Under the Radar

00:00:00 ◼   ► Welcome to Under the Radar, a show about independent iOS app development.

00:00:04 ◼   ► I'm Marco Arment.

00:00:05 ◼   ► And I'm David Smith.

00:00:06 ◼   ► Under the Radar is never longer than 30 minutes, so let's get started.

00:00:09 ◼   ► So, Apple launched their search ads program last fall in the App Store, and a lot of developers,

00:00:15 ◼   ► I think us included, didn't really know what to make of it at first.

00:00:19 ◼   ► It certainly was a little bit intimidating, a little bit off-putting, the idea that we'd

00:00:24 ◼   ► have to pay to have our ads shown at the top of search results in the App Store.

00:00:30 ◼   ► And we've now had a lot of time.

00:00:31 ◼   ► I've actually been using search ads for Overcast since the day they launched, and I'm still

00:00:37 ◼   ► using them today.

00:00:38 ◼   ► It's a little bit back in the news now because yesterday Apple launched, I think by surprise

00:00:44 ◼   ► for most of us, Apple launched a new search ads basic, which is a version of search ads

00:00:51 ◼   ► that is so far US only, but they said it's going to expand to the rest of the world,

00:00:55 ◼   ► and that's where the ads show up, not who can buy them.

00:00:58 ◼   ► So if you're a developer outside the US, you can buy these ads, but they will only start

00:01:02 ◼   ► showing in the US App Store at first.

00:01:04 ◼   ► But anyway, so now they have search ads basic, and the old one is renamed search ads advanced.

00:01:09 ◼   ► I think it's worth discussing, now that we had, at least from my perspective, and from

00:01:15 ◼   ► yours, David, to the extent that you've used it, now that we have almost, I have over a

00:01:19 ◼   ► year now of search ad data, it has slowly expanded from, initially the main offering

00:01:24 ◼   ► was also US only, and then it expanded over the last maybe six months or so to a few other

00:01:31 ◼   ► different marketplaces.

00:01:32 ◼   ► Now you can have Canada, UK, Australia, Switzerland, and New Zealand, and it's a little bit cumbersome

00:01:39 ◼   ► to manage them all, honestly, but I think I've come up with a number of key insights

00:01:44 ◼   ► and best practices that might be useful to share.

00:01:46 ◼   ► What's your experience with them?

00:01:47 ◼   ► I mean, you have way more apps than I do, because all I can say is what my experience

00:01:53 ◼   ► has been with my single app, and it's a free app, so it's gonna be different than paid

00:01:57 ◼   ► upfront apps, and it's fairly popular among certain people, so that's gonna be different

00:02:01 ◼   ► from an app that's totally unknown.

00:02:03 ◼   ► Have you had more diverse experience than I have with this?

00:02:07 ◼   ► - A little bit.

00:02:08 ◼   ► So I've tried a variety of different campaigns and things with this.

00:02:15 ◼   ► I've tried it for my paid apps, for my free apps, and overall, my experience is, in aggregate,

00:02:22 ◼   ► is just very mixed, I would say.

00:02:24 ◼   ► Some of the time, it seems to work and pay off, sometimes it really doesn't, and the

00:02:31 ◼   ► thing that I struggle with the most with these is trying to, it's like, there's initially

00:02:38 ◼   ► the obvious, naive reaction to something like advertising, is viewing it as a, in a weird

00:02:47 ◼   ► way, almost like a direct investment, where I take this amount of money, I put it into

00:02:52 ◼   ► this, and then I expect to get a certain amount, hopefully a greater amount, back.

00:02:57 ◼   ► And that was straightforward, especially for my paid apps, for example, where it's very

00:03:01 ◼   ► straightforward that search ads are based on the cost per install, and if it's a paid

00:03:07 ◼   ► app, my revenue per install is 70% of whatever the purchase price is, and I could go in and

00:03:15 ◼   ► say, I'm essentially willing to break even, say, for all of these app installs, and I

00:03:21 ◼   ► try that for a little bit.

00:03:22 ◼   ► But in general, that didn't seem to work at a very high volume, anyway, where I would

00:03:32 ◼   ► get some amount of install and traffic, but it didn't seem like I was paying enough for

00:03:38 ◼   ► those installs that my ads are being shown very often, that my terms are very low volume

00:03:44 ◼   ► or whatever, but it didn't seem like it was one of these things that's like, great, I

00:03:47 ◼   ► can spend a dollar and get a dollar forty back, but I can only have three or four transactions

00:03:55 ◼   ► a day that are doing that, and so it's like, well, that's net positive, but not particularly

00:04:01 ◼   ► interesting.

00:04:02 ◼   ► So I had that experience on the paid side, and then on the free side, because it's so

00:04:08 ◼   ► hard to have this feeling of what an install is worth, and I pursued it a little bit, but

00:04:14 ◼   ► at the end, it kind of felt like I was just sort of throwing money away, or at least putting

00:04:18 ◼   ► money into something that was so hard for me to quantify that over time, I just sort

00:04:23 ◼   ► of backed away from it and decided that it's okay, and I'm probably helped a little bit

00:04:30 ◼   ► in this that my most important apps tend to have reasonably good just general search optimization

00:04:39 ◼   ► or whatever, like they show up fairly high anyway, and so that helps a lot for not needing

00:04:44 ◼   ► this necessarily, because I'm now the second, say if I was previously the top result for

00:04:51 ◼   ► a search, now with ads, I'm the second result, because the first one, the paid result is

00:04:56 ◼   ► going to always be on top, but at least being on that first screen is probably useful enough,

00:05:02 ◼   ► but I don't know, I go back and forth on how useful they are, and then in the back

00:05:07 ◼   ► of my mind, I kind of feel like I'm probably missing, in some ways, the fundamental realities

00:05:16 ◼   ► of advertising, where I'm sure if you were just a typical, a large brand advertiser, your

00:05:24 ◼   ► goal is not necessarily direct, net positive cash flow from each advertising buy, it's

00:05:30 ◼   ► a much more broad something that you're taking, you have a marketing budget whose purpose

00:05:37 ◼   ► is to just increase brand awareness and to increase usage of the app as this kind of

00:05:42 ◼   ► more general concept, in some ways, even kind of per our conversation a couple episodes

00:05:47 ◼   ► ago where I was talking about why I'm making workouts plus plus free, where it's like,

00:05:51 ◼   ► there's this general value to being well known beyond the immediate return on that

00:06:00 ◼   ► investment that you see within a few days, and so I go back and forth if I'm just doing

00:06:07 ◼   ► it wrong or not, but my experience has been kind of mixed, and mostly I've kind of backed

00:06:12 ◼   ► away with it. I'm going to try the new basics approach, just to see if there's anything

00:06:17 ◼   ► interesting there, and Apple's giving us $100 free credit, so great, sure, I'll take

00:06:23 ◼   ► that $100 and spend it, but overall I'm still kind of mixed about it.

00:06:28 ◼   ► - I think a lot of developers didn't give it a fair shot when it first launched because

00:06:33 ◼   ► we felt kind of put off by it or kind of like it was a trick or a negative thing for Apple

00:06:40 ◼   ► to be doing. I think we have, there's kind of two parts of that argument that are worth,

00:06:46 ◼   ► I think, internalizing now. Number one is that, well, it's done. It's clearly here to

00:06:51 ◼   ► stay there, not only keeping it but expanding it, so I think that that ship has sailed,

00:06:56 ◼   ► much like initially when we were all mad that App Review was a thing when the App Store

00:07:00 ◼   ► first launched. That ship has sailed, and so has this one, so you might as well get

00:07:05 ◼   ► on board with at least being okay with it existing. Even if you decide not to participate

00:07:09 ◼   ► in it, you should at least realize that this is now a factor in the App Store that's probably

00:07:13 ◼   ► here to stay. And number two, while it does feel pretty bad to pay to reach people who

00:07:22 ◼   ► were already seeking you out, so if somebody searches the App Store for overcast, someone

00:07:28 ◼   ► else's ad is gonna be up top there if they outbid me on that keyword that day. Probably

00:07:32 ◼   ► audible, but someone else's ad is gonna be up there, and it's probably gonna be a bigger

00:07:37 ◼   ► company than me, and I'm gonna feel kind of like that's a little bit unfair, but the reality

00:07:42 ◼   ► is that could also be a smaller company than me. That could also be an indie startup who

00:07:46 ◼   ► wants to get into the same market I'm in and wants to have a chance at paying to take away

00:07:52 ◼   ► some of the searches from me, and I'm doing the same thing. I'm bidding on generic terms

00:07:58 ◼   ► like podcast, but I'm also bidding on terms like audible, because I'm trying to get some

00:08:04 ◼   ► of that same audience, and search ads largely have worked for me, and so it is the kind

00:08:11 ◼   ► of thing where you can still promote an app for free through traditional PR channels and

00:08:18 ◼   ► marketing channels, and you probably should, but this in a way gives an opportunity for

00:08:23 ◼   ► small apps to get seen in some way and get started. And yes, of course, it does take

00:08:30 ◼   ► money, and therefore that will price out a lot of developers, but if you manage it responsibly

00:08:37 ◼   ► and conservatively up front, it doesn't have to take that much money, and if you look out

00:08:43 ◼   ► with your terms and if you control it well enough, you can actually make it profitable

00:08:47 ◼   ► per user. So this leads into I think the first major challenge of using search ads, which

00:08:54 ◼   ► is you have to know what a user is worth to you, and that's very hard to do if your app

00:08:59 ◼   ► is not paid up front, if there's any more complex business model than paid up front,

00:09:05 ◼   ► and honestly I don't think paid up front apps work that well in search ads, because you

00:09:09 ◼   ► don't pay per install, you pay per tap on the ad. So the cost per install varies depending

00:09:17 ◼   ► on how many people tap your ad and then decide not to buy it or not to download it, and so

00:09:22 ◼   ► that can obviously vary based on things like your screenshot, your description, your ratings,

00:09:27 ◼   ► but also I think if you have a price there, an up front price, I would guess that causes

00:09:34 ◼   ► much lower buy through rates or conversion rates from the tap to the purchase than a

00:09:41 ◼   ► free app would, and so that's going to dramatically affect your cost right there. So if you're

00:09:47 ◼   ► paid up front, I think it's easier to calculate what an install is worth to you, but it's

00:09:53 ◼   ► also harder to get those installs. So I'm going to now focus for the rest of my discussion

00:09:56 ◼   ► of this on free versions, on free apps up front, because that's what I know, that's

00:10:01 ◼   ► what I have, and I think that's going to be mostly what search ads are good for, because

00:10:06 ◼   ► there's just that one less big barrier to entry for people who are learning about it

00:10:10 ◼   ► for the first time through that ad to then click there and download it. So the question

00:10:15 ◼   ► then becomes how do you value a user if your app is free up front? So if you have a scheme

00:10:21 ◼   ► like I do, which I think at least part of this is common if not the whole thing, where

00:10:26 ◼   ► I have ads in the app and I get paid for those ads, and I have an in-app purchase that can

00:10:31 ◼   ► be optionally bought to remove the ads and add some premium stuff, but mostly it's

00:10:35 ◼   ► removing the ads. And I think this is common enough that I can tell you some lessons I've

00:10:40 ◼   ► learned from this. Number one, when you're trying to figure out what each user is worth

00:10:45 ◼   ► to you, you have to make some assumptions that might be uncomfortable or cruel or revealing.

00:10:52 ◼   ► For instance, what percentage of people who download your app will actually ever launch

00:10:56 ◼   ► it? It's not 100 percent news-fledged. If you weren't aware of this yet, and if you

00:11:03 ◼   ► already have some analytics in your app, you probably already know this kind of thing,

00:11:06 ◼   ► but yeah, it's not 100 percent. You might get two-thirds or 80 percent or something

00:11:12 ◼   ► like that, or half. It might not be a very kind ratio of people who actually ever launch

00:11:18 ◼   ► the app after downloading it. And then after that, you have to then work through, like,

00:11:24 ◼   ► if you're valuing the user based on the likelihood that they're going to buy an in-app

00:11:28 ◼   ► purchase or whether they're going to see your ads or follow through on any other kind

00:11:32 ◼   ► of monetization, what other barriers are there between getting the app and running the app

00:11:37 ◼   ► and getting to that point where they're making you money? Do they have to create an

00:11:40 ◼   ► account? Do they have to sign in? Do they have to enter some data? And all of those

00:11:45 ◼   ► steps are going to reduce that percentage even further of how many of those downloaders

00:11:49 ◼   ► actually make it to the part of the app that makes you money. Not to mention the fact

00:11:54 ◼   ► that once they get, you know, assuming they're actually using it, then you have the

00:11:58 ◼   ► question of what percentage of actual users will then buy the in-app purchase? Or if it's

00:12:04 ◼   ► ad-based, what percentage of them will see the ads? And then ads have another massive

00:12:09 ◼   ► complex guess that you have to make or data you have to collect, which is how long will

00:12:15 ◼   ► each person be using my app? You know, will the average user use it for a day, a year,

00:12:20 ◼   ► ten years? So those numbers can dramatically change what you think a user is worth.

00:12:27 ◼   ► And I would urge you that if you don't know this number, you know, if you are making

00:12:32 ◼   ► estimates and guesses on a lot of these numbers, which you might have to, you know,

00:12:36 ◼   ► if you don't have really good, really creepy analytics, like I don't, you have to

00:12:40 ◼   ► guess on a lot of these numbers or just wait and see the results. And for me, I found

00:12:46 ◼   ► that guess very conservatively on these numbers. Like every step of the way that might

00:12:51 ◼   ► reduce the amount of users that get to the next stage, cut it in half. Like really,

00:12:56 ◼   ► like be very aggressive. Like just assume half the downloads will result in launches.

00:13:01 ◼   ► Assume that half of the launches will result in accounts being created or, you know,

00:13:05 ◼   ► signups happening or proceeding to the next step. And so these numbers start to,

00:13:10 ◼   ► start to, you know, divide out and math starts catching up to you pretty soon.

00:13:14 ◼   ► And so you have to be very, very careful when making any kind of assumption about

00:13:19 ◼   ► what a user is worth to you long term. And keep in mind that long term might really

00:13:24 ◼   ► be a day. For the majority of your users, your average duration of how long a user

00:13:31 ◼   ► sticks with your app might be a day or a week or something pretty short. So you can't,

00:13:35 ◼   ► for instance, assume like that you have like six months or a year of ad revenue to

00:13:40 ◼   ► attribute to that download because you probably don't.

00:13:42 ◼   ► The nice thing though is we do get a lot of those numbers now from App Analytics and

00:13:46 ◼   ► iTunes Connect. That's true, yeah. Like at least we do have a better starting place

00:13:50 ◼   ► than, like you don't necessarily need the creepy analytics for a lot of that kind of

00:13:55 ◼   ► stuff. Because I'm just thinking like with the things that we get from there are like

00:13:59 ◼   ► a retention percent, which is a really useful measure for being able to get a sense of like,

00:14:05 ◼   ► you know, so once someone has downloaded the app, what is it like, you know, what percentage

00:14:10 ◼   ► of those people will open it the next day? And what percent will open it a week later,

00:14:15 ◼   ► two weeks later, three weeks later? I think they go up to 28 days. And so at that point

00:14:19 ◼   ► you get a sense of like what your retention curve looks like there. And so at least you

00:14:24 ◼   ► can kind of, you know, infer that, you know, if your retention has kind of flattened out

00:14:30 ◼   ► by the time you're a month out, like you've probably gotten that user. If you're still,

00:14:34 ◼   ► you know, whatever it is, say you have 10% retention after a month, it's like you probably

00:14:38 ◼   ► have them for that, but even if you just capped it at a month and tried to work out, you know,

00:14:44 ◼   ► how many sessions does the, what will the user have, you know, if they, if 10% of them

00:14:51 ◼   ► are there at the end of a month and maybe 20% of them are there after a week, you know,

00:14:56 ◼   ► you can kind of do a little bit of math. But yeah, it's, it's, it is a really tricky thing

00:15:00 ◼   ► and I think being conservative here makes, is by far the better way to err on because

00:15:05 ◼   ► it's far more important for it to just be a reasonable number than it is for it to be

00:15:09 ◼   ► an exact number.

00:15:11 ◼   ► Exactly. And all this, by the way, this is all making a pretty big assumption, all this

00:15:16 ◼   ► advice so far that you don't want to lose money on each installation, but there are

00:15:21 ◼   ► types of businesses and apps and situations where that might not be the case. You might

00:15:25 ◼   ► actually want to lose money on or you might be willing to lose money on every installation

00:15:31 ◼   ► if maybe it's only a little bit of money or if you're, if you have a plan to make it up

00:15:36 ◼   ► down the road or if you really want to just focus on growth of user base right now or

00:15:41 ◼   ► if you haven't monetized yet, if you're like, you know, a startup that's pre monetization

00:15:44 ◼   ► strategy and you just want growth, you know, so you might have reasons where you don't

00:15:48 ◼   ► care about whether you're going to make up the, you know, make it up properly, you know,

00:15:52 ◼   ► on the other side or not. Or you just might be betting for your future and betting, you

00:15:55 ◼   ► know what, I can take a small loss on each user right now, but some percentage of these

00:16:00 ◼   ► users will tell their friends and spread the app and cause my organic installations to

00:16:04 ◼   ► go up. So it's, it's very, very vague and based on a lot of estimates and assumptions

00:16:10 ◼   ► that this is true of all advertising on all sides for everything. So all this is to say,

00:16:17 ◼   ► like, you know, try to quantify some of these numbers, try to make good estimates, try to

00:16:21 ◼   ► have as much data as you can to, to make good decisions here. But your decisions might not

00:16:26 ◼   ► have the same, you know, needs as someone else's. Anyway, before I get into specifics

00:16:31 ◼   ► of what I do in the app store with app search ads, we are brought to you this week by Eero.

00:16:35 ◼   ► This is another kind of ad by Eero. Never think about Wi-Fi again. Eero has created

00:16:39 ◼   ► the dream Wi-Fi setup, a fast, reliable connection throughout your house and even into the backyard

00:16:44 ◼   ► if you want. And now is the best time to get on board with Eero as they've just released

00:16:48 ◼   ► their new super slick second generation devices. This brand new tri-band second generation

00:16:53 ◼   ► model, along with the new Eero beacon allows you to build a Wi-Fi system that is perfectly

00:16:57 ◼   ► tailored to your home. The new second generation Eero includes a third 5 gigahertz radio, making

00:17:02 ◼   ► it twice as fast as before. This lets you do more than ever, whether it's your Wi-Fi

00:17:06 ◼   ► needs. Eero has the power to blanket your entire home in fast, reliable Wi-Fi. The base

00:17:12 ◼   ► unit sits flat on any surface. Just plug it into the wall with the included power adapter

00:17:16 ◼   ► and you're ready to connect to your Eero either with ethernet or wirelessly. And the new Eero

00:17:21 ◼   ► also includes a thread radio, which lets you connect to low power devices such as locks,

00:17:25 ◼   ► doorbells and more. And the new little satellite units are now called Eero beacons. And this

00:17:30 ◼   ► is new as well. All you do is plug it into a wall. It looks like kind of like a large

00:17:34 ◼   ► nightlight almost. It sits flush right on the outlet, right against the wall. There's

00:17:39 ◼   ► no wire needed. Plugs right in and it even includes a nightlight on the bottom to help

00:17:43 ◼   ► you out. So you can plug it in anywhere you have an outlet. You don't need like a table

00:17:46 ◼   ► or a surface or a cabinet to hide it in. These are wonderful. I love these new beacons. And

00:17:51 ◼   ► the new Eero app lets you manage your network from the palm of your hand. And you can also

00:17:55 ◼   ► easily create and share a guest network too if you want that. And if you need any help,

00:17:59 ◼   ► Eero's customer support is amazing. You can call them and get a Wi-Fi expert on the phone

00:18:03 ◼   ► in just 30 seconds. I've used Eero and it is by far, you know, I've used other multipoint

00:18:08 ◼   ► access point things before. And first of all, you definitely want multipoint Wi-Fi. But

00:18:12 ◼   ► Eero is by far the easiest and fastest to set up I've ever seen. It's wonderful. And

00:18:18 ◼   ► the fact that you don't have to run wires between the base stations is also just incredible.

00:18:21 ◼   ► So definitely recommend it. It's very, very easy to use and performs very well. So the

00:18:26 ◼   ► new Eero starts at just $399 for one second generation Eero base station and two beacons.

00:18:31 ◼   ► And that is everything you will need for almost any situation. You can add more if you want,

00:18:36 ◼   ► but you probably won't need to. Listeners of this show can get free overnight shipping

00:18:39 ◼   ► to the US or Canada when you head to Eero.com. That's Eero.com and use promo code radar.

00:18:44 ◼   ► That's Eero.com promo code radar for free overnight shipping. Thank you to Eero for

00:18:49 ◼   ► supporting this show. So I wanted to get into a little bit more specifics of how to optimize

00:18:54 ◼   ► search ads on the App Store. Now that you know roughly what you want to spend per installation,

00:19:00 ◼   ► because you figured that out during my ad read I guess, I don't know, you figure it

00:19:03 ◼   ► out sometimes. We'll assume you have that number.

00:19:05 ◼   ► We did the math.

00:19:06 ◼   ► Yeah, yeah, roughly what you were able to spend per installation. You know, it's worth

00:19:10 ◼   ► now learning kind of how the system works. So the main thing with search ads is that

00:19:16 ◼   ► it's dependent very, very heavily on Apple's own App Store relevancy engine. And as we

00:19:23 ◼   ► know from App Store search, it's not that good. It works on a fairly coarse level, but

00:19:30 ◼   ► there's a lot of mismatches, a lot of misjudgment of relevancy. Ranking isn't so great. And

00:19:38 ◼   ► this applies therefore to the search ads matching algorithm as well. So you need to do some

00:19:43 ◼   ► tailoring and you need to know that going into it that the relevancy matching is not

00:19:48 ◼   ► strong. And this, you know, therefore this could work for or against you. You know, your

00:19:53 ◼   ► app might rank weirdly high or weirdly low. One thing Apple has done, which I think is

00:19:58 ◼   ► a smart move, is, and Phil announced this on stage when they announced search ads or

00:20:02 ◼   ► whatever that was, but the idea is that even if you bid a high enough price, if Apple's

00:20:11 ◼   ► algorithm deems your app not relevant to a query, it will not include it no matter what,

00:20:15 ◼   ► no matter what you bid. And this is good because this prevents something like Clash of Clans

00:20:19 ◼   ► bidding on the term podcast because it might be becoming a high volume term and they want

00:20:24 ◼   ► to get in on that even though their app is not relevant to podcast at all. So that's

00:20:28 ◼   ► actually a very good prohibition. The downside of that is that if the search algorithm guesses

00:20:33 ◼   ► wrong on whether your app is relevant, you could be on the wrong side of that. So something

00:20:37 ◼   ► to keep in mind, there's not much you can do about that unfortunately. But that is nice

00:20:42 ◼   ► to know that that's a thing. When you are picking your keywords, I have found that it's

00:20:48 ◼   ► best to mostly just do search match. That if you do too much keyword selection, you

00:20:53 ◼   ► can get weird results. If you do search match, you're kind of letting Apple do it for you.

00:20:58 ◼   ► And this is actually, so when they launched yesterday, they launched search ads basic,

00:21:02 ◼   ► that's all it is as far as I can tell. It seems like it is a search match only ad account.

00:21:08 ◼   ► Is that right? >> Yeah, you just give it a budget and a maximum

00:21:12 ◼   ► cost per install. Like those are the only two numbers you put in and then it just does

00:21:17 ◼   ► the rest. It'll do, it'll try and find you whatever the best opportunities are and, you

00:21:24 ◼   ► know, sort of do its own thing. But you have no visibility into what those terms are or

00:21:31 ◼   ► anything like that. It's just you just give it, basically you just tell it how much money

00:21:35 ◼   ► overall are you going to give to this campaign and then how much, what's the maximum amount

00:21:39 ◼   ► that you're willing to pay for an install and then it just does the rest for you.

00:21:43 ◼   ► >> The only thing about that that I will say, and this is why I'm a little bit worried about

00:21:47 ◼   ► search ads basic, is that one of the great things about the search ads implementation

00:21:52 ◼   ► in the advanced mode, that used to be the only mode until yesterday, is that you can

00:21:56 ◼   ► go in, so run it for like a few weeks, you know, with the, you know, or whatever time

00:22:00 ◼   ► you want, you know, run it for a little while and then you can go in to the reporting area

00:22:04 ◼   ► and you can see if you go to, if you go within the ad group, you go to a tab called search

00:22:09 ◼   ► terms and it will list the terms that people actually use to search your app and you can

00:22:15 ◼   ► rank that by things like conversion rate or average cost per installation, which they,

00:22:21 ◼   ► it's CPA, cost per action in the advertising world, but it's cost per installation. And

00:22:25 ◼   ► so you can sort and rank the keywords people are actually using that Apple is matching

00:22:30 ◼   ► to you, you know, for you and you can see like, you know, certain ones are going to

00:22:35 ◼   ► be better than others in terms of what you're actually spending for those users, what percentage

00:22:38 ◼   ► of them are actually installing your app, et cetera. And if you see the keywords, like,

00:22:44 ◼   ► you know, I'll tell you, like some of mine, like I start out as, so if I sort by spend,

00:22:49 ◼   ► which is the amount of money you have spent on it for that time interval, my first one

00:22:53 ◼   ► is podcast, which is good, that is my most expensive one, but that's also my highest

00:22:56 ◼   ► volume one. And I'm bidding high on that because I want, I want to get those people who are

00:23:00 ◼   ► searching for podcast. Then there's low volume terms, which is kind of their bucket for everything

00:23:05 ◼   ► else that's too small to matter. Then I have things like podcast app for iPhone. Great,

00:23:10 ◼   ► that's what I want. The next one down is FM transmitter. Well, I'm not sure I want that

00:23:16 ◼   ► one. That's not that relevant. And sure enough, FM transmitter has a way lower conversion

00:23:22 ◼   ► rate than the other ones. So I have to decide, like, is this actually, do I actually want

00:23:27 ◼   ► to be paying for people who search for FM transmitter? And if you don't, one thing you

00:23:32 ◼   ► can do in search ads advanced is you can go through this list, you can check off whichever

00:23:37 ◼   ► ones you don't think are worth bidding on that are not relevant enough. And you can

00:23:42 ◼   ► say add as negative keywords. And if you say exact match, which you probably should do,

00:23:47 ◼   ► then it will not put you on those bids. And that I have found is a necessary part. If

00:23:54 ◼   ► you're going to do search match, if you're not going to just bid on exact keywords, if

00:23:57 ◼   ► you're going to do any search matching at all on the app store, you must go in and add

00:24:02 ◼   ► as negative keywords the things that are not relevant to your app that are costing you

00:24:06 ◼   ► a lot of money. And my, because, you know, because Apple's relevancy engine isn't

00:24:11 ◼   ► that great, you get a lot of these. And some of them you actually might want. So like,

00:24:16 ◼   ► one of the very common ones that I get is things like music player without Wi-Fi, free

00:24:23 ◼   ► music downloader, offline music player, FM radio offline. And I'm like, well, actually,

00:24:29 ◼   ► those are people who want to listen to stuff offline, some of which is radio. So I think

00:24:35 ◼   ► I actually might want to keep bidding on a lot of those. Whereas one of them, Overwatch,

00:24:40 ◼   ► I don't want to bid on because that's a video game that has nothing to do with my

00:24:43 ◼   ► app and it's being matched by some kind of linguistic coincidence. And I definitely

00:24:48 ◼   ► don't want to bid on Overwatch terms. So I definitely add that as a negative keyword.

00:24:52 ◼   ► But again, keep in mind that you're dealing with a fairly low sophistication search

00:24:58 ◼   ► algorithm and relevancy algorithm. So be careful. One of the reasons I said only add

00:25:03 ◼   ► negative keywords as exact match is that earlier in the summer I had a problem where I

00:25:08 ◼   ► added too many things to negative keywords as like regular like broad match, like

00:25:13 ◼   ► relevancy match, thinking that was the right thing to do. And I started getting matched

00:25:17 ◼   ► on very, very few things. All of my numbers just plummeted. And I contacted the search

00:25:21 ◼   ► ads team and they're very nice. There's actual humans there who will like help you

00:25:24 ◼   ► out. And they ran amongst numbers and they recommended that I like remove some of

00:25:27 ◼   ► those and just switch to a more relevancy based approach, which I did. And so I would

00:25:31 ◼   ► recommend only using negative keywords in exact match mode to avoid that situation.

00:25:35 ◼   ► Anyway, I end up coming out for a long time until the last couple of months. I have

00:25:44 ◼   ► come out ahead in what I believe a user to my app is worth versus what I'm paying per

00:25:50 ◼   ► install from the search ads. And the search ads have made up approximately 10 to 15

00:25:56 ◼   ► percent of my weekly installs. And so that's not bad.

00:26:01 ◼   ► - That's a lot.

00:26:02 ◼   ► - Yeah, exactly. And that's 10 to 15 percent of my installs per week. So that

00:26:07 ◼   ► increases over time. And like those users are sticking with the app, I think. I

00:26:12 ◼   ► don't know. There actually is an API somewhere in iOS. There's an API where you

00:26:17 ◼   ► can attribute installations to your app to whether they came from search ads or not.

00:26:23 ◼   ► And I think even what keywords they came from. And then you can tell in your app

00:26:26 ◼   ► analytics like what those people are worth to you. I have not done that. That seems a

00:26:30 ◼   ► little bit too fiddly for me. I don't think I really care that deeply. And it's a

00:26:34 ◼   ► little bit creepy, I think. So I haven't done that. But you can do that. And you

00:26:38 ◼   ► can get into deeper measurements. But I have found for the most part this has been

00:26:43 ◼   ► worth it for me to do for a while. Unfortunately, over the last couple of

00:26:47 ◼   ► months, the podcast related keywords have really heated up. Thanks, Audible.

00:26:52 ◼   ► From like big companies, thanks, Audible. Who don't seem to have a limit to their

00:26:57 ◼   ► budget, thanks, Audible. And so I'm losing a lot more bids. And I've had to increase

00:27:02 ◼   ► my bids if I want to keep winning them. And I'm now pushing to the point where I'm

00:27:06 ◼   ► losing a little bit of money on each install that comes from a search ad. And

00:27:10 ◼   ► so, so far I've decided, well, I kind of want to wait and see and watch how these

00:27:15 ◼   ► numbers go. I kind of don't want to give up that 15%. So maybe I'm okay losing a

00:27:19 ◼   ► bit of money on those people because then the other 85% of my users will make up

00:27:23 ◼   ► for it. So I'm not entirely sure what I'm going to do about this long term yet. But

00:27:27 ◼   ► the only reason I've been able to keep costs under control at all is because I'm

00:27:32 ◼   ► going in and negative key wording the searches that are providing me with really

00:27:36 ◼   ► bad conversion rates and really high cost per install. And because search ads basic

00:27:41 ◼   ► does not offer that ability, I do hesitate to recommend it because I don't think

00:27:47 ◼   ► you're going to get a lot of control over that kind of thing. And Apple's search

00:27:52 ◼   ► relevancy engine is not good enough that I would trust it to do it without any

00:27:56 ◼   ► intervention from me. What do you think?

00:27:59 ◼   ► - Yeah, I mean, I think, like I'm intrigued to go, to use the basic system, but

00:28:06 ◼   ► mostly insofar as it's an, like because it's the only input I give is what the

00:28:11 ◼   ► maximum amount I'm willing to pay for installation is, I would just put that

00:28:15 ◼   ► number low enough that I feel comfortable with it and understand that, like, your

00:28:21 ◼   ► approach of the highly tuned, like going in there and looking at keyword by

00:28:26 ◼   ► keyword, like dialing it in, you are going to end up with a better result for the

00:28:32 ◼   ► same, like, cost per install, but it takes a lot of work and thoughtfulness and so

00:28:39 ◼   ► on versus the approach of just saying it's like, I'll put that number low and if

00:28:43 ◼   ► the number is so low that it's not getting high volume, okay, like, that's fine. I

00:28:47 ◼   ► didn't have to do much for it, but if I am getting installs from that, great. And

00:28:52 ◼   ► so I kind of view it as a, what you're doing makes perfect sense and I think it's

00:28:57 ◼   ► the ideal way if this is something that you want to spend time on, if you want to

00:29:01 ◼   ► sort of have it be a focus. If not, if you're sort of like me and you're just

00:29:05 ◼   ► like, this is an interesting kind of more amusement, it's a really convenient way

00:29:10 ◼   ► to just kind of like set it up, set it at a number that you're comfortable with and

00:29:13 ◼   ► then just see what happens. And that's how I expect to proceed from here.

00:29:17 ◼   ► Well, best of luck. I do, I strongly encourage everyone to use that free

00:29:21 ◼   ► credit. Look, you might as well. They're giving you a free credit, so use it and

00:29:25 ◼   ► see how it goes. If you want to be strategic about it, wait until after

00:29:29 ◼   ► everyone else uses theirs, then use yours. Then the prices will be lower. Anyway,

00:29:33 ◼   ► we're out of time this week. I hope this was helpful to you. Thanks for listening,

00:29:36 ◼   ► thanks for listening everybody, and we will talk to you next week. Bye.

00:29:40 ◼   ► All right.

00:29:40 ◼   ► [BLANK_AUDIO]