Everything here is an estimate
No broker reports its traffic, its customers or its spending to us. Everything on this site is measured from the outside, from what is publicly visible about a brand rather than from its books. That kind of measurement shows direction and relative position well. It is not accounting.
Read a number as "roughly this, compared with the others here". Never as an exact count. A brand share of three in ten means that brand is about three of every ten branded searches we can see in that market. It does not mean it holds three in ten of the customers.
Where it comes from
There is no single source. Some of it the platforms publish themselves. Some we collect directly. Some reaches us through specialist providers. Which route carries which figure changes over time, because a provider gets replaced or something we used to buy becomes something we measure ourselves. The site is built so a figure can change route without changing meaning.
Collection runs on a regular schedule, each source at its own rhythm, and little is ever thrown away. A new measurement joins the ones before it, so a series gets longer instead of being replaced. Together that is many gigabytes of measurements. Most of it exists so the handful of numbers on a page can be put in context.
Rankings compare inside a market; indexes compare across them
Within one market every broker is measured the same way over the same period, so a ranking there is solid. Raw totals do not travel between markets. Language, vocabulary and the number of active brokers all differ, so more searches in one country than another does not mean a broker is doing better there.
That is why figures meant for comparison between markets are indexes against their own normal: a broker a fifth above its own year is a fifth above it in every country. The market size index is the exception. It sets the searches in our measured keyword set of a market against the median market, and that set follows the language and how many investing keywords it has, so the index gives the order of size between markets, not an exact ratio.
How to read an index and a label
An index puts a figure against a point of comparison, and 100 is that point. Under every index the page names what it is compared with: the broker's own usual level, the same period a year earlier, or the typical market. Above 100 is more than that, below 100 is less.
Words such as Busy, Quiet, Growing or Declining sit on top of those comparisons. They change only when the move is clear, so a small wobble keeps the word it had. The number beside the word shows how large the move is.
A figure shown with ≥ is a lower bound: at least this much, and possibly more. A figure shown with ≤ is an upper bound: at most this much. A share shown as less than half a percent is above zero but too small to print. est. marks a brand share estimated from Google Trends, for a brand that Google Ads gives no search volume for.
Share and growth are two different questions
A brand's share can rise while the brand itself shrinks, if the market around it shrinks faster. It can fall while the brand grows. Share on its own does not tell you whether a broker is reaching more people.
Where the site answers that second question, it compares like with like in time. Attention has a season. The same periods are busy or quiet year after year, so a comparison against a nearby period mostly measures the season rather than the broker. The broker's own change and its market's change are shown together, because the pair is what makes either one readable.
Some companies are more than a broker
Several of the companies here are banks or platforms whose business covers far more than investing. Public sources see the whole company, so these figures do too. Where we know that is the case, the page says so next to the figure. A count, such as searches or visits, is then an upper bound for the investing part. A share or a rating is not: the investing part alone can score higher or lower than the whole company. Read those figures as the company, not as a like-for-like comparison with a pure broker.
What we leave blank
Where a measurement is missing, the page says so. We never fill a gap with a zero. "Not measured" and "measured as zero" are different facts, and only one of them says something about the broker. Where there is too little to measure reliably, that part stays empty instead of showing a number built on a handful of observations.
A sudden jump that nothing else confirms is held back. The previous level stays on the page until a later measurement shows the same new level.
How current it is
Each measurement has its own rhythm, and some sources publish well after the period they describe. Rather than asking you to remember which is which, every figure and chart names the period it covers. If it does not say today, it does not mean today.