Track your rankings on a reading you can compare month to month
Your rankings move every day, and a single reading does not say which way. We take the measurement daily on your tracked keywords, attach to every number the configuration that produced it, and show the trend against a median reference rather than against yesterday’s reading.
Access opens in waves: we email you when yours is ready.
Our numbers, with their provenance
-
100results captured per query ceiling
-
10competitors per project ceiling
A ranking is quick to read, a trend takes proof
A ranking varies from one day to the next for reasons that are not always about you: an engine test, personalisation, a result block appearing above you. The three situations below almost always come before the sentence “we lost rankings”, said without anyone being able to name the day.
-
01
Today’s number contradicts yesterday’s
A single reading mixes real movement with engine noise. Deciding on that basis means fixing a page that was fine, and leaving alone the one that is actually slipping.
-
02
You hear about a drop from your client
Between two monthly exports, four weeks go by. A query your business runs on can fall off the first page early in the month: you find out at the end, once the traffic loss is already in the numbers.
-
03
Two readings taken differently end up compared
Language, country, device: change one of the three and you are measuring another quantity. Meeting decks overlay both curves anyway, and nobody can tell which one describes what.
What we measure
Four quantities, each with its date
The rank, what surrounds it on the results page, the domains that appear there, and the reference everything is compared against. Each carries the configuration that produced it: language, country and device travel with the number, otherwise nothing is comparable after a settings change.
- The rank, read every day (every tracked keyword goes out once a day, to a depth published in our claims registry. Beyond that depth the product writes “outside the reading”: a dated finding about your position, and it reads as one.)
- The result blocks that take the click (rich snippet, local pack, people-also-ask, featured answer: the reading records what occupies the screen above you. A first place under several blocks is worth less than a second place under none.)
- Competitors, pulled from your own readings (the domains that surface on your queries are read from the results pages of your own tracking, without one extra call. You see who is climbing on your keywords without ordering a second measurement.)
- A median reference, to decide on (the comparison runs against the median of your readings, never against yesterday’s. One isolated accident would raise the threshold and make a real drop look like a return to normal.)
Further reading
Rank tracking: building a reading that compares over time
The guide: building a rank reading you can compare over time.
Read ComparisonNessFlow vs Semrush
What an all-in-one suite records on rankings, and what its model asks in return.
Read GlossaryGlossary
SERP, result block, average position: the definitions.
ReadOther modules
Frequently asked questions
How often are rankings read?
Once a day, on every tracked keyword. An hourly reading would measure mostly noise: served results shift from one minute to the next, with no decision that could rest on them. The date of every pass sits next to the number, so you can see whether the curve you are reading is complete.
What happens when a reading fails?
The cell stays empty and the screen shows “not measured”. This is the module’s most important distinction: “outside the reading” is a finding about your position, while a collection that failed is nothing at all. Confusing the two would sink your curve on the day of a network outage, and trigger a meeting about a collapse that never happened.
Is the reading taken on mobile or on desktop?
On one of the two, chosen per project, and the device travels with every reading. The choice matters: the results served differ from one device to the other, and a curve mixing both would describe no reality at all. The product therefore makes you decide when configuring the project, rather than blending two measurements into an average that would match nothing.
How are the tracked competitors chosen?
They come out of your own readings: the domains that actually surface on your queries are extracted from the results pages of your own tracking, without ordering one more measurement. The number kept per project is a configuration cap, published in our claims registry together with the file that holds it. So you can check it instead of taking our word.
The scope of this reading, and its limits
A rank reading is a dated measurement, not continuous surveillance. The cadence, the device and the depth are bounded because of it, and writing them here beats letting anyone imagine unlimited watching.
What we cover
- One reading per day and per tracked keyword, with the date of every pass
- The result blocks recorded at the same time as the rank
- Competitors pulled from your existing readings, at no extra spend
Stated limits
- Desktop or Mobile, chosen per project: one project measures a single device
- A missed pass shows as “not measured” and stays distinct from a lost position
- No score out of a hundred: this module carries a trend, which the scorecard leaves out of its calculation
- No automatic explanation: the reading says a position moved, the cause is still yours to find
Your rankings, with the date of every measurement
A daily reading, its configuration attached, a median reference: enough to say what moved, when, and on which query.
Access opens in waves: we email you when yours is ready.