Lior Barak
Data consultant, Tale About Data
Where to hear and read about Lior Barak1 episode or test
Lior Barak is a data consultant behind Tale About Data who helps subscription app teams turn marketing data into daily decisions. The most useful thing to learn from him is to track only the few numbers that trigger a clear action and bury the rest.
what is the action I'm driving out of this data I'm collecting
To Increase App Subscriptions You Need to Look At This Data with Lior Barak
Lior Barak: the story and the lessons
1How Lior Barak times retargeting on subscription apps
Barak starts by finding when the revenue event happens. He groups installs into cohorts and measures the days users take to subscribe, then designs drip campaigns to fit that window.
His example is a client app with a seven-day conversion window. Without this data, he says, the retargeting decision is made blindly.
He investigates users who don't return twice in the first week. On subscription apps, most users who haven't come back at least twice in the first seven days delete the app within 30 days.
A user with only the install event and no return visit by day seven has a problem to diagnose and a trigger to find.
Barak retargets to reopen the app first and chases reinstalls later. In the first 30 days he runs campaigns to get users to open the app. Reinstall campaigns start after 30, 60 or 90 days.
This spreads a limited budget according to how users actually react.
He caps win-back spend at what a user is worth. If a user brings around $15 in revenue, spending $100 to win them back makes no sense.
Barak points to Peak as a model of retargeting with popular content. The brain training app retargets heavily in the first seven days, with ads that show its most popular games.
The aim is a return visit inside the window where subscribing is most likely.
2Why Barak limits daily KPIs to three
Barak allows no more than three daily KPIs. More than three makes decisions impossible, and the more a client spends, the more their data must be compressed into a few usable signals.
He applies the cap to every client, including those spending tens of thousands of dollars, but the choice of KPIs depends on the client.
His daily picks are campaign ranking score, revenue and spend. Read together, spend and revenue show whether a change in spend moves revenue. If spend falls and revenue falls with it, something in the setup is wrong.
If cost jumps while revenue stays flat or declines, that is a warning he can act on the same day. The three KPIs can change over time, but the habit of watching how revenue responds to spend stays.
Barak moves installs to a weekly review. Fridays run low and Sundays peak, when bored people at home search for apps, so a daily install count triggers no action.
A weekly check works better, for example 50,000 installs in a week as a sign of a good job.
He prefers logic-based metrics over emotional ones. Installs are an emotional feed. Revenue, clicks or ads viewed carry a rule behind them, such as 15 to 25 ads viewed per user per day to turn a profit on an ad-funded app.
Organic users spend no money, so a cost-per-install number says little for them.
Barak compresses many session events into one campaign score. A data scientist builds a model that rates each campaign's chance of success against the others, turning tons of data points into one number such as 8.7.
The index signals the worth of a session and of the user just brought in.
He uses one source per measure and never blends them. Clicks come from Facebook, Google, Adjust and AppsFlyer, and they never match. Barak takes impressions and clicks from the ad partner, not from attribution tools.
Mixing impressions from one source with clicks from another gives a wrong score, and he calls it the common mistake.
Barak buries data that drives no action. Being data-driven means knowing your sources and asking what action each number triggers.
Unused data goes in a "graveyard" until it is needed, because looking at it only mixes you up.
3Barak's data discovery workshop and three dashboards
He makes teams cap their measures before building anything. In a data discovery workshop, teams first ask any question, then mark which measures are must-haves and which are nice-to-haves, under a limit.
A team starting with 50 or 60 measures has to choose with care, and the chosen measures decide how the tables are built.
Barak builds three dashboards for three rhythms of decision. The daily one is action-driven: which campaign to kill, which to improve. The weekly one adds research, and the monthly one is a deep dive for the team.
The monthly view covers what went wrong and what to fix next.
He reallocates budget about once a month. Budgets are usually fixed for the year, so changing stuff too often hurts campaigns and clients.
His example is a monthly review showing that more money should have gone to Facebook because Snapchat soaked up spend.
4Why Barak wants teams to trust and automate their reporting
Barak says teams waste 60% to 70% of their time on automatable work. That share goes to campaign creation and optimization. He puts a data team between engineers and marketers to automate it.
Engineers build the databases and automation, while marketers keep launching campaigns so the machines can learn and readjust. The data team runs the automation but does not run campaigns.
He sees distrust of numbers as the top issue with new clients. Even clients with Looker or Power BI download data into Excel and rework it, which wastes time and adds errors.
One wrong VLOOKUP can spoil every decision. His fix is a data dump built the right way, so nobody has to dig through it.
Barak matches the fix to whether a data engineer is on the team. Without an engineer, find a tool that automates the work. With one, put them to work turning the data into something usable.
The goal is data that people can actually use and trust.
To Increase App Subscriptions You Need to Look At This Data with Lior Barak
Teaches which data to watch to grow app subscriptions: find when users convert, retarget inside that window, and limit daily KPIs to three that drive action. The payoff is a simple system of daily, weekly and monthly views that replaces blind spending and spreadsheet work.
Barak starts by finding when the revenue event happens. He groups installs into cohorts and measures the days users take to subscribe, then designs drip campaigns to fit that window.
His example is a client app with a seven-day conversion window. Without this data, he says, the retargeting decision is made blindly.
He investigates users who don't return twice in the first week. On subscription apps, most users who haven't come back at least twice in the first seven days delete the app within 30 days.
A user with only the install event and no return visit by day seven has a problem to diagnose and a trigger to find.
Barak retargets to reopen the app first and chases reinstalls later. In the first 30 days he runs campaigns to get users to open the app. Reinstall campaigns start after 30, 60 or 90 days.
This spreads a limited budget according to how users actually react.
He caps win-back spend at what a user is worth. If a user brings around $15 in revenue, spending $100 to win them back makes no sense.
Barak points to Peak as a model of retargeting with popular content. The brain training app retargets heavily in the first seven days, with ads that show its most popular games.
The aim is a return visit inside the window where subscribing is most likely.
Barak allows no more than three daily KPIs. More than three makes decisions impossible, and the more a client spends, the more their data must be compressed into a few usable signals.
He applies the cap to every client, including those spending tens of thousands of dollars, but the choice of KPIs depends on the client.
His daily picks are campaign ranking score, revenue and spend. Read together, spend and revenue show whether a change in spend moves revenue. If spend falls and revenue falls with it, something in the setup is wrong.
If cost jumps while revenue stays flat or declines, that is a warning he can act on the same day. The three KPIs can change over time, but the habit of watching how revenue responds to spend stays.
Barak moves installs to a weekly review. Fridays run low and Sundays peak, when bored people at home search for apps, so a daily install count triggers no action.
A weekly check works better, for example 50,000 installs in a week as a sign of a good job.
He prefers logic-based metrics over emotional ones. Installs are an emotional feed. Revenue, clicks or ads viewed carry a rule behind them, such as 15 to 25 ads viewed per user per day to turn a profit on an ad-funded app.
Organic users spend no money, so a cost-per-install number says little for them.
Barak compresses many session events into one campaign score. A data scientist builds a model that rates each campaign's chance of success against the others, turning tons of data points into one number such as 8.7.
The index signals the worth of a session and of the user just brought in.
He uses one source per measure and never blends them. Clicks come from Facebook, Google, Adjust and AppsFlyer, and they never match. Barak takes impressions and clicks from the ad partner, not from attribution tools.
Mixing impressions from one source with clicks from another gives a wrong score, and he calls it the common mistake.
Barak buries data that drives no action. Being data-driven means knowing your sources and asking what action each number triggers.
Unused data goes in a "graveyard" until it is needed, because looking at it only mixes you up.
He makes teams cap their measures before building anything. In a data discovery workshop, teams first ask any question, then mark which measures are must-haves and which are nice-to-haves, under a limit.
A team starting with 50 or 60 measures has to choose with care, and the chosen measures decide how the tables are built.
Barak builds three dashboards for three rhythms of decision. The daily one is action-driven: which campaign to kill, which to improve. The weekly one adds research, and the monthly one is a deep dive for the team.
The monthly view covers what went wrong and what to fix next.
He reallocates budget about once a month. Budgets are usually fixed for the year, so changing stuff too often hurts campaigns and clients.
His example is a monthly review showing that more money should have gone to Facebook because Snapchat soaked up spend.
Barak says teams waste 60% to 70% of their time on automatable work. That share goes to campaign creation and optimization. He puts a data team between engineers and marketers to automate it.
Engineers build the databases and automation, while marketers keep launching campaigns so the machines can learn and readjust. The data team runs the automation but does not run campaigns.
He sees distrust of numbers as the top issue with new clients. Even clients with Looker or Power BI download data into Excel and rework it, which wastes time and adds errors.
One wrong VLOOKUP can spoil every decision. His fix is a data dump built the right way, so nobody has to dig through it.
Barak matches the fix to whether a data engineer is on the team. Without an engineer, find a tool that automates the work. With one, put them to work turning the data into something usable.
The goal is data that people can actually use and trust.