John Koetsier
Mobile Economist at TUNE, writer for Forbes
Where to hear and read about John Koetsier1 episode or test
John Koetsier is Mobile Economist at TUNE and a near-daily Forbes writer who backs his views on app marketing with large datasets on ad networks, keywords, ratings and Fortune 1000 apps. The most useful lesson from him is to assume paid installs are partly fake and to judge every channel by what users do in the app, not by download counts.
15% of the clicks are just not real
Insights on the Future of App Marketing with John Koetsier
John Koetsier: the story and the lessons
1Who John Koetsier is and how he works
Koetsier is TUNE's Mobile Economist and writes for Forbes almost daily. He studies app marketing through data and sees the industry's claims first-hand, since nearly every press release he reads boasts of download numbers.
He has also written about influencer marketing for Forbes, which he calls "huge".
He builds his conclusions on very large datasets. He studied 702 ad networks, with solid data on 455, covering 25 billion clicks over three months. A separate study covers about 13,000 Fortune 1000 apps, 95 billion installs and 4.5 billion ratings and reviews, matched to financials and stock performance.
Koetsier says app marketing is getting harder on every front. Search, ratings, paid ads and big rivals no longer favor small apps, and voice assistants will change the game again. His advice is to keep adapting as if it is always day one.
2Koetsier on ad fraud and paid acquisition
Koetsier tells builders to assume about 15% of paid clicks are fake. Average fraud across his sample was 15%, and almost a quarter of networks had more than 20% fraud.
Well-known, respected networks are affected too. Because so many traffic sources feed any one network, some fraud is guaranteed.
He checks four signs of fraud in each network's installs. The signs are installs that arrive too fast, installs that do not follow a natural curve, installs from outside the targeted geography, and installs from known data centers.
Among the worst 34 networks, 5% were 100% fraud. His test is to graph a network's installs and ask whether the curve looks natural.
Koetsier prefers automated fraud screening to manual checks. Watching usage and in-app milestones does catch fraud, but the feedback cycle is long and it takes work. A tool should flag odd patterns across ten or a hundred ad partners quickly.
He says to show ad partners the data and ask for refunds. Uber, Twitter, Amazon and Supercell, which work with TUNE, suspected a third of their installs were fraudulent and took data to their ad partners.
The partners did not argue. As Koetsier puts it, "they just cut a check" or offered replacement traffic.
Koetsier judges acquisition by engaged users, not download counts. Press releases boast of 50 or 100 million downloads, which he says means almost nothing. Money comes from engaged, value-creating users.
His rule is "front-end metrics mean nothing", so measure what people do inside the app.
3Koetsier on search, ratings and organic growth
Koetsier treats search as the main install channel. Apple says search precedes about 65% of app downloads. Google says 83% of quality installs come from search.
He argues ASO and search visibility deserve serious investment.
His keyword research shows big brands own the top search terms. All of the top 10 iOS keywords by volume are branded, and so are 90% of the top 50. Large companies own about 88% of the top 50.
To get that figure he first removed some firms. One "start-up" had $70 million in VC and a market cap near $400 billion. His conclusion is that small apps should expect to buy more users.
Koetsier says app marketing must reach beyond App Store optimization. It used to be a niche with its own core techniques. Now it also needs PR, social, web search and influencer marketing.
One speaker at the event gets many organic installs from SEO.
He finds ratings no longer separate top apps. Top Android apps average 4.16 stars with a variance of 0.3, and top iOS apps average 3.81 with a variance of 0.6. The figures cover the top 240 Android and top 150 iOS apps in each category.
Good ratings get an app into the top tier and set it apart from bottom-feeders, but top apps must stand out in other ways.
Koetsier says top apps ask happy users for reviews, and you should too. About 86% of Google Play ratings are four or five stars, because apps prompt the right user at the right moment.
He says "they're gaming the App Store they're gaming Google Play", and that skipping the tactic means losing out.
4What Koetsier expects next for apps
He expects big enterprises to compete seriously in mobile. Exxon Mobil has 400,000 mobile users and Caterpillar has 355,000, using its app to order parts and log service. Strong mobile performers had stock valuations up to 15% better.
Amazon has 440% more mobile customers than all its competitors combined. Big brands want a one-to-one relationship with customers, so small developers cannot count on being the nimble ones.
Koetsier expects assistants and bots to take over multi-app tasks. One request to Alexa could plan a Chicago trip with Cubs tickets, dinner at Morton's, a hotel and a Friday flight after 3 p.m., replacing four or five apps.
He rejects Nadella's line that bots are the new apps, but says booking or buying in Messenger or SMS has less friction than downloading an app. About 70% of American digital time is on mobile, and about 90% of that is in apps.
He says the gold rush is over and user acquisition is harder. People already have the apps they need and keep five or six casual games. Adobe data shows US app growth down 38% since 2014.
He notes Adobe's data leans toward big enterprise. He also floats glasses as a possible eventual replacement for the phone.
Koetsier uses Pokemon Go to show why a follow-up must be ready. In one month of 2016 it had more daily usage time than Facebook, Twitter, Instagram, Snapchat and Tinder combined. It still earns, but far below its potential.
It lacked new content ready, such as gyms players could teleport into, or play for rainy days.
He reads Bezos's day one idea as a warning for every app company. Koetsier calls Bezos's yearly shareholder letters a cheap MBA that costs only time. Day two is stasis, then irrelevance, then painful decline and death.
Few companies return from day two to day one. Koetsier names Apple as a probable case and Microsoft as a possible one now.
Insights on the Future of App Marketing with John Koetsier
App marketing is getting harder: search is dominated by big brands, ratings no longer differentiate, and ad fraud is widespread. TUNE's data covers 25 billion clicks across 455 ad networks, finding 15% fraud on average, which makes it worth hearing before you spend on paid acquisition.
Koetsier is TUNE's Mobile Economist and writes for Forbes almost daily. He studies app marketing through data and sees the industry's claims first-hand, since nearly every press release he reads boasts of download numbers.
He has also written about influencer marketing for Forbes, which he calls "huge".
He builds his conclusions on very large datasets. He studied 702 ad networks, with solid data on 455, covering 25 billion clicks over three months. A separate study covers about 13,000 Fortune 1000 apps, 95 billion installs and 4.5 billion ratings and reviews, matched to financials and stock performance.
Koetsier says app marketing is getting harder on every front. Search, ratings, paid ads and big rivals no longer favor small apps, and voice assistants will change the game again. His advice is to keep adapting as if it is always day one.
Koetsier tells builders to assume about 15% of paid clicks are fake. Average fraud across his sample was 15%, and almost a quarter of networks had more than 20% fraud.
Well-known, respected networks are affected too. Because so many traffic sources feed any one network, some fraud is guaranteed.
He checks four signs of fraud in each network's installs. The signs are installs that arrive too fast, installs that do not follow a natural curve, installs from outside the targeted geography, and installs from known data centers.
Among the worst 34 networks, 5% were 100% fraud. His test is to graph a network's installs and ask whether the curve looks natural.
Koetsier prefers automated fraud screening to manual checks. Watching usage and in-app milestones does catch fraud, but the feedback cycle is long and it takes work. A tool should flag odd patterns across ten or a hundred ad partners quickly.
He says to show ad partners the data and ask for refunds. Uber, Twitter, Amazon and Supercell, which work with TUNE, suspected a third of their installs were fraudulent and took data to their ad partners.
The partners did not argue. As Koetsier puts it, "they just cut a check" or offered replacement traffic.
Koetsier judges acquisition by engaged users, not download counts. Press releases boast of 50 or 100 million downloads, which he says means almost nothing. Money comes from engaged, value-creating users.
His rule is "front-end metrics mean nothing", so measure what people do inside the app.
Koetsier treats search as the main install channel. Apple says search precedes about 65% of app downloads. Google says 83% of quality installs come from search.
He argues ASO and search visibility deserve serious investment.
His keyword research shows big brands own the top search terms. All of the top 10 iOS keywords by volume are branded, and so are 90% of the top 50. Large companies own about 88% of the top 50.
To get that figure he first removed some firms. One "start-up" had $70 million in VC and a market cap near $400 billion. His conclusion is that small apps should expect to buy more users.
Koetsier says app marketing must reach beyond App Store optimization. It used to be a niche with its own core techniques. Now it also needs PR, social, web search and influencer marketing.
One speaker at the event gets many organic installs from SEO.
He finds ratings no longer separate top apps. Top Android apps average 4.16 stars with a variance of 0.3, and top iOS apps average 3.81 with a variance of 0.6. The figures cover the top 240 Android and top 150 iOS apps in each category.
Good ratings get an app into the top tier and set it apart from bottom-feeders, but top apps must stand out in other ways.
Koetsier says top apps ask happy users for reviews, and you should too. About 86% of Google Play ratings are four or five stars, because apps prompt the right user at the right moment.
He says "they're gaming the App Store they're gaming Google Play", and that skipping the tactic means losing out.
He expects big enterprises to compete seriously in mobile. Exxon Mobil has 400,000 mobile users and Caterpillar has 355,000, using its app to order parts and log service. Strong mobile performers had stock valuations up to 15% better.
Amazon has 440% more mobile customers than all its competitors combined. Big brands want a one-to-one relationship with customers, so small developers cannot count on being the nimble ones.
Koetsier expects assistants and bots to take over multi-app tasks. One request to Alexa could plan a Chicago trip with Cubs tickets, dinner at Morton's, a hotel and a Friday flight after 3 p.m., replacing four or five apps.
He rejects Nadella's line that bots are the new apps, but says booking or buying in Messenger or SMS has less friction than downloading an app. About 70% of American digital time is on mobile, and about 90% of that is in apps.
He says the gold rush is over and user acquisition is harder. People already have the apps they need and keep five or six casual games. Adobe data shows US app growth down 38% since 2014.
He notes Adobe's data leans toward big enterprise. He also floats glasses as a possible eventual replacement for the phone.
Koetsier uses Pokemon Go to show why a follow-up must be ready. In one month of 2016 it had more daily usage time than Facebook, Twitter, Instagram, Snapchat and Tinder combined. It still earns, but far below its potential.
It lacked new content ready, such as gyms players could teleport into, or play for rainy days.
He reads Bezos's day one idea as a warning for every app company. Koetsier calls Bezos's yearly shareholder letters a cheap MBA that costs only time. Day two is stasis, then irrelevance, then painful decline and death.
Few companies return from day two to day one. Koetsier names Apple as a probable case and Microsoft as a possible one now.