The resource centre

What a summary is for

Summaries are not there to teach you something new; they are there to put what you know in order. Ten minutes to read, a few sentences left behind, and a stronger hand walking into a meeting.

Long texts have a problem: a week after reading one, not much is left. And what is left is usually the text's most striking sentence rather than its most useful one. Summaries were written to close that gap.

A summary is not a shrunken version of a long text. A long text explains something; a summary gives the same thing in its decision-moment form. The metadata guide explains the causes of character waste one by one; the metadata summary says "check these five things, this many characters come back". One teaches, the other reminds.

So a summary only helps if you have learned the subject once. Reading a summary of a subject you have never read is like listening to directions without having seen the map: the sentences are clear but where you are going is not.

The second use of a summary is inside a team. Reading a guide end to end is one person's job; but that person has to explain their decision to three others. The summary is that explanation, ready-made.

The shared skeleton

Every summary here is made of the same four parts. Because the skeleton is fixed, whichever one you open you find what you are looking for in the same place.

A one-sentence claimThe sentence the whole text reduces to - "the keywords you rank 11-30 for are an app's cheapest growth". That sentence is not a slogan but a measurable claim, and the rest of the text supports it.
What we measuredThe measurements the claim rests on and their limits. What we could not measure is written here too, because the boundary of a claim matters as much as the claim.
What to do this weekThe summary's single action item. One, because a summary with two action items is no longer a summary. Usually a check or a list that can be done in an hour.
Where to look nextWhich guide or report to go to if you want to go deeper. A summary is a door; a long text stands behind it.

Which summary before which meeting

The moment a summary is most useful is right before a meeting where something will be decided. The pairing below shows which summary strengthens your hand in which meeting.

Meeting	The summary to read
Release planning - is the metadata changing?	The metadata summary. It gives how many characters are wasted and what one move costs, on a single page.
Opening a new country	The country priority summary. It shows how the same phrase differs between storefronts and where the opening is.
A rival got ahead - what do we do?	The rival tracking summary. It sets out how to tell a rival's move apart from the market's.
Is the time we spend on this worth it?	The return summary. It says what has a measurable counterpart and what does not.
Should we switch tools?	The tool comparison summary. It shows which tool takes which data from where, and who is estimating what.

The six sentences that recur

Every summary comes back to the same six sentences. They are the compressed form of how rankcusp thinks about measurement, and you will meet them whichever text you read.

An unmeasured value is not zero. Zero is a reading and people decide on readings; a short dash stands in place of the unknown.
Relevance and rank are different things. If Apple has associated you with a phrase the job is half done; if it has not, the job has not started.
The first page is where a search ends. The gap between eleventh and tenth is larger than the gap between thirtieth and twentieth.
A move's effect can only be measured against a control group. If the phrases you did not touch moved the same way, your move did not produce that effect.
Phrases fold by country. In Turkish and Azerbaijani a capital I becomes a dotless ı; apply the rule everywhere and you corrupt the phrase.
Rival reviews are the keyword source no tool looks at, because users describe a product in their own words.

The limit of reading summaries

Summaries are fast, which also makes them dangerous. Trying to learn a subject from a summary means missing its most important part: the exceptions.

Most rules in ASO have one. "Do not repeat a word from the title in the keyword field" is a correct rule - but where the app name is written differently in different countries, the repeat may be necessary. "Change one field at a time" is a correct rule - but if the app name is changing entirely, the subtitle has to follow it. A summary cannot carry those exceptions, because carrying them would stop it being a summary.

The second limit is that a summary assumes context. "Pick the three most popular phrases" means something to somebody who knows what popularity is and why it is relative. Somebody who does not takes it for absolute search volume and picks the wrong phrase.

The third is timing. A summary assumes a world where measurement has already happened. "Pull the band and order it" means nothing when you hold no measurements; measuring comes first, and that is a guide's job.

In short: use summaries in the areas you know, and go to the guide in the areas you do not. Reading them in the wrong order wastes more time than not reading them at all.

Writing one for your own team

We suggest rewriting these summaries with your own app's data. A general summary stays general; one written with your own numbers produces a decision.

Build the one sentence from your own data"The 11-30 band is cheap" is general. "Fourteen of our phrases are at 11-30 and six of them are in the top quartile for popularity" is specific. The second gets argued with; the first gets approved and forgotten.
Write the measurement dateA number written without a date is taken as true forever. Put the measurement date at the top so that read two months later it is understood to need re-measuring.
Put what you could not measure on its own line"We could not measure the popularity of these five phrases" may be the most valuable line in the summary. It stops the team deciding about those phrases - which they should not.
End with one action itemOne job at the end of the summary, with a named owner. A two-item summary ends with neither of the two being done.

The measurements behind the summaries

A summary is short but there is a long measurement chain behind it. Every number in a summary comes from somewhere in the chain below.

Search results. A phrase is searched in a country and the first 200 are read. Rank, density and title share come from there. Measurement runs once a day; measuring more often raises the noise, not the information.

Search suggestions. The relative value for how much a phrase is searched is derived from these. Where no suggestion returns, popularity stays blank - not zero, blank.

App metadata. The app name, subtitle and description are read; phrase extraction and the metadata audit both feed from there. Character waste is not calculated, it is counted.

Ratings. Score and count are recorded over time. It does not carry rank directly, but a drop after a release is the fastest signal that something in that release broke.

Rival reviews. Review text is scanned, sentences are broken up, and phrases that look like searches are extracted. Ones containing first- and second-person pronouns are dropped, because people write "it really helped me" and nobody searches that.

What the five share is that they are all open. None comes from a private arrangement or a closed data set; you can verify every one of them by hand. A number that cannot be verified does not go into a summary.

The decisions summaries usually produce

Whether a summary worked is measured by the decision that follows it. These are the decisions these five most often produce in practice.

From the metadata summary, most often: rewrite the keyword field in the next release. The decision is easy because the cost is low - clearing a field is editing a text, not building a feature. That it can be undone makes it easier still.

From the cusp band summary, most often: shorten the target list. Teams usually try to track thirty or forty phrases and touch none of them. When the summary shows the band the list drops to five, and those five actually get worked on.

From the rival tracking summary, most often: cancel the panic move. When a rival gets ahead the first reflex is to touch metadata. The summary says to check whether the movement came from the market, and the answer is usually that it did - in which case the best thing to do is nothing.

From the country priority summary, most often: postpone opening a new country. Opening one while your existing keyword fields are half empty only increases the number of empty fields. When the summary counts that, the decision makes itself.

From the return summary, most often: stop a piece of work. Dropping time spent on an unmeasurable area gains more than adding time to a measurable one. That decision is hard, because the work being dropped is usually somebody's favourite.

Clear the keyword field - low cost, reversible, measurable.
Cut the target list to five - working thirty phrases is working none.
Cancel the panic move - check the control group first.
Postpone the new country - until the existing fields are full.
Stop the unmeasurable work - usually the largest gain available.

Frequently asked

Do summaries replace the long texts?: No. They exist so you can recall a subject you know. Reading a summary of a subject you have not read leads to deciding on incomplete information - especially because they cannot carry the exceptions.
How long does one take to read?: Ten minutes. If it takes longer it is not a summary but a short guide, and there is no such category in this library.
Can I share them with my team?: Yes, that is what they were written for. The most productive use is to share the summary and add one page with your own data: when your own measurements sit beside the general sentences, the discussion becomes concrete.
Why are there no percentages in the summaries?: Writing the average effect of a tactic as a percentage requires long measurement across many apps. That measurement has not accumulated. It will be published when it does; until then a percentage would be invention.
Which summary should I start with?: Whichever subject you most recently made a decision about. Summaries are a decision-support set rather than a learning sequence; they do not have to be read in order.