Sandi MacPherson, Editor-in-Chief at Quibb

Read what else Sandi MacPherson is reading for work

Quibb lets you share what you're reading for work. Use Quibb to share news about your industry, discuss what matters, and see what colleagues are reading.

Our mission is to connect professionals over business news and informed commentary — targeting every industry, profession, and geography.

Likes Comments
Managing Director at DFJ

Sandi MacPherson, thanks for putting this together! One interesting theme to come out of the discussion above is that Facebook Messenger is the only obvious success of unbundling we can collectively cite at this point when using installs / engagement as the key metrics. The asian messenger apps usually point to third party apps afaik so that eliminates a broad swath. The elephant in the room, besides measuring success via different metric than installs, is Google. They have an app constellation across Maps, Chrome & Gmail but I don't know how to measure success of it given all where large established properties prior to secular shift to mobile. On iOS, they all work "better together" because of default deep linking which in theory should also be driving higher engagement. Would love to hear smart folks opine on this as I don't have any good insights.

Product Manager at InsightSquared

Interesting that Instagram is not mentioned w/r/t Facebook's constellation. Whether the KPI is installs, MAUs, or mindshare, it's a huge part of their mobile strategy. Facebook's constellation also points to a kind of decoupling of their brand too. It's not Facebook Instagram, Facebook Paper, and Facebook Slingshot.

Reply · Tweet ·
Principal, Business Development, Games at Amazon

I uninstalled 4sq for the first time. See how frustrating it is

Reply · Tweet ·
Quibb, Uber

I kept Foursquare but uninstalled Swarm, and that seemed to downgrade my UI within Foursquare to the "classic" version which I prefer. But yeah, tricky problem for them.

Growth Lead at Pinterest

Ooh, good idea. I hate when I'm transitioned to Swarm.

4sq and yelp are not battling on my screen, but 4sq and swarm are. Users are bound to favor one star in the constellation.

Product Manager at InsightSquared

Swarm is on my home screen. Foursquare is in the travel folder. I'm more likely on any given day to check in someplace than look up tips for where to eat. Also re: Swarm, I really like the concept of the location-based planning. It would have felt like feature bloat in pre-split Foursquare, but makes sense in standalone Swarm. Haven't seen much usage in my social graph in downtown Boston yet. Anyone else tried using this feature?

Reply · Tweet ·

A couple examples where app constellations have worked well: 1) Kakao, the maker behind KakaoTalk (150M users) has successfully launched about a dozen additional apps, many of which number in the millions of users within just days. For them, having KakaoTalk as a platform has made distribution simple, and made it easy to introduce new apps and features which can find success (or failure) on their own without disrupting the experience of KakaoTalk. 2) RunKeeper launched Breeze earlier this year as a separate step-tracking app which has done well and is important to keep separate from RunKeeper since it targets a rather different demographic (applicable to anyone who wants to get just a bit more active, vs who want to analyze their workouts), behavior (always-on tracking with built in motivation, vs self-motivated working out) and device (it's only available for iPhone 5s). Way better to issue this as a separate product, especially when it involves a higher degree of data collection and analysis. The choice is up to the user.


Thanks for this, as it's very illuminating insights on the challenges app makers are facing. Would love to see how Google's constellations of apps are doing. It still points to a broken app discovery system I believe or maybe, just maybe, people are pretty happy with the apps they already have and you have to be truly great to be added to their phone.

Reply · Tweet ·, Twitter

I'd argue that the constellations strategy has been successful:
All the "failed" constellations are probably failures (so far) because the app is still figuring out product-market fit. Had the feature been in the main app (Carousel in Dropbox, or Paper in Facebook, for example), it would have made the main app more confusing, less compelling and hurt the core business. (No data to back this; just a hypothesis)

As others have said, these offshoot apps are experiments. We shouldn't consider the failure of an experiment to be the failure of the experimentation strategy.

CEO and Co-Founder at Fetchnotes

I think some of these could rightfully (albeit still speculatively) be called experiments (Paper for sure, unclear about Carousel), but others are definite unbundling of their product into separate core apps. When a main button in one app takes you to the other (like Foursquare-->Swarm and Facebook-->Messenger), it's not really experimenting anymore.

Reply · Tweet ·, Twitter

Agree: some of it is unbundling rather than testing a new product. But unbundling makes way for further experimentation.

FB Messenger (the app) has been successful likely because it was already successful as a product, inside the FB app and on the web before.

Re: Foursquare / Swarm: Their strategy confuses me as much as their apps. It seems more than unbundling - almost like a tear-down of their existing story into a new one.

It seems like they're not working cause the apps aren't worthwhile.

Releasing a bunch of half baked or even bad products and then calling it an unbundled constellation- a mobile distribution and engagement strategy is just lots of fancy words that have no bearing on the fact that you've released shitty apps. The FB / Dropbox / etc brand and marketing machine can get you some installs but at the end of the day that's not building a business- that driving some downloads.

Sounds like Evernote is the only company doing it right across the entire constellation (don't know this firsthand mind you), maybe they're the only ones who have genuinely useful apps across the board?

Reply · Tweet ·
Co-Founder at iobeam

I was thinking the same thing. Building/maintaing one valuable app is hard enough, let alone multiple apps. And most of these constellation apps just aren't that useful.

Yes, you can drive downloads from your core app to constellation apps, but that doesn't promise engagement - which is likely why so many of these charts start strong and then trend down.

That said, sometimes you need to try something crazy new without disrupting your core business. Constellation apps are great for that. But we should expect most of them to fail.

Maybe instead of "Constellation Apps" we should start calling them "Pilots" or "Spinoffs", like new TV shows. Don't let the parent company's name fool you. These apps are mostly experiments.

Spinoff seems tonally correct. Don't want to ruin Friends with Joey, but still wanna know if you can get that solo act money.

Reply · Tweet ·
PM, Growth at SurveyMonkey

MapMyFitness are on to this by separating different mapping activities with a cluster of MapMy________ apps. This is a slightly different use case for a constellation - rather than unbundling functionality within an app they're offering a largely similar app in different flavors, each designed to appeal to a specific target market.

Reply · Tweet ·
Co-Founder at Beautylish

The app engagement chart for Swarm doesn't tell the whole story. It's a shift in strategy for Foursquare and both apps don't have to be successful for it to work.

It's hard to kill products. Users complain. People inside the company have a hard time saying goodbye to their baby.

It would be worse to try to combine both apps into one and do both poorly. This lets Foursquare keep all their existing users and try a new strategy while not explicitly killing the old functionality.

I don't think there's any definitive proof in the market that app constellations don't work. Too early in the game. Broader definitions of app constellations have worked fine in the past (networks of games, utilities etc.). My view on unbundling/app networks as applied to emerging markets are here: .

Reply · Tweet ·

Perhaps this is an obvious point, but an app constellation strategy is not one-size-fits-all. What works for one app will most likely not work for another, given the different use case and context. So everyone in this forum is correct, as long as they don't acknowledge only one side of the coin.

A similar parallel can be drawn around last year's trend of making apps free (in order to minimize the friction of signup) and then trying to make money off in-app purchases. While this worked for some companies (esp in gaming, where people are willing to spend money deeper into the game experience), it hurt premium brands like Nickelodeon who implicitly signalled a lower quality experience with a free app. These latter companies subsequently reverted back to their existing pricing after losing a big chunk of revenue.

Perhaps today's app are jumping on the app constellation bandwagon too quickly without first thinking about their unique business and testing such a major product change with their audience. Or, they may not realize that a successful app constellation strategy takes time to catch on (behavioral change is difficult, after all), in the same way they built their original communities.

Marketing Manager at Appsfire

I think that having several apps work only if you have a key value proposition for those apps or/and if you invest money & time to promote it.

App constellations are likely a result of app developers to having to wait too long to have widgets on iOS. A core app could have a brand line extension and drive it's own downloads of a new feature set, and it would make it easy for disparate teams at a large company to build a subset of features that are deeplinked together with the core app.

If instead of separate apps there were many simple widgets for Facebook users might have an easier time dealing with them - deleting a widget wouldn't make the app functionality break down which we see with 4sq and swarm. Also, Evernote has great widgets for android but had to build separate apps for iOS, even though it's not likely except in the case of Skitch that someone would use an Evernote app if they weren't already an Evernote user.

For me I am much more interested in the next two trends 1) how apps start building iOS extensions (widgets) instead of new apps 2) those apps that actually bundle, instead of unbundle features, and how those apps are deleting needless features to streamline workflows on mobile. More on that here:

Reply · Tweet ·

The CBInsights article asks, "App Constellations - Are They Working?" and apparently wants us to conclude they're not...but the data they've chosen don't actually address the question, for two reasons: (1) they've misunderstood Fred Wilson's point and (1) they're treating a heterogeneous data set as though it were homogeneous.

(1) Fred Wilson describes App Constellations as a distribution strategy, not a retention one: "If you own a leading constellation, you can use your apps and your relationship with the users of those apps to promote and distribute new apps that you either build or buy." And nearly every app in the CBInsights post starts out at the top of the charts, meaning the strategy is working. He doesn't suggest App Constellations will drive engagement; the developer is still responsible for building something that meets user needs effectively. So the fact that many of those apps don't stay there isn't a problem with App Constellations, it's a problem with the app itself (or simply the result of it settling into a niche market).

(2) Like Alex Schiff, I see two categories of app here: spin-outs of existing features, and experiments. I find it difficult to believe that Paper or Slingshot was intended as a permanent fixture in the Facebook ecosystem. R

Reply · Tweet ·

[continued...not sure why my comment autosaved mid-sentence]... They're experiments: Facebook can take advantage of the distribution channel to try stuff out, much like a big A/B test. They learn something whether the app succeeds or fails.

Messenger, Carousel, Camera, and Swarm, on the other hand, are core features spun out into new apps—additions to the constellation. Skitch, Penultimate, and Mailbox are similar but came in via acquisition. Again, though, success or failure is dependent not merely on the distribution power of the constellation but also on their viability as apps. The constellation just gets them that top rank out of the gate. So to me it's no surprise that Mailbox, which serves its target users well, is maintaining a solid rank; or that Messenger, a proven feature with a familiar experience and a network effect, is staying solidly in the top ten; or that Carousel, which I think failed to differentiate itself sufficiently from the built-in Photos app, has dropped in the rankings. (And while, as Bubba Murarka said, Dropbox can probably "re-up" the distribution in various ways, it'll be a temporary fix as long as Carousel doesn't keep users engaged.)

In other words, if CBInsights' data do anything, it's to support Fred Wilson's original thesis: regardless of app quality, these large companies can use the constellation approach to drive distribution.

Reply · Tweet ·
VP Product / Co-Founder @hellochime at Sittercity

If we classify installs / adoption as the primary metric of success, then there are only few examples of "success" and more of "failures". However, I think we should look at this a bit differently, in the context of where we are in mobile (vs. the web). Perhaps a more apt terms than "constellations" is really "app families", where you have "parent" apps (the main one to get traction) and "children" apps (the additions to the family over time).

In this context, I can see two primary approaches as to why companies "start a family" each with different goals, and therefore success metrics:

1) UNBUNDLING: In this case the parent app spins-off functionality from it into a standalone app (or very linked). For the most part, since the *primary* acquisition channel is from parent -> child, we can assume that in the dominant case, both parent & child apps would coexist on the user's device. Examples here are Facebook & Messenger, Foursquare & Swarm, Instagram & Layout, Twitter & Periscope, etc. The dominant strategy here is to break off the specific use cases to either optimize them (i.e. make it simpler / easier than in the main app), or to let them take on a life of their own, and thereby remove them from the shackles of the parent app. In this case, it is highly unlikely that the child app will ever gain the traction of the parent app, and instead we should look at whether existing users of both apps are more engaged than just the parent app users (based on either time spent, frequency, or activity depending on the use case).

2) EXTENSION: Here, is where the "child" app is usually acquired (e.g. Facebook & WhatsApp, Dropbox & Mailbox) or as a new line of business (think of a new "tab" or "section" on the homepage in the web world). The strategy here is to test & enter new markets and use cases, so by definition this strategy will never go away, in my opinion. With this strategy installs could be an acceptable metric; however, when testing new markets / opportunities, or warming them up, where more internal metrics (e.g. revenue) may be a better yardstick. One additional overlooked case is where these extensions span different geographic markets, or completely different customer segments, rather than multiple apps to the same user.

So bottom line, I think it's way too soon to say to judge winners & losers of app constellations yet -- especially if we judge by installs. Moreover as businesses scale, acquire companies, and enter new markets, I believe that these "app families" become inevitable -- especially the "extension" strategy.

Reply · Tweet ·
Andrew Chen, Quibb, Uber Andrew Chen
Quibb, Uber
Andrew Goldner, Quibb, GrowthX, Kauffman Foundation Andrew Goldner
Quibb, GrowthX, Kauffman Foundation