Reading your research

A finished research page holds six readings of the same market. They answer different questions, and the commonest mistake is to read one of them as the answer to another. This page goes through them in order and says what each figure is — and, just as often, what it is not.

Finding your way around

A result is split into six readings, listed down the left on a wide screen and across the top on a narrow one. Only the one you have selected is shown. Click another and the page changes to it; the address bar changes with it, so a link to #earnings opens on the earnings and can be shared or bookmarked.

Section	The question it answers
Score	How crowded is this niche, and why?
Angles	What is the market already selling?
Earnings	What would one sale here pay me?
Book concept	What could I actually write?
Save	Where is this kept?
Competing titles	Which books is all of this read from?

The score

The number

The big figure is the opportunity score out of 100. Higher is better. Beside it is a plain-language band:

Score	Band
70–100	Promising
50–69	Worth a look
30–49	Difficult
0–29	Crowded

The score measures competition — how crowded and how beatable the niche looks. It does not measure demand. Nothing the app collects can see how many people want a book on a subject, and it will not pretend otherwise. A score of 80 means "few strong books are in your way", not "many readers are waiting".

The four facts beside it

• Factors measured — how many of the eleven factors could be read here. 9 of 11 means two were left out rather than guessed at.
• Organic results — how many genuine (non-sponsored) results the score was built from.
• Collected — when the market was last read. Research repeated within twelve hours is answered from this and costs nothing.
• Scoring model — which version of the model produced the number, so two scores can be compared fairly.

What moved this score — the two figures per row

This is the part most often misread. Each row carries two different quantities.

• The bar is how this niche did on that factor. Green is good for you, amber is middling, red is against you.
• The percentage is how much that factor counts towards the total — its weight in the model. It is not a score.

So a row reading Competitive entry rate — 9% does not mean the niche scored nine out of a hundred. It means that factor is worth 9% of the total, and the bar beside it says how the niche did on it. A page of green bars is good news even when every percentage is small.

Under each factor name is the actual reading in words — "1 of the 11 dated results were published in the last year" — so you never have to take the bar on trust.

The eleven factors

Factor	What it reads
Competitive entry rate	How quickly new titles are reaching the top. Counted as a cost: a rank won today is contested tomorrow.
Median book age	How old the competing titles are. An ageing page can be beaten on presentation and currency.
Price headroom	What the market will pay. Higher accepted prices need fewer sales to be worth publishing into.
Review concentration	Whether review weight sits with a few titles or is spread evenly. Concentrated counts in your favour — the typical competitor is beatable.
Competition depth	How much review weight a new title must overcome, on the median rather than the mean, so one runaway bestseller does not condemn an enterable niche.
Shelf crowding	How many matching titles Amazon reports. A competition measure, not a demand measure.
Competitor satisfaction	How happy readers are with what exists. Mediocre ratings across a page is the clearest sign readers want better.
Search result stability	Whether the same titles hold the page over time. Needs at least two readings.
Rank volatility	How much competitors' Best Seller Ranks move. A volatile niche has soft positions. Needs rank history.
Seasonal demand	Whether the niche has a season. Needs a full year of observations.
Publisher concentration	Whether one imprint owns the page. Needs publishers, which come from collected detail pages.

Not measured yet, and confidence

A factor that cannot be read honestly is excluded, not estimated. Each one says why. Seasonality needs a year of history; rank volatility needs weeks of tracking; publisher concentration needs enough detail pages collected.

The confidence badge beside the score combines two things: how much of the model's weight could be read at all, and how much the collection step trusted the reading it took. A full set of factors read from a shaky collection does not earn high confidence, and neither does a clean collection missing half the model.

Confidence	Meaning
High	0.80 and above
Moderate	0.55 and above
Limited	0.30 and above
Early	Below 0.30 — treat the score as provisional

A keyword the app has only just seen can never reach full confidence, because some factors need time that has not passed yet. That ceiling is deliberate.

Angles

Angles reads the competing titles and reports the patterns actually in use — never invented ones. They are grouped by what kind of pattern they are: Who it is for, What the book is, What it promises, The method or school, Genre and tropes, Subject matter.

Each angle carries how many of the page-one titles use it, the median reviews and median price of those titles, and a badge saying how entrenched it is:

Badge	Meaning
Thinner competition	Titles using this angle carry far fewer reviews than the rest of page one (40% or less of the page median). An opening — or a sign readers are not asking for it.
Typical for this page	About as established as everything else.
Well established	Far more reviews than the rest of the page (1.6× the median or more). Competing here means competing with books readers already trust.
Not enough data	Too few of these titles report review counts to say.

Research this beside an angle runs it as a keyword of its own. That is deliberate: the analysis can say a phrase recurs and how entrenched it is, but only a real search can say whether it is a viable niche.

Where this page is thinnest

A shortcut to the angles with the least entrenched competition. Read the wording carefully — it is a gap in the competition, not evidence that readers want it. The two are not the same thing, and the app will not blur them.

Earnings

This applies KDP's published royalty maths to what the niche already charges. For each format you get:

• The price used — the median price of the organic results, not a price you chose.
• What one sale pays you — the large figure.
• The rate — 60% for paperback, 70% for Kindle in the eligible price band.
• What you keep of the cover price — after printing cost, which is why paperback keeps far less than Kindle on the same price.

Printing cost depends on page count, so the paperback line shows the page count used: the median of the titles whose detail pages were collected. Sponsored placements are excluded, exactly as they are from scoring.

This is per sale, never per month. It does not know and does not guess how many copies you would sell. It answers one question: at what this market charges, is a book here worth the work?

Book concept

The one part of research that writes rather than measures. It takes the score, the angles and the page-one titles and drafts a concept: five title options, positioning, a description, keywords and categories.

Two things to keep in mind:

• It is charged as its own step, and the price is on the button along with your remaining balance.
• It is written by a language model from the evidence on the page. It is a starting point to react to — keep, reject or argue with the five title options. The page says so above the draft.

Save

Every result is kept automatically, so research cannot be lost by leaving the page. This section says which project it went into, and lets you move it somewhere else or remove it. Removing the shortlist entry keeps everything collected for the keyword.

Competing titles

The evidence underneath everything above: the page-one results exactly as collected. The columns are position, title (with its ASIN), author, price, rating, review count and publication date. A dash means Amazon did not report that field.

Sponsored rows are dimmed and labelled "Sponsored, excluded from scoring". They are shown because they were on the page you would be competing on, and excluded from every figure above because a paid placement says nothing about how strong a book is.

If a number anywhere else on the page looks wrong, this table is where to check it — every median, every count and every score comes from these rows.

What research never tells you

One limit is worth repeating, because it is the one that matters:

Nothing here measures reader demand. Not the score, not the angles, not the earnings. Every figure is read from the competition — who is already there, how entrenched they are, and what they charge. A thin, cheap, uncrowded niche can be all three because nobody wants the book.

What research does give you is an honest reading of what you would be up against, with its workings shown and its gaps named.

See also Research for the Opportunity Finder, the Vault and saved projects, and Credits for what the concept draft costs.