Instagram and the interest graph
The killer app and the app killer
Of course, the free and open web is prone to beg, borrow, and steal your attention. Reading blogs, news, and research has always been an inefficient user experience — like finding needles in haystacks — but now Annotote is the antidote, so check it out: Don’t waste time or attention; get straight to the point.
Mark Zuckerberg had a lot to say about the evolution of Instagram on Facebook’s most recent earnings call last month:
[T]here is still a lot to improve in Feed, but we’re increasingly focused on other experiences as well. But on Instagram, instead of focusing on communities, we’re very focused on helping you explore your interests.
—Highlights saved with Annotote
Therein, Facebook’s cofounder and CEO is drawing a bright line between social graphs and interest graphs. That says a lot about his future strategy for monetizing Instagram — and the app’s value proposition relative to the rest of Facebook’s portfolio. Stratechery’s Ben Thompson nailed the upshot:
This gets at why Instagram is so valuable: as I’ve noted in the context of Twitter, the interest graph is even more valuable than the social graph, but it is harder to build. Instagram, though, has done exactly that and, at the same time, unquestionably been aided by being able to piggy-back off of Facebook’s social graph.
Perfect. He’s right: Instagram has certainly found its own, unique product/market fit in the valuable, but indefatigable, interest graph. But, hold on a second, Instagram’s execution of its strategic plan is a separate peace — with its outward growth starting to defy the magical alchemy that provided it this defensible moat in the first place…
Social graphs vs interest graphs
There’s something really interesting about the distinction between social and interest graphs, especially within context of Instagram’s taxonomy therein. Part of why a killer app for the interest graph has been so elusive is because of how easy the open web’s many layers of abstraction make it to game-the-system. Juxtapose that with Instagram’s wholly owned and operated walled-garden, which engenders higher-efficacy signals about user interests thanks to first party data and quality control.
With a social graph, you can derive a lot of signal from just looking at users’ first party interactions right there on the platform. Intention is as pure of a signal as you can get, and digital communications are highly motivated by this intent, since comms are both the means and the ends. Compiling a social graph is more like algebra than probability — wherein known quantities can be used to solve for most of the unknown variables.
In contrast, you’re dependent on far more approximation when it comes to an interest graph. It’s more probability than algebra — rife with unintentional engagements and highly subjective indicators (e.g. semantic language instead of literal interconnections) that create a shaky co-dependence among many unknown variables. Whether a URL itself, a destination webpage/article, or metadata, all of a traditional interest graph’s inputs are drawn-from extraneous sources, and are therefore far more prone to false positives than the explicit intent of interpersonal connections.
Semantic/NLP technical challenges aside, the accuracy of these predictive models suffers from the fact that Twitter, Google, and Facebook have to scrape much of their “interest” data from off-platform properties — like those aforementioned URLs, 3rd party webpages, and metadata. Despite Herculean efforts toward quality control, things like SEO and clickbait can throw scalable algorithms off-the-scent of pure signal. (Which is not to write-off how impressive the companies’ algorithms really are — or how far their data collection tentacles really reach — but interests are just far more subjective than mere contacts!)
As of 2018, the interest graph seems to have been more of an app killer than a killer app.
The URL as a gateway drug
Enter Instagram. Engagement on IG occurs almost entirely on-platform and generates almost all first-party data. Until recently, it wasn’t even possible/practicable to append URLs to Instagram’s user-generated content — beyond the obligatory “link in profile” marching orders. To the extent that clickable links are even functional in today’s Instagram (i.e. “navigable URL links” are now available, but only in profile bios and Stories), user engagement is still almost entirely predicated by the on-platform content in their feeds and stories. And, to preserve the sanctity of those spaces, Instagram only populates the main feed with content from people you follow directly, plus it doesn’t let users repost into their friends’ feeds. All told, this adds a lot of friction to the spread of corrosive content.
As a corollary, users generally “like” posts and follow others based upon explicit intent: Either the user is part of your social graph or their content is part of your interest graph; you don’t get fooled into clicking-through and reading a misinformed/misrepresented/clickbaited article.
Of course, outbound links aren’t ruining Instagram, nor was everything perfect before their arrival. Instagram itself still faces aforementioned technical challenges in translating data from visual content into interests. Relatedly, algorithmic curation in IG’s “Explore” tab is another work-in-progress. Plus, users themselves have always had latitudes for gaming-the-system via comments and hashtags. However, without clickable URLs, bad actors had far less incentive for malfeasance, because their reward was merely social currency, which is certainly valuable, but not as directly monetizable as clickbaiting an audience, en masse, to feed a CPM-based shark.
This isn’t an indictment of ad-supported media or the free and open web; nor is this an analysis of fake news and misinformation online. I’m merely pointing-out how difficult of an equation the interest graph is to solve — a balancing-act for which the incentives and misincentives of ad-based monetization are part-in-parcel.
Instagram’s jobs to be done
That all said, over and above commercial clickbaiting, one of social currency’s most fruitful applications is influencing political sentiment. So, how can Instagram protect its interest graph from being gamed toward political ends?
Its initiative starts with jobs-to-be-done. From “The Future or Reading, Listening, and Watching”:
Anthony Downs classified the types of knowledge we consume as follows:
“[There’s a difference between] entertainment information — stuff you read because you enjoy it — to production information — stuff you read because you think it’ll help you make money. [There are only] four types of information…
1. Production information helps you make smarter business decisions; if you’re a stockbroker, The Wall Street Journal is production information for you.
2. Consumption information makes you a better consumer; if you’re going to a movie this weekend, Rotten Tomatoes is consumption information for you.
3. Entertainment information is … anything you consume primarily to be entertained, whether high culture (a great novel) or low (a Kardashian).
4. Political information is anything that makes you a more informed voter [which is] the toughest one to sell…”
Much like the original Facebook concept, Instagram traffics in #2 and #3 and almost entirely eludes #1 and #4. To be sure, IG has nothing structurally prohibiting information classified as production and politics — nor anything structurally assuring that its users stick with the consumption and entertainment genres. However, photos do lend rather naturally to consumer and entertainment info; and photos lend rather poorly to productivity and politics — especially relative to other mediums/platforms that are better optimized for these jobs-to-be-done. (For example, interest graphs serving production information to knowledge workers.)
To wit, this reminds me of the bundling/unbundling underway in cable television, as described by Ben Thompson in “The Jobs TV Does”:
It’s in jobs-to-be-done, however, where the unbundling [of TV] that matters is happening… two of the jobs TV has traditionally done are now done far better, and far more cheaply, by personal devices like computers, tablets, and phones. That is disruption…
It’s attention that is key; our attention is a zero sum resource — every minute I spend playing a game, for example, is a minute I don’t spend watching TV. And, if any company “cracks” TV, it’s not that they’ve figured out how to do TV better, but that they’ve figured out how to win a greater and greater share of consumer’s attention by doing the same jobs that TV does, but better.
While TV’s supply/demand mechanics are much different than user-generated content’s, Ben’s idea of the pendulum swinging from one extreme to another (i.e. a new medium unbundles an integrated incumbent then eventually rebundles the disintegrated landscape that results) is manifest in Facebook’s own trials and tribulations. Facebook Core was originally unassailable for entertainment such as social networking, but given their marketshare of attention, they almost inertially expanded to subsume other verticals like news. In fact, I think you can trace the seeds of this imperialist expansion strategy back to when FB first allowed users to add clickable hyperlinks to posts. (Was it 2010?)
In sum, Instagram can start to inhibit the propagation of undesirable information on its platform by re-focusing on and delving deeper into its entertainment niche — for which clickable links serve no purpose. URLs are not a necessary component of an IG post or story; they don’t improve the app’s UX; and whatever jobs-to-be-done that they actually address are better served by competitors.
Intention is nine-tenths of the law
Again, the key is intent. Sure, few people open Facebook with the intent of getting informed — or misinformed for that matter. But, FB has become synonymous with the internet for many of its users worldwide — in fact, it has become the internet for most users — thus undertaking the web’s messy battle with (mis)information comes with the territory. This fits an observation I made in “Facebook, Cambridge Analytica, and Other Weakest Links”:
[Facebook] assumed the difficult — but necessary — task… an impossible undertaking that nobody could perform flawlessly. Stuff is going to fall through the cracks, and [Facebook’s] going to get blamed for it. Such is the plight of the supermassive content aggregator, although the spoils are immense.
Zuckerberg never planned on Facebook assuming its current role as a social supermarket. He always wanted (and still wants?) Facebook to be a social network instead of social media — to connect people, not inform them. Competitive dynamics and, frankly, users’ needs led Facebook to integrate so much functionality that it grew into the Swiss Army Knife of a social hub that it is today. (That point of view gives Facebook the benefit-of-the-doubt for “lowering the barrier-to-entry for new users and Pages”.)
But, once upon a time, on a quarterly earnings call back in 2015, Zuckerberg actually talked about the need to segregate user-generated content from organic brand interactions in order to preserve the app’s singular focus on person-to-person comms:
That other [monetization] opportunity is Instagram, Messenger, and WhatsApp… But it’s really important to get this right and not rush it…
[Messenger today is] about where Facebook was in around 2006 or 2007 [when it was] really just a consumer product. There were no businesses in the ecosystem. And a lot of people were telling us go put better ads in. And that felt wrong. I didn’t think that that was going to be the right way to build the product or build the business.
So, instead, what we did was we built Pages, which was a way for businesses to interact for free in the system and start creating organic interactions between people and businesses so we could figure out what the people using Facebook wanted from businesses within Facebook. And we built more tools for Pages and businesses to engage.
Our recent success with advertising is really just built on some of those organic interactions… we’re still at the early end of that curve where the interaction is still primarily people to people, and businesses are starting to figure out… what the organic interaction is. But we’re going to have to go through a whole cycle of figuring out how that works before it really makes sense to start monetizing them…
—Highlights saved with Annotote
In other words, his design was to ring-fence brands and businesses on Pages, where consumers could find them if they were looking for them, but leave consumers to interact with one another in the News Feed. (It’s almost like an inversion of B2C, as Pages were originally conceived as a C2B destinations for reverse inquiry and News Feed strictly C2C comms.) Perhaps too idealistic to foresee the problems of misinformation and false positives, Zuckerberg was intent on separating users and brands as a solution to a UX problem.
Obviously, Facebook veered off that path in the years since those comments, as commercial interests have figured out “what the organic interaction is” to break out of their segregated Pages and into the News Feed. But, since Facebook has assumed that “difficult but necessary” role, that doesn’t have to be Instagram’s lot. Few people open Instagram with the intent of getting informed; people open Instagram with the intent of getting entertained; and Instagram should predominantly aim to entertain them.
Do your job
Bringing this full-circle, Instagram has a chance to scale-up a really good interest graph for a really good niche. That niche also monetizes really well for bottom-of-the-funnel conversions. In trying to fulfill that promise, IG doesn’t need to encumber itself with all the baggage that comes with competitors’ territories — be they Google, Twitter, or even Facebook Core.
Sure, unpaid organic content on Instagram can accrue social currency for indirect monetization, but IG could at least make commercial marketers pay for the privilege of directly monetizing that. Since direct conversions require outbound links (or native buy buttons), IG could make advertisers and influencers pay for those capabilities. That way, commercial interests would be taxed for the rent they extract from a user’s interest graph (i.e. ad impressions without conversions) and rewarded for the value they add therein (i.e. conversions).
That’s different than other platforms and aggregators who operate in the direct response and targeted advertising space, because it would disincentivize the noise of unpaid organic brand engagement — pushing that onus/opportunity back on to Facebook Core, et al. Were they to explicitly signal such intent, users could still engage with businesses and brands’ organic content, but eliminating clickable URLs in everyone’s organic content re-emphasizes that explicit intent, which would be derived directly from users’ observable interactions (algebra) rather than approximated from estimates of estimates (probability).
In that vein, back on Facebook’s quarterly earnings call last week, Sheryl Sandberg answered a question about the long-term revenue potential of Instagram vs Facebook Core. She frames not only the breadth of Instagram’s opportunity, but also the depth of its niche:
When you look at the Instagram Shopping experience, we’re seeing some really nice growth. We have 90 million people tap to reveal product tags and posts every month to learn more about them, and we’re putting real investments behind this. In Q3, we rolled out Shopping in Stories globally and began testing the Shopping Channel in Explore. And so, we think the opportunities are big.
As you think about commercial intent in Facebook versus Instagram, there’s so much activity on both. We think there’s a lot of opportunity for people to have commercial intent, if not have it when they start, but develop it because they see things they’re interested in, in both. Instagram can be more interest based in some places than Facebook, so there are places in Instagram like Fashion or like Shopping that have very high signal, and that gives us I think a very strong opportunity there.
—Highlights saved with Annotote
She tagged all of the themes I’ve discussed so far — interests, intent, jobs-to-be-done (specifically consumption and entertainment info), organic vs paid, signal vs noise, etc. Yet, here again is the problem that I opened with: Instagram isn’t implementing the same strategically-sound plan it keeps talking about. Instead, it’s creeping ever closer toward assimilation, becoming a redundancy of Facebook Core. The longer Instagram leaves itself prone to production and political information, the more it vacates its own defensible moat.
I don’t want to overindex to clickable URLs in isolation. Not to eschew Occam’s Razor or The Butterfly Effect, but it’s too neat and easy to blame such big ramifications on such a small feature. Plus, given the narrow scope of that feature’s rollout to date, nobody can say that Instagram has irreparably lost its way. It seems more like they’re just experimenting. It’s also a truly impossible task to thwart motivated malevolents from using an open platform for ideological means — and the more attention Instagram aggregates, the more of a target it becomes.
But, the point is that “navigable URL links” are a fruitless trial, because it doesn’t fit Instagram’s vision or value proposition as the first killer app for the interest graph. Retreating from that frontier back to the safety of Facebook Core’s fat-and-happy social graph will leave the interest graph opportunity wide-open for new entrants who have less baggage to lose in blazing new, adventurous trails — like TikTok, the most recent forager to lean into its interest graph (“For You” algorithm) at the expense of its social features.
Were Instagram to monetize advertisers and influencers by charging them for the right to directly convert their audiences via outbound URLs, buy buttons, etc — as discussed herein — they would still face the formidable challenge of verifying these users, paying them, and moderating their content both on and off platform. Some of these challenges, like moderation, are intractable; yet, others, like verification and payments, are suited for nobody better than Instagram/Facebook.
Know your role and get back on the strategy track.
Since we’re talking about your interests…
