#13 Mastering App Review Management: AI, Automation, and Hybrid Strategies for Success
Anatoly Sharifulin
CEO at AppFollow, App Reputation Automation Platform | B2B SaaS
During our recent visit to the WN Conference (a return after 5 years!), discussed many different topics relating to app reputation management. I have learned a lot in these 5 years, yet the topic's core has never changed.?
To get you up to speed on what I’ve talked about, here’s a recap, spiced up with our extra thoughts on the matter. Enjoy!?
Users talk everywhere—not just on the app stores
During simpler days of the past, you would expect to hear from your community where it resided mostly: a single forum, and then you would get contacted via your support lines: email and the review section of one or two app stores you were in.
Things have surely changed since then. Today, you can hear about your app in so many different places: social networks, a Discord community, a whole variety of app stores, and the old and trusted email support line.
Generally, we’ve split these channels into three tiers, by community presence and level of review/feedback:
Thus, the biggest shift we’re seeing is that aggregation of user reviews and opinions about your product must be from more sources than just app stores and emails.?
In these other channels that are so often not even given a single look, so much happens. Users share their opinions, suggestions, grievances, monetization concerns—anything that they want to talk about, and yet there is no reliable way to aggregate this data and act on it.?
Discord communities, in particular, are a unique channel that often has a lot of intense and quick feedback—regardless of how popular the game is. This kind of feedback is so important for your Product, Support, and Marketing teams to see—otherwise, bugs may be missed, suggestions not implemented, and other cues towards improvement simply missed.
What’s more, every app, depending on the category and community, will have a different share of feedback across available channels. Some apps get most of their feedback via Review sections, and some see most of it on social media. At the same time, all of this feedback is important—you shouldn’t ignore it.?
For many teams working on these apps, review feedback has a whole host of KPIs they work with and measure their success with. 4.0 star rating is a great mark to strive for, and 4.5 is the ultimate success benchmark.
At AppFollow, we’re making sure you will be able to work with all these channels soon. For the time being, reviews in app stores are not going anywhere, and it’s still one of the best places to look for feedback on your app.
The true power of both ratings & reviews
Aren’t ratings and reviews one and the same? Indeed not. At least, not really. While it’s the same system, what you can consider a “rating” is a quick review that doesn’t say much. A couple of words, a one-liner, anything with a star rating attached to it that’s not telling a story. A “review”, in our books, is a star rating with an expanded opinion on the matter. Based on reviews, you can measure sentiment scores, discover suggestions, learn about bugs, and tackle many other things your users feel like telling you.
Both ratings and reviews are quantifiable; improving the star rating, reply speed, reply effect, and other metrics is something you can set as a KPI. Above, we’ve talked about different channel tiers; Tier 1 and Tier 2 are difficult to quantify as these are all different formats of feedback. Tier 3 is easy to measure.?
Since it’s a lot easier to leave a positive rating without elaborating enough for it to be a full-fledged review, you can also quantify the difference between the total rating and the total review rating. The reviews might then show that not everything is as peachy as you would expect.
Based on continuous market analysis, we’re also confident to say that as soon as your app rating (total) drops below 4 stars, you can expect your conversion rate to dip under 20%. Much like the $99 deal, the psychology of the first number you see is a force to be reckoned with.?
Scaling up
Depending on how big your apps are or how many of them you have in your portfolio (or both!), you will also be dealing with the toughest bottleneck to handle: growth. Let’s take a few super publishers, and compare them to popular apps.?
Say, Supercell (one of the biggest game app publishers out there) and Rovio (same deal) all have tons of apps under their wing. Let’s take a look at their biggest winners:
All these apps, combined, bring tons of reviews every day. Let’s say about 130k a month combined. Some apps have a gajillion reviews coming their way during their launch—like Call of Duty Warzone, which had about 240k reviews during the launch month. Now let’s take an app everyone uses—WhatsApp. That app alone brings 300k reviews per month—a single app!?
All of these apps, with the right approach, can capitalize on this volume of reviews to get product feedback funneled, analyzed, and taken advantage of. Not to mention speedy responses bringing the brand’s image and star rating up.
To do that, you need to understand well what the users are talking about. How do you analyze feedback coming in at such a scale and act on it? AI, of course—what else excels at pattern analysis? We’ll get to that below. But first!
Hearing your user base
We’ve learned that in gaming, 50-80% of all reviews are simple 5-star reviews. It’s nice to be so praised! But even in the best of cases, 20% of total feedback is still neutral or negative sentiment. And, users do pay attention to your responses!
What’s more, we’ve witnessed that among gaming apps, more and more people use negative feedback within 5-star reviews…because they believe that’s how they can improve the odds of the developers seeing it.?
There’s a whole host of things people talk about, but some are more popular than others. “Thank you”, feature requests, UI/game improvements, monetization issues, bugs, and crashes are the most common. For a gaming app, we’d say that you must focus on two big categories:
These are the types of reviews you must focus on first. Especially because the difference in how the game is perceived by your users can be so drastically different between similar titles from the same developer.
Yes, this is when you have to jump at the opportunity to respond as quickly as possible to ensure that the first reviews are always positive or at least transform into positive. The biggest thing to be on the lookout for here is a sudden drop in rating: say, 5 stars to 2 stars, or 1 star. You have to know what brings it on—a monetization change, an unpopular feature, or something else. Analyzing this kind of trend and getting hard data on it (e.g. 30% of reviews in the last 7 days complain about the monetization change) is your best bet at resolving the issue fast (and understanding how big of a deal it is in the first place).
Indeed, we have seen many apps that did not do that at first or later apps fall behind when they are not automating their review strategy featured reviews and negative reviews must be your primary focus. If you can't respond to reviews everywhere negative or positive at least focus on these two groups.?
But is it too late to start when you’ve already got a ton of reviews? No. Based on our findings, the average benchmark for the reply rate is 20%. Good results start to really show up at the 50% reply rate benchmark, with a 100% reply rate to featured reviews. If you start now, you can still catch up.
For gaming, Board, Casino, Card, Puzzle, and Trivia are the most intense categories when it comes to the review volume, with a high satisfaction rate. People love these games! This is where you really have to respond to everything because it’s so easy. Another little gem that we have learned—up to 30% of users update their reviews if they have received a response from the developer.
The power of AI and automation
Regardless of the volume of the reviews you're dealing with, so long as it's high you have to respond to everything, ideally. Sometimes it's okay if it's just the main categories of reviews (featured, negative) if you feel like you can’t tackle more. Except you can—what you need is the tool to automate the entire process or at least parts of it
AI, especially, coupled with automation is the ultimate answer here. These are two different things, AI and automation, however. Automation is, for instance, using templates with pre-existent processes—such as auto-tagging reviews based on the content, or automatically posting a template response to a review that fits the category.
AI, on the other hand, is something you use to tackle things where it would excel—-positive reviews, feature requests, bug reports, or questions that you can answer with the help of AI.? This is the optimal use case, as the response speed is instant. What's more, you can vary the templates with the help of AI—for instance, you've got three templates and you would like to use different words every single time. AI can use different wording and make sure that your customers are engaged and react well to your responses—even though it's the same template.?
And here’s the solution to that:
Of course, AI is also helpful when it comes to tackling the whole world! By that, we mean reviews in different languages. This is probably the easiest thing for it to do, as translation is something it’s incredible at.?
Oh, and you can report reviews that violate terms & conditions that way too—automatically! Saves you the hassle, and the potentially bad mood reading what nasty people sometimes:
And let’s not forget to use AI as a broadcasting tool! For instance, if you have a large number of bug reports, all talking about the same thing, you can use AI to provide a personalized response to every single review. Or perhaps you’d like to update users on a new feature they’ve been asking for? Just go for it—carefully.
AI pattern analysis is one of its best features. It's how it works, period. It can help you instantly understand what's going on, especially when you're drowning in hundreds of thousands of reviews.
Is the issue you're dealing with just smoke? An indication of something that might happen, but hasn't yet? Maybe it won't even happen at all. Or is it a fire? An immediate problem that needs your full attention and a swift response?
AI pattern analysis cuts through the noise and tells you straight up what you're dealing with. No guesswork, no wasted time.
The hybrid model
Our top performers all use the hybrid model, which is a mix of manual responses, automation, and AI. Each of these features works best for certain scenarios. AI isn't the answer to everything, and neither is automation. Doing everything by hand will just slow you down. Here’s how we recommend splitting the approach into manual, automated, and AI.
So, with all these enhancements, you can achieve the ultimate efficiency when it comes to tackling reviews and responding to every single one of them as quickly as possible. Focus on the important ones first, using human responses where it matters, automation where it makes sense, and AI for everything else.
The results will speak for themselves:?
KPIs
The last thing you need to do is measure your team's efficiency. Agent level, manager level, whatever. You need concrete KPIs to work with.
With other tiers of community feedback, you're flying blind. But with this, you can measure reply effects, average rating, rating increases, and rating drops over time. You'll get all these numbers thrown at you.
But how do you make sense of it all? How do you know why any of this is happening? Simple. You use AI as well!
Here's what we recommend measuring with the help of AI, specifically AppFollow AI:
Agent-level:
Manager-level:
AI insights drive game development. From bugs to features to monetization, AI cuts through the noise and tells you what you need to know. With AI, you can improve your game faster. Squash bugs, reduce crashes, and resolve account issues. Use AI to measure what matters.
Afterword
If you want to make the most out of your community's feedback, here's what you do:
That's it. That's how you get stronger community relations, better customer experience, better app quality, and higher revenue. Thanks for listening. Good luck!