As Apple messes with attribution, what does growth marketing look like in 2021?

It’s said that you should measure what you value, and for founders, nothing is more valuable than growth. Growth provides revenues, venture capital, prestige and scale — ultimately driving the success of every business. Yet, measuring growth is complex and challenging — and it’s only getting tougher. Changes to attribution in iOS 14 and further refinements in iOS 15, plus other privacy-preserving initiatives in the industry, have forced growth marketers to rethink how they define their growth analytics engines.

On the Extra Crunch stage at TechCrunch Disrupt 2021, we convened a distinguished panel of growth experts: Jenifer Ho, vice president of marketing at Elation Health; Shoji Ueki, head of marketing and analytics at Point; and Nik Sharma, the owner of Sharma Brands, a notable DTC growth marketing consultancy and investment shop.

Help TechCrunch find the best growth marketers for startups.

Provide a recommendation in this quick survey and we’ll share the results with everybody.

We talked about the crisis around attribution and how startups should define their models, how to acquire the first customers in a business and when growth marketing should really start, and finally, how companies should select the right metrics for their business.

The fundamentals of growth marketing: Building the attribution model

Our conversation started with a deep dive into the world of attribution, which is a hot topic for growth marketers these days. Attribution is all about connecting marketing spend on channels like Facebook and Google to actual purchase actions made by a customer. If you display an Instagram ad for shovels, and that user then buys a shovel two days later, attribution models are designed to capture that activity.

Purchasing decisions can be quite involved for businesses and consumers, which is why defining attribution is so important for a startup’s success. When it comes to how to model it, Ueki emphasized one word to describe all aspects of attribution: Simplicity.

I think one of the main things I’ve learned from experiences with attribution is that it’s really important to start basic and simple and then try to evolve from there. I’ve definitely been in companies where we’ve made that mistake before.

The example I like to give is, if you think about attribution, it’s a vehicle. It’s really enticing to try to build that Ferrari of attribution systems, but a lot of times when you’re just starting out, all you really need is a skateboard, right? It’s something that’s simple, and it gets the job done.

One example from a few years ago at SeatGeek: We tried to build a media-mix model before we had something much more basic in place. We invested a lot of time and effort into that with our data science team, but after all of that, it was actually never used and never adopted.

At the end of the day, the ultimate goal of attribution is to measure incrementality, or how much incremental revenue is being generated by your marketing efforts. A complex media-mix model or multitouch attribution model is intended to do that, but it’s hard to do it right. For us, we had built one that actually didn’t serve that need of measuring your incrementality as much as possible. (Timestamp: 2:48)

Ho noted that for Elation Health, the key is connecting attribution to company goals.

The way I think about it is, a lot of our attribution ties back ultimately to sales or acquisition of new physicians coming on to use the Elation platform. [ … ] For us, when we actually see that person convert to a known user, that is when we really start to get a lot of strength in our tracking. (Timestamp: 1:24)

Sharma, who is much more focused on consumers, notes that measuring incrementality is also a way of dodging the new Apple privacy restrictions.

We’ve actually found that the best way, especially with iOS 14 and 15 (which we’re very not happy about, especially with those emails being hidden), the best way to do it is actually look at attribution from an incrementality testing standpoint. So understanding with holdouts [if you were to turn a channel off, what happens to that acquisition cost?] how incrementally good a channel is for us.

Outside of just giving us the data of what the platform says, the incrementality attribution platforms can actually help us to verify that and really understand, “Oh, you know, is the platform telling us something that makes them look good? Or is it telling us a real holistic number?” (Timestamp: 5:42)

Ho agreed with the focus on simplicity and argued that even simple models can provide a lot of value even quite late in a startup’s life.

I agree that in most cases, you do have to start simple. Jumping into a multitouch attribution model, even on the B2B side, can be quite complex.

For example, at Elation, we’re very simple right now, it’s a first-touch model. We realize it is crediting a lot to first touch and it’s first touch when the person becomes known to us, but it’s a starting point, right? And it gives us at least some directional data to understand if that person brings in other people into the buying committee. (Timestamp: 10:44)

I asked how Elation arrived on that model, and Ho explained.

There was actually no structure before. So one of the learnings I saw when I kind of got into the systems and looked was, I think there was a strong attempt at trying to do attribution, but there wasn’t really strong execution across it. When I first came in, there was just a lot of data that was showing one story that didn’t actually align with what was really happening.

So we decided, let’s start with first-touch attribution, it was faster for us to execute so we can start seeing, and now we’re having discussions as we go into 2022 — do we start to move more toward a first-and-last touch or multitouch attribution? There’s some tooling that we would need to put in place as well, if we didn’t move toward multi. (Timestamp: 12:45)

For Point, Ueki noted their approach was last touch, given that consumers can spend months or longer considering a mortgage before committing. So first touch made less sense in his specific context.

The conclusion is to lock in a very simple model, and perhaps incrementally add more complexity over time, but to never allow elaborate systems to overtake the basic logistics of understanding key moments in a customer’s lifecycle.


Getting your first customers

We then pivoted the conversation into how startups can seek out their first customers and when growth marketing starts to kick in for scale.

Ho emphasized that a lot of growth can be organic in the early days.

I actually think it has a lot to do with whether your product or service has product-market fit. That’s like the first thing I think of. And if you do have fit, the first handful or couple dozen customers will come in pretty organically, because you’re addressing an actual need. (Timestamp: 16:04)

Ueki said that it’s okay to hustle for early customers, but then you need to actually start quickly locking in the right areas to spend time on growth.

It’s really important to ensure that core metrics are in place before starting to try to scale. So whether it’s retention rate, repeat purchase rate, conversion rate — whatever makes sense for your business, are roughly in line with, say, industry averages. I think at that stage, when you’re just trying to get the first 10 or so customers, then you can do anything, like hustle types of things, anything that doesn’t scale. But once you’re past that point, and you’re ready to step on the gas pedal, I think it’s really important to focus on strategies that can scale — finding growth loops that can lead to exponential growth even without an exponential increase in effort associated with them. (Timestamp: 17:03)

Unsurprisingly, given what he said during attribution, Ueki doubled down on simplicity.

One of the things I’ve learned is to focus on as few channels as possible, especially at the beginning. I think that for most products, it’s really one or two channels that are going to drive the vast majority of growth. The more effort and time and research you can put in figuring out what those channels are, I think that’s really helpful. (Timestamp: 17:56)

Meanwhile, Sharma noted that it’s not sufficient to just get early customers — you have to figure out your brand’s message before that.

I would say for the consumer world, it’s definitely a little different. The one thing I’ve noticed — especially the last couple years as the drug of Facebook ads is slowing down — is that you really need to build the foundations of a brand before you decide to turn on these growth efforts. So many of the brands that I see launched today, essentially their plan for acquiring their first 50 or 100 or even 10,000 customers is just Facebook ads. And the problem with that is you are essentially now a product company. (Timestamp: 19:19)

Instead of building a relationship with a customer, the company has become a widget seller. And no one wants to be a widget seller.


Selecting the right metrics for your business

Growth marketers can choose from dozens of metrics to benchmark their efforts. But which metrics matter, and how do you get consensus within a company on what the OKRs are? This is a problem that can often get worse over time as a company scales and more people join the growth function and senior leadership teams.

Ho emphasized communication and transparency.

It actually starts with having a shared understanding of the metrics that matter to the business. So whether that be the board or the CEO or senior leadership, but having that shared understanding of what matters to the business. That’s where I always begin — get a really strong shared understanding of the metrics, and then start to think about what the metrics are that actually matter to the marketing team or the growth team.

For example, at Elation, we’re not solely focused on revenue; we’re more focused around bookings. So for some people, they hear that and they’ll say, “Interesting, so you care more about acquiring the customer, regardless of how much they’re paying?” And the answer is, “Yes.” That’s the focus right now. And trust me, you can argue like, “Okay, well, you know, that’s short-term thinking, what about later?” But our board and our CEO at this moment, that’s what we’re focused on, which is new logo acquisition and acquiring new physicians onto our platform. (Timestamp: 27:15)

Ueki agreed, but noted it was important to coordinate goals with finance.

I think that it’s also really important to be very, very explicitly aligned about not just what the metrics are, but what the trade-offs are across them as well.

I have an example from when I was at SeatGeek a few years ago. There was a particular month when marketing was in charge of our gross revenue number. We actually had a pretty good month, but we ended up burning through a lot more cash than we anticipated. So while marketing thought it was a pretty good month, the finance side thought it was a horrible month and an alarmingly bad month. So marketing and finance were working toward these sort of opposing goals.

So we moved to a new structure where we had this gross revenue goal that we’re trying to maximize as much as possible, but we set a cash burn constraint. (Timestamp: 29:21)

Sharma noted that the same battles materialize in consumer companies as well, and founders need to handle trade-offs effectively.

With CPG, performance and brand are often the two teams that are really fighting. While performance and brand fighting is usually a sign of growth, it also does things where you might spend less today on brand marketing, and move that money to performance to hit a revenue number on an Excel sheet. Now, let’s say you’ve spent these performance dollars, the brand dollars that you would have spent this month may have had a better effect on the business in six months, and so there are a lot of trade-offs.

I think the biggest advice and maybe more of a plead I would say to the finance teams is, please be realistic with a lot of the goals that are set. (Timestamp: 32:01)

You can read the entire transcript here.