Building Product Channel Fit
Like Product Market Fit, but easier. Part 3 of our Facebook Guide.
Welcome to the conversation. This is a newly launched discussion on business and product strategy driving parabolic growth. Please take a moment to provide feedback, share your thoughts, or pass along to your friends.
Product Channel Fit is how well your business aligns with its distribution and/or marketing channel(s).
Unless you own the distribution or marketing channel, you are fitting your product to the channel, not the other way around. For advertisers, this means there is a competitive advantage in how well you can adapt marketing and business strategy to the ad platform. As disruptive companies increasingly apply new monetization strategies to old products and innovate through new business models entirely, advertisers that plan to drive growth through Facebook acquisition must build for product channel fit.
Not sure if you’re building towards strong product channel fit? Ask.
Product Channel Fit on Facebook starts and ends with signal.
I ask myself the following two questions when auditing my clients:
Within the constraints of the platform, how well can this business express desired business outcomes?
Within the constraints of the platform, how well can this business express desired business value?
Expressing Desired Outcomes
Facebook’s conversion optimization allows advertisers to optimize for an outcome passed back via the Pixel, Facebook SDK, or Conversion API*. The events passed back by each data source are matched back to a user, and then attributed to ads that were served to that user. This feedback loop informs the ad platform that the ad was successful in driving one or more actions tracked by the advertiser. More importantly, this tells the system how effective the impressions served by a campaign are in driving results for the advertiser.
*At the time of writing this article, Conversion API server-side events are treated as Pixel events, and require deduplication. Anecdotally, optimization for server side events alone have generally not outperformed traditional Pixel events alone, and my recommendation is to pass events through both data sources, deduplicate or merge overlapping events, and optimize for the combined signal.
Lakers in 6, Business Outcome in 7
As of October 2020, Facebook’s ad optimization is limited to a 7 day post-click, 1 day post-view** window. In plain English this means Facebook can reliably optimize for outcomes that happen 7 day after clicking or 1 day after viewing an ad impression. Advertisers should look to:
Have the vast majority of desired outcomes (e.g. Purchase) occur within 7 days of ad impression. Yes, this biases towards businesses with shorter consideration periods and higher frequency of conversion.
Identify a strong proxy event with extremely high correlation (R2 > 95 PLEASE.) to whatever down funnel conversion occurs outside of the 7 day optimization window. A R2 > 95 does not mean that 95% of people who take the proxy event end up converting. We simply want extremely high correlation on the variance between the proxy and the end goal.
Build a pLTV model that is capable of passing back a real time spectrum of values for the likelihood a proxy conversion will lead to their desired outcome.
All of the above.
The long-term answer really is D since any mature business ends up bidding for value anyways. More later.
Anything that isn’t captured inside the 7 day optimization window is just lost signal. Remember this?
Total Value (or eCPM) = Advertiser Bid * Estimated Action Rate + User Value
If Estimated Action Rate is loosely defined as Conversions Matched / Impressions Delivered, any Conversion outside of the 7 day window reduces the Estimated Action Rate, increasing the Advertiser Bid required to maintain the same competitiveness in the auction.
**It’s worth noting that the implication of a 1 day post-view optimization window is that signal loss decay starts to set in very quickly if the majority of your attributed conversions are post-view. I would tinker with your attribution windows to understand what the last touch attribution reveals about your user behavior.
The Reg. Gate Dilemma
Someone smarter than me probably once said:
A sign up is a customer, an guest checkout is someone you’re going to lose to Amazon
Full disclosure, I made that up but it’s still true. Brands have tossed around the registration gate conversation for a long time. Yes, signing up is a pain in the ass and some users will bounce, especially at the beginning. Chances are you’re weeding out low intent, low LTV users anyway, and those who are willing to sign up have already taken the first step to qualifying themselves as valuable customers.
More importantly, those users are giving you their personal information. Personal information, that if coupled with their actions on site, can supercharge the machine learning powering your ad delivery.
Back to Estimated Action Rate, which is Conversions Matched divided by Impressions Delivered. Match rate is heavily dependent on advertisers passing back parameters that can be used to map an event back to a user on Facebook’s identity graph. These Advanced Matching parameters will become increasingly important as legislation and browsers look to increase privacy and limit tracking through cookies, fingerprinting, etc.
Collecting these parameters is important, but so is the point in the customer journey that these parameters are collected. The earlier the data is provided by the users (*cough* registration gates *cough*), the earlier the data can be associated with subsequent user actions for matching. This unlocks a number of opportunities for advertisers who want to more effectively optimize for upper and mid-funnel events or to populate retargeting audiences on what would normally be anonymous user events pre-purchase.
Expressing Value
Facebook’s ads metrics, which flow into Ads Reporting and Ads Manager surfaces, are based on last touch attribution, with click being dominant. These reporting surfaces also don’t take into account any other marketing channels, which is often criticized since it almost guarantees that Facebook is over-crediting itself unless an advertiser only invests in Facebook for acquisition.
A better way of using Ads Manager performance numbers is as an indication of relative optimization performance, or how efficiently the ad delivery system is able to drive outcomes based on the inputs and signals you provide it.
Value != ROAS
Most, if not all, businesses eventually get to a conversation about bidding for value, or acquiring higher quality, higher LTV users. While Facebook has Value Optimization as an easy to use, easy to understand product for many advertisers, it’s important to understand the relationship between ROAS and Value, and the tradeoffs an advertiser makes when optimizing for Value.
In Ads Manager, ROAS is simply Conversion Value / Amount Spent.
From an optimization perspective, Value Optimization can be conceptualized as an added layer of machine learning that attempts to predict not only the future Conversion Rate, but the expected range of Conversion Values from those future conversions.
Therefore, ROAS bidding becomes limited by how closely AOV aligns with actual business value, and depending on your exclusion set up, it can either be limited to initial AOV (e.g. All Time Purchaser exclusions) or to AOV over the 7 Day optimization window (no exclusions, maximize rolling 7D Conversion Value for $ Spent).
Businesses that have LTV calculations with low correlation to AOV, or are dependent on AOV and/or other metrics outside of the 7 Day optimization window should proceed with caution.
Value Optimization comes with the price tag of being ineligible for post-view optimization, which is one of the larger optimization unlocks for most businesses.
Bidding for Value
Armed with an understanding of how Value Optimization works, advertisers have two basic options when bidding for Value.
Value Optimization - Most appropriate when initial AOV or total Purchase value within 7D is (or is most causal) to user LTV.
Conversion Optimization - Generally most appropriate when A) does not apply. Advertisers should look to express Value by mapping relative bids to subsets of signal, either through Signal Segmentation or Bid Multipliers.
Bottom Line on Product Channel Fit
Businesses must be intentional about the channel they use to acquire customers, whether it’s a paid channel like Facebook, through organic virality, content marketing like YouTube, or others. Consumer businesses can achieve insane growth by mastering a single channel alone (e.g. Wish + Facebook or Kylie Cosmetics + Virality). If the primary growth channel is Facebook:
The bad news: most advertisers have very little influence on molding the channel to their business.
The good news: the mechanisms of the channel are very clearly defined. Learn the rules of the game and you can optimize your business to have the greatest chance of success.