Mobile performance marketing campaigns are often evaluated on the basis of Cost per Install (CPI), or the cost per app download (and sometimes some further action, such as opening the app or reaching some stage in the on-boarding process).
Pricing campaigns on any cost-per-action (CPA) basis – versus a Cost per Mille (CPM) or Cost per Click basis — helps advertisers manage risk: by only paying for app installations (and not impressions), an advertiser can arbitrage per-user adoption costs against predicted lifetime customer value. This is the foundation of the LTV > CPI equation.
Of course, advertising networks can’t actually moderate install volumes on a CPI basis: they have no control over the ad creatives used by advertisers, and publishers are generally paid on a CPM basis (meaning they receive some set amount of money per 1,000 ad impressions served).
This is why CPI campaigns are “fuzzy”: ad networks can’t guarantee the number of installs at a certain per-install price because they have no idea how many impressions, which are sold at a fixed price, will convert. In other words, an advertiser can buy a fixed amount of impressions at a fixed price, but since the number of installs generated from that fixed amount of impressions is unforeseeable, the number of installs can merely be estimated.
One of the determinants of CPI is Click-through Rate (CTR), or the percentage of people that click on an ad upon seeing it. A higher CTR produces a higher number of people being taken to the app store listing for an app. If an app’s ad is shown 1,000 times and it receives 100 clicks, its CTR is 10%.
The other determinant to CPI is the install rate, or the percentage of people that install an app after having clicked on an ad for it. If an app’s ad is clicked 100 times and produces 10 installs, its install rate is 10%.
CTR is affected by ad creative, but the same can’t be said for the install rate. Once a potential user has landed at an app’s store listing, their decision to install the app or not is almost entirely influenced by what they see there (the app’s screenshots, its rating, its reviews, etc.).
For this reason, CTR and install rate might move in opposite directions: as an app’s ad creative becomes more generic, its CTR might increase while its install rate decreases. This is intuitive: broadly appealing advertising will attract more interest than precisely targeted advertising, but if the app itself isn’t broadly appealing, that interest won’t necessarily convert into installs.
But install rates don’t necessarily decline to the same extent that click-through rates increase when advertising targeting is broadened; some of the increases in clicks may convert to installs. When this happens – an advertising campaign is made more broadly appealing, but its install rate doesn’t decrease to the same degree that CTR increases – then CPI decreases.
This is the CTR conundrum: how aggressively should an advertiser chase clicks, and how can the interplay between click-through rate and install rate be optimized?
Consider the following three scenarios:
In scenario 1, when CTR doubles against the baseline and the install rate subsequently halves, CPI remains the same. But in scenario 2, when CTR doubles against the baseline but the install rate only decreases by 40%, CPI decreases by 17%. And in scenario 3, when CTR doubles against the baseline but the install rate decreases by 60% (ie. it decreases to a greater extent than CTR increases), CPI jumps by 25%.
The optimal CTR / install rate combination is obviously that which minimizes CPI at a scale approaching the app’s total addressable market. For broadly appealing apps, there’s no reason that an increase in CTR needs to be accompanied by a decrease in to install rate. But for niche apps with small or very specific total addressable markets, rendering ad creative more accessible may significantly impact install rates and thus drive CPI up without delivering additional installs (as in scenario 3).
The allure of increasing CTR is undeniable, but CTR optimization independent of install rate and overall CPI is meaningless. An increase in CTR could potentially result in 0 additional installs if ad creative is uninformative (or, worse yet, misleading). Thus campaign optimization can’t be undertaken with a singular focus on CTR; it necessitates a holistic strategy with an end goal of minimizing CPI at scale and maximizing overall profit, which itself might preclude CTR increases.
Recent