Keyword research at the centre of your strategy

Why rival tracking is a measurement instrument

Rival tracking is not curiosity, it is measurement. To see the effect of your own move you have to know what the phrases you did not touch, and the apps moving independently of you, did that week.

Rival tracking is usually explained by competitive instinct: let us see what they are doing. That is the least important part of it. The real reason is this: your rivals' movement is the noise in your own measurement.

You made a metadata move and a week later the phrase dropped two places. Two explanations fit. Either the move was wrong, or two rivals passed you that week. Without separating them you cannot say anything about the move - and if you say the wrong thing you will get the next one wrong too.

The same holds upward. If you celebrate a rise and repeat the move, you may not notice that a rival pulled their app. The gain was not yours and it will not repeat.

So a rival list is not an interest list, it is a control group. The apps you did not touch, competing with you for the same phrases, are the only source that tells you what the market did that week.

The practical consequence: ASO done without rival tracking is ASO without measurement. Moves are made, ranks move, nobody knows what did what, and three months later no learning is left.

Why measuring them costs almost nothing

Rival tracking is assumed to be expensive: a separate measurement, a separate request, a separate cost per rival. Set up correctly it is not.

A rank measurement works like this: a phrase is searched in a country and the first 200 results are read. Your app is in that list and so are your rivals'. One search returns the rank of every app tracking that phrase.

So measurement has to run per phrase, not per row. Adding a rival does not raise the cost; what raises it is the number of phrases and the number of countries.

A method doing the opposite - one search per app - ties cost to the number of apps tracked. That arithmetic does not hold: the requests per second to Apple are capped, and as the rival count grows the measurement stops finishing daily.

The practical consequence: you can build the rival list without fear. What costs is phrases and countries.

There is one exception: reading a rival's metadata or reviews needs separate requests. Rank comes free; those two do not. That is why metadata tracking runs weekly and review mining monthly.

What rival metadata teaches

A rival's app name, subtitle and description are public. Reading them is not for copying; it is for understanding positioning.

The first thing to read is the phrase in the app name. A rival with a generic phrase there is taking it seriously, and competing for it will be expensive. A rival whose name is only a brand is holding generic phrases in the subtitle and keyword field, which means more flexibility there.

The second is the shape of the subtitle. A sentence, or a word list? Subtitles that are word lists are optimised for indexing and usually convert worse - which is ground you can take with a readable one.

The third is the first two lines of the description. The user arriving from search sees those first. If every rival's first two lines say the same thing, saying something different is an advantage on its own.

The fourth is change over time. Record a rival's metadata weekly and you see when they changed what. Seeing their metadata change in the week they rose on a phrase confirms your hypothesis.

What you should not try to read: the keyword field. Apple keeps it private to the developer. Tools claiming to show it are inferring backwards from rank data, and inference is estimation.

The limits of rival tracking

Rival tracking says a lot and it does not say everything. Knowing the limits keeps you from deciding on ground that was never measured.

Download count is invisible. Apple does not give it to third parties. The estimates in circulation are computed backwards from rank and rating count, and their error bars are wide and category-dependent. If you see a number saying how many times larger a rival is than you, it is an estimate.

Revenue is invisible. In-app purchase and subscription revenue is entirely closed. You can see the price, not how many paid it.

The keyword field is invisible. Which words a rival wrote is private to the developer. Backwards inference is possible, but presenting an inference as a measurement makes people decide wrongly.

Ad spend is invisible. A rival appearing at the top of search through an ad is not showing you their organic rank. Organic measurement is done separately, but how much they spend cannot be measured.

All four look fillable with an estimate. We do not fill them, because the moment an estimate goes on a screen the reader takes it for a measurement. Leaving a cell blank is better than filling it wrongly.

• Downloads and revenue are closed; the numbers offering them are estimates.
• A rival's keyword field is invisible; tools showing it are inferring.
• A top position from an ad is not an organic rank.
• Leave the unmeasurable blank; an estimate presented as a measurement breaks decisions.

Frequently asked

How many rivals should I track?

Five to eight. More costs nothing in rank measurement but it costs readability; nobody looks at a twenty-app list weekly, and a list nobody looks at is not a control group.

Does adding a rival raise my measurement cost?

On the rank side no: every app tracking the same phrase is read out of one search. On the metadata and review side yes, because those need a separate request per app.

Can I see a rival's keyword field?

No. Apple keeps it private to the developer. Tools claiming to show it infer backwards from rank data; that is an estimate, and unlabelled it makes people decide wrongly.

Can I use phrases from rival reviews directly?

No, they have to be measured first. Coming from a review is not a privilege; the phrase passes the same elimination. And do not take a phrase describing something your app does not actually do.

My rival is above me on every phrase. What should I do?

Look at the strength threshold first. If their rating count is far above yours this is not a gap metadata will close; the work has moved to the product. If it is close, look at the metadata difference: which phrases are in their name and subtitle that are not in yours?