value customers: unless told otherwise, the platform algorithms don’t know that a given customer actually returned their product, or purchased a low-margin item, or was an actual net loss to your business; all the system sees is a customer who made a purchase. Once again, shifting the initial conditions results in targeting rapidly diverging from where you might want it to go (for instance, in the graphic above, from the red (high-value customers) to the yellow (trial kits, which are often loss-leaders). 

By supplementing your website/app data with business data (upsells/returns/profit margin/lead qualification), you are effectively creating your own insurance against happening to you, while also re-orienting the system in a way that favors you long term.

OK, I’m convinced. How do I set up Enhanced + Offline Conversions?

It’s honestly very simple! If none of this convinced you, let me close with one point that perhaps will: enhanced + offline conversions are here to stay, and will only become more prevalent as ad platforms build out more robust integrations with more offline systems. We’ve already seen Google Ads expand support from Salesforce to more than 1,000 other platforms via their Zapier connection. Meta Ads has rolled out an incredibly easy to install system for CAPI. 

That means more of your competitors will be using enhanced + offline conversions – and if you’re not, you will fall behind. And once you fall behind, the cold, hard math of machine learning makes it exponentially harder to catch up. 

As Jeremy Irons says in Margin Call (one of my absolute favorite movies), “There are three ways to win in this business: be first, be smarter, or cheat. And I don’t cheat. And while I think we’ve got some pretty smart people here, it sure is a hell of a lot easier to just be first.” 

So, if you haven’t built out your enhanced conversions + offline conversions data flows just yet, there’s no better time to start.