You are currently browsing the tag archive for the ‘ltv’ tag.
[Also published on Medianama]
It has been only five years since the launch of the iPhone App Store in July 2008. Feels like fifty dog years. In reality this is not a long time, compared to nearly twenty years since the launches of Yahoo (February 1994), Amazon (July 1995) and the IPO of Netscape (August 1995). Over these twenty years, not only have startups innovated on product/design and business models and but also on demand generation/user acquisition strategies. Yet only five years after the launch of the App Store, the pace of innovation in mobile app user acquisition seems to have hit a brick wall… in the search for increasingly efficient methods of marketing, we seemed to have hit the efficient frontier.
In India, efficient user acquisition is a key problem area for developers targeting Indian users as well as global users. Fortunately, marketing has gone online, along with placement, onboarding, monetization and payments. And mobile marketing can be done at world-class levels right in India.
So, what is this efficient frontier? What are the best practices for mobile user acquisition?
To provide some concrete pointers, I organized several founders-only sessions on enterprise/SMB SaaS user acquisition as well as mobile user acquisition in Bangalore, Delhi and Mumbai. This month, I also organized and moderated a session on mobile user acquisition with TIE in Delhi. Our eminent panelists included Harinder Takhar (CEO of PayTM), Pathik Shah (Head of Growth, Hike), Jamshed Rajan (Chief Product Officer, Nimbuzz) and Chandan Gupta (founder/CEO of PhoneWarrior).
So here’s a summary of what we discussed – please note the tone of the conversation was more around hacks and learnings from practitioners as opposed to some over-arching strategic viewpoint on mobile user acquisition. Many of these tips fall into the non-scalable bucket but some are more scalable. I will leave it up to you to decide which is which. Also, it was assumed that developers were tracking efficiency of marketing campaigns and funnels through some form of app instrumentation, whether through commercial solutions like Mixpanel, Apsalar, Flurry, Google Analytics etc or home-grown analytics.
CONVENTIONAL PRE-MOBILE TECHNIQUES
These include traditional PR/media outreach, analyst relations, direct selling and tradeshows/conferences. These techniques are fairly inefficient and out-of-date for mobile apps as most target users/consumers are not reached through these means.
Blogs/websites: Chasing Techcrunch and other tech blogs does not have nearly the same effect it had a few years ago – previously, a post on Techcrunch could drive 50-100k visitors/downloads – now, this number is down to 100-500 downloads.
Vernacular newspapers: Targeting vernacular media outlets across India, as opposed to the English and Hindi dailies could provide some advantage. Regional language papers are hungry for technology news and can be quite effective in reaching regional audiences.
Localization: On a related note, for some apps, it makes sense to provide app store listings in several different languages, sometimes backed up by the product being localized as well but not necessarily.
TV: In India, it could be useful to get onto NDTV Cell Guru and other such shows. These media outlets also have Facebook, web, mobile and video assets to drive awareness.
Print: Some panelists had tried this. It does not have any meaningful impact on mobile app downloads.
Offline: Some panelists had tried stationing people on campus to get some initial adoption. It does not work and the message gets diluted/warped when temporary employees are hired to do this.
MOBILE 1.0 TECHNIQUES
These include OEM/mobile operator distribution, mobile advertising and search engine optimization (SEO).
A basic deterrent is app size – especially in Tier 2 towns and beyond, people are wary of downloading apps greater than 10MB in size. Really need to minimize app size.
Factory loading: Average OEM/carrier deals take 5-6 months at least and have to be positioned as helping the OEM/operator differentiate. Most OEMs are now looking at apps/services as revenue streams so this should be baked into the business case for them, perhaps as a rev-share. Some panelists mentioned Rs 5-10 per install as what Indian OEMs are asking for.
If the app is already factory-loaded onto the product, this doesn’t drive activation either – factory-loading has to be on the homescreen and accompanied by an above-the-line marketing campaign (e.g. advertising or logo on device box) preferably paid for and driven by the operator/OEM.
Some of the smaller/newer OS/OEMs providers are being more aggressive in courting developers. These include Tizen, Intel, Amazon, and Blackberry. If you build your app for these, you will maybe get an advantage and may get paid to build out on their platform. The flip-side, however, is that these platforms have small audiences and will most probably not drive a meaningful amount of downloads/usage. Panelists mentioned Parag Gupta at Amazon, Annie Mathew at Blackberry and Priyam Bose at Microsoft/Windows.
Mobile advertising: General consensus is that users acquired through paid advertising tend to be less loyal than users acquired organically. One exception may be advertising to users of competitor apps on Facebook and the use of promoted posts on FB. Panelists mentioned Google/Admob, Inmobi, Flurry, Tapjoy, Yieldmo, HasOffers etc.
Mobile advertising gets an initial burst of downloads to move up into the top rankings on the app stores and then some drip marketing is required to keep rankings high. Some people expressed an opinion that any burst marketing should be done on one day rather than over several days and perhaps should be done on a Friday so the boost in rankings persists over the weekend. The key is to get into the top 10.
One needs 100k-200k installs per day to get into the top of the charts in India. Can’t get there through paid advertising. Advertising is not cheap. Especially given the messaging wars between Line, WeChat, Whatsapp, Hike and others, mobile inventory seems to be sold out in India.
If you measure real CPI (i.e. CPI taking into account successful download rates, activiation rates and 3 or 6 month churn), actual cost of customer acquisition (CAC) ends up 4-10x as high as CPI quoted by ad networks. In India, iPhone CPIs are under Rs 120 ($2) and Android Rs 30 ($0.50) at low scale.
There are mediation layers from Flurry,
Hasoffers, Mopub and others available so that developers don’t have to integrate multiple ad network SDKs into their apps. All these SDK providers have their own ad networks but also connect with other ad networks. Meanwhile, publishers use SSPs to route between ad networks. It’s a complete spaghetti-like mess.
Incentivized downloads: Tapjoy/Flurry used to provide this but have moved away from this. Panelists urged developers to not even think about trying incentivized downloads as CPIs are high as are uninstall rates, given that users are downloading without any intent to use.
Search engine optimization: Most developers mentioned that web SEO did not work for them. Content on the web does not bring traffic from the web to the app stores. Some people mentioned content marketing e.g. blog posts and posting presentations on Slideshare as a way to drive some traffic.
Mobile web: Make sure you have a http://get.yourwebsiteURL.com mobile-optimized website up and running. Apple and Android have special HTML widgets to include here that you insert once you know the OS of the device (through the header). These widgets redirect to the relevant app in the relevant app store.
Social media is not very effective for user growth. It is somewhat effective for engaging existing users as well as a support channel. Adding social network sharing within apps does generate some virality, especially if sharing is encouraged at points within the app where users get a delightful experience. Apps with social as their core may benefit from Facebook, including automated actions posting to Facebook (e.g. ‘read’ or ‘play’).
Virality: Startups should track their k-factor/viral-factor and viral cycle time. Even a k-factor of 0.2 really helps if it can be sustained over several months/years. A viral factor anywhere close to or greater than 1 is phenomenal but can only be sustained for a short period of time.
MOBILE 2.0 TECHNIQUES
App store optimization (ASO): The panel talked about platform stores (like Google Play, iOS App Store, Blackberry App World and Amazon), indie stores (like Getjar, Opera, UCWeb and Appia) and operator portals/stores. Most indie stores have a paid/sponsorship model but CPIs are the same as ad networks.
Platform app stores require carefully crafted keywords (repeated in title and description), creative content (which is mostly only read by loyal users), quality screenshots/logos and a good demo video for Google Play (linked through Youtube). Do not to go overboard here e.g. do not stuff keywords in the title/description – you will look desperate. Best tools for ASO include Google Trends, Searchman SEO, AppCodes. Reverse engineer the search algorithms on the app stores by typing in keyworks to see output of apps appearance.
Getting featured is obviously great but is driven purely through relationships (for Apple and Amazon) and algorithmically (for Google Play) with the curation teams for each platform, sometimes on a geography-by-geography basis. Always make sure to comply with the design guidelines provided by each platform – this makes it more likely you will get picked up for featuring. Use AppFigures and Appannie to track your performance and reviews.
App updates also drive additional downloads and push up ranking for a short period of time. Since there are no well-proven A/B testing methods for mobile apps, it makes sense to try several variations with each app update.
Cross-promotion: Companies like Outfit7, Zynga and Google have very effectively used their large network of apps to cross-promote new app launches. Outfit7 has been able to get to one billion+ downloads and has cross-promoted new launches to tens of millions of downloads in a few weeks.
Barter: Many developers don’t think about this, perhaps because it only applies when their apps get to some scale (several million MAUs). The trick is to find mobile app properties that (1) have tens of millions of MAUs; (2) have users in demographics/regions that you are targeting; and (3) have a large proportion of unsold or remnant inventory i.e. low sell-through rates. Bilaterally trading this remnant inventory can then be quite an efficient, not to mention cashless, way of driving downloads.
Referral schemes: Virality can be driven through incentives that provide an individual relevant app-specific user benefits in inviting people successfully. Examples include Hike (free SMSs for each successful invite), Dropbox (additional storage for each successful invite), Evernote (one month of free premium service with each successful invite) and Paypal ($5-10 for each successful invite). These schemes do not work for single-user utilities if you hand out real money. Users will try to hack around this system.
Beyond a certain point, only word-of-mouth/virality works, can’t use paid. This does not apply necessarily in the case of apps where the lifetime value (LTV) of an average user has been quantified, as can be done with many user-paid models like games, ecommerce and subscription services.
Use social influencers: If you can identify and target social influencers, it sometimes works to make them proponents of your app.
Gamification: Leaderboard-based incentivization does not impact new user acquisition. Make sure that gamification works even if the user does not have any friends using the same app.
Push SMS marketing: CPIs end up being within 25% of where the ad networks are, so not much different in price. Historically, SMSs went out to non-data, non-smartphone users as well so were not effective. This can contribute to cheapening the brand. Also, TRAI has specifically banned sending spam SMSs to users on the DND list.
Restricted invite lists: This is what Mailbox did, as have many others. A permanent beta is a less extreme example of this. This make sense for apps like email which need to be scaled up slowly given their complexity. However, restricted invite lists only make some sense when there is a lot of PR and noise generated some other way to drive artificial scarcity.
Review sites: At small scale, this helps. Some developers pepper comments throughout review sites such as Appolicious, AppTurbo and AppBrain to drive some downloads. This also build links into the developer’s website to drive Google search rankings.
Here’s the presentation I gave at the IAMAI Digital Commerce event this morning:
[Published in Pluggd.in]
I am speaking at IAMAI’s conference on Digital Commerce later this week in Mumbai. I thought I would put down a few thoughts here that I believe affect ecommerce in India as an industry.
There has been an increasing amount of debate recently around the sustainability of ecommerce companies in India. I believe that a key driver of sustainability is a sharp focus on long-term customer value and a deep understanding of customer metrics. Delivering strong value to customers results in high repeat purchase rates and low customer acquisition costs, while an analytical orientation enables companies to measure key metrics and take important business decisions based on real data.
Everyone understands intuitively that repeat customers are good for business. Yet very few e-commerce companies develop a deep understanding of customer behavior, measure and analyze key metrics and tailor their business strategy and internal focus accordingly.
Understanding the value of a customer – and how to measure it – is perhaps one of the most important questions for anyone running an ecommerce business. If you don’t know what a customer is worth, you run the risk of not knowing: (i) how much you should spend on marketing/customer acquisition, (ii) how much you should spend on customer support (customer service, fulfillment etc), and (iii) the levers that drive an increase in the value of your customers (and therefore the value of your business). Companies that don’t understand customer value may be able to grow rapidly, but this will be unsustainable over time, because the costs they incur to acquire, reacquire and support their customers may greatly outweigh what those customers are worth.
All e-commerce entrepreneurs I meet share a repeat customer metric (indicating an appreciation that repeat customers are valuable), but more often than not these metrics don’t reveal insight into customer behavior. Some examples include:
- “33% of the orders last month were from repeat customers”, OR
- “50% of the customers that bought last year bought again this year”.
OK, this sounds great, but what does it actually mean and how is this data actionable for the business?
Let’s take the first statement: “33% of the orders last month were from repeat customers”. Take a company that has been in business for a couple of years and during that time served over 200,000 customers. Assume that they currently serve 21,000 customers per month. So this statement means that 7,000 customers (or <4% of their cumulative customer base) in the month had bought at some point before. Is that good? Bad? What does it mean? What actions should be taken? It’s tough to tell because this statement is meaningless without more context.
Let’s take the second statement and apply it to the same company above: “50% of the customers that bought last year bought again this year”. This implies that 100,000 existing customers bought again this year. I think we would all agree that sounds pretty good. But is it actionable? What did they buy? How much did they spend? How often did they buy again?
To help answer some of these questions and drive specific, actionable insight for an online business, it helps to think of customers a little differently. On the internet, a customer is like a store in the physical world – you need to make upfront investments (acquisition cost on the internet; capex for a physical store) which will yield a margin stream in the future. Understanding the ratio of the upfront investment to the expected margin stream, as well as how you can reduce the investment and increase the margin stream, is critical. Just as the operator of a physical store thinks about time to break-even and pay-back, operators of online stores can do the same with their customers.
Once you accept this premise, you can use cohort analysis to estimate customer lifetime value. Cohort analysis tracks the behavior of a specific ‘batch’ of customers (for example the January 2010 cohort means customers acquired in January 2010) over time. You can do this for multiple cohorts as follows:
The table below illustrates the behavior of the January cohort, which is the customers that were originally acquired in January.
The next table indicates the order value generated by the January cohort.
What we see is that the January cohort of 1,000 buyers spent a total of Rs. 29L during the year. If this business has a contribution margin (i.e. shipped revenue less COGS and variable costs such as discounts, payment gateway, shipping & handling) of 15%, then the cohort of 1,000 customers generated Rs. 4.35L of contribution margin in the year or Rs. 435 per customer. Because the number of transacting customers as a % of the starting base has begun to asymptote in 12 months, you can project this forward and calculate lifetime value over 2 or 3 years (note, this is NOT year 1 value multiplied by 2 or 3). Lets assume the 3-year lifetime value of a customer based on this analysis is approx. Rs 800. This should become the maximum allowable acquisition cost you pay to acquire a customer in steady state.
Many of the most successful e-commerce companies in our portfolio achieve lifetime value to customer acquisition cost ratios (LTV/CAC) in excess of 3:1, which enables them to grow rapidly and profitably through aggressive marketing. Since acquisition costs can only be controlled to a certain extent given media costs, competitive dynamics etc, they do this primarily by focusing on customer economics, and specifically increasing: (i) frequency of purchase, (ii) margins, and (iii) order values (through effective retention marketing and initiatives).
Most companies that measure this data carefully allocate more of their resources to retaining customers and improving customer economics than companies that don’t. After all, it’s much easier to ramp up customer acquisition if you already have systems in place to maximize the value of those customers than it is to change the DNA of an organization that focuses only on customer acquisition.