Raw search data is not yet insight. It becomes a steering instrument only once you have agreed in advance which value triggers which action. This article reads the eight Search Console views and the six SERP views of the Semalt panel against that one requirement, starting where Amsterdam figures depart from the textbooks.
Access to data stopped being the problem years ago: every verified domain delivers daily numbers on impressions, clicks, click-through rate and positions. What is missing is the translation — the moment somebody turns a number into a task with a name and a date on it.
That step has two enemies. The first is the composite figure that squeezes too many situations into one value. The second is the missing threshold: without an agreed boundary, every observation stays a topic of conversation. Both are tackled below, across the fourteen views in the rebuilt Semalt panel.
Why search data from Amsterdam needs an extra split
Amsterdam is a capital city and an international business centre at once. An accountancy firm on the Zuidas serves Dutch mid-sized clients and the subsidiaries of foreign parents from one building. A hotel in the centre takes nearly all its bookings from people who speak no Dutch.
That means two markets side by side on one domain, each with its own competitors, volumes and maturity. Search Console throws them onto one pile by default. A service can rank fourth in Dutch and twenty-seventh in English; the report says 15.5 and describes neither.
The country filter offers no way out: both languages are searched from within the Netherlands. The expat looking for a bookkeeper in English sits in the same country row as the business owner two streets away. The split has to come out of the URL structure, and so out of the Pages view.
Two languages, one property
As long as the language versions live in directories of their own, the Pages view is your language filter. Without that structure the separation cannot be made.
- Split first, judge afterwards
- Each language gets its own thresholds
The Netherlands is not Flanders
Flanders shares the language but not the market. Results differ per country, and a Flemish visitor rarely becomes a client of an Amsterdam firm.
- Weigh the Belgian share separately
- Growth there hides standstill here
English does not mean abroad
Much of the English-language search traffic in the Randstad comes from people who live and work here. That is traffic with spending power, not noise.
- Do not filter it away as irrelevant
- Do judge it on its own terms
Mobile as a market of its own
Hospitality, tourism and local services are searched for mainly on a phone. A combined figure irons out precisely that difference.
- Read desktop and mobile apart
- A gap above two positions is a signal
Average position leaves out too much
Average position is the most quoted value in Search Console and the least usable. Not because of the arithmetic, but because of three properties that make it unfit for steering.
It weights by impressions, not by value. A query with 9,400 impressions at position twenty-six drags the average down harder than a commercial query with 260 impressions at position three pulls it up, though the second delivers the enquiries. The relationship with revenue is not evenly spaced. Twelve to seven changes traffic materially; thirty-one to twenty-six changes nothing, yet both are five places. It punishes growth. Every keyword you newly appear for enters low and pushes the average down, while the added visibility is a gain.
Position bands answer all three objections by replacing a continuous scale with groups. Instead of one value you get a distribution: nine keywords in the Top 3, thirty-eight in the Top 10, a hundred and twenty-one in the Top 30, the rest outside. Such counts survive month-on-month comparison, because newcomers no longer cloud the picture.
An example shows the difference. An Amsterdam consultancy sees its average position move from 12.7 to 12.3 in a month: slight improvement, no action. The distribution says otherwise. Seventeen keywords entered the Top 30, nearly all long variants with a handful of impressions, while five left the Top 10, two with three-figure monthly volumes. The weak arrivals masked two heavy losses.
Keyword dynamics: arrivals and departures around three boundaries
The eighth Search Console view does what the other seven do not: it counts transitions instead of states. Not which position a keyword holds today, but which boundary it crossed since the last measurement. That is the difference between explaining afterwards and intervening in time.
Keyword dynamics
Counts who has crossed the Top 3, Top 10 or Top 30 boundary since the previous measurement, in both directions.
- Entering the Top 30. The first reliable sign that the search engine considers a page relevant. It often arrives weeks before a single click does.
- Entering the Top 10. From here you get clicks you can count. These are the most rewarding candidates for a sharper title and description, because the effect shows up straight away.
- Entering the Top 3. Where the bulk of the clicks lands. A jump in total click volume can almost always be traced to a newcomer in this group.
- Leaving the Top 10. The most urgent signal of all fourteen views. Behind the first page, traffic on a keyword all but disappears within days.
- Leaving the Top 30. A late bulletin. The cause lies weeks back, and the way home takes far longer than the fall did.
The value sits in the pile-up, not the individual case. One keyword leaving the first page is noise. If four leave at once from the same URL or theme, there is a shared cause: a technical change, an edited template, a refreshed competitor page, or a search engine update.
Eight views, eight reasons to open one
The division into eight views looks like a menu but works as a division of labour. Each answers one question well and every other question badly. The table ties each view to the reason for opening it and the action that follows.
| View | Reason to open it | Action that follows |
|---|---|---|
| Dashboard across all sites | Does one of my projects need attention today? | Open the property that stands out in the health markers |
| Site overview | Which way is this domain heading over 28 or 90 days? | Set movement against your own changes and date it |
| Keywords | Which queries perform below their potential? | Rewrite the search result where impressions are high and clicks are low |
| Pages | Which URL carries the result, and are two competing? | Merge them and link internally to the stronger URL |
| Devices | Does mobile perform structurally worse than desktop? | Go through speed, layout stability and controls technically |
| Countries | Which share of the volume comes from the target market? | Judge the Netherlands, Belgium and the rest separately |
| Traffic | How is the click volume composed and how do I report it? | Export to CSV, JSON or PDF for the monthly meeting |
| Keyword dynamics | What has actually changed this week? | Diagnose grouped departures from the Top 10 |
Two of the eight deserve a word. Pages matters more for Amsterdam sites than elsewhere, because it carries the language separation: with the English versions in a directory of their own, it becomes your language filter and you read keyword figures per market. Traffic does not analyse but reports: it carries the numbers into the meeting without screenshots.
The background processes of the Semalt panel synchronise continuously, so every view is current without a manual refresh. That lowers the barrier to looking in between times — the same barrier that explains why reports elsewhere get read once a month.
Six views of the field around you
What Search Console cannot know in principle is how the domain two places above you is doing. About other people's domains it holds no data. The ranking module exists for that, with six views built on comparison with others rather than with your own past.
Ranking score across all sites
One composite value per domain across the full tracked keyword space. Meant for comparing a domain with itself.
Average position and keyword count
The two headline figures beside the score. In practice the number of ranking keywords moves faster and earlier than average position.
Top keywords
Ordered by search volume and current position. Shows where one place gained is worth more than ten places elsewhere.
Best pages
Which URLs carry the most positions. This list almost always differs from what the editorial team regards as the important pages.
Competing domains
The sites playing in the same keyword space, with Domain Authority and the number of shared keywords alongside.
- Exposes unexpected opponents
Global ranking with a trend
One picture across the whole portfolio, with a curve over 28 days. Meant for dividing up hours and attention.
- Which project deserves more hours next month?
The competitor view surprises most often, because the real opponents are rarely the companies management names. A boutique hotel in the canal belt competes online with booking platforms and city guides, not with the hotel around the corner; a Zuidas service firm loses visibility to comparison sites and trade media. That never comes from your own performance figures, only from the ranking module Semalt provides.
Your own performance against the competitive field, and the two days in between
The two sources get confused constantly, and when they contradict each other neither is usually wrong. Search Console averages real impressions across locations, devices and personalisations; the ranking module runs a standardised, depersonalised query and reads what appears. A reported 9.4 beside a tracked 6 is not an error but two measurement set-ups.
| Question | Search Console | Rank tracking |
|---|---|---|
| Where do the figures come from? | Official Google connection on your own property | Standardised reading of the results page |
| What stays invisible? | Everything you never appear for | What visitors actually click |
| Who is in the picture? | Your own domain and nothing else | Also the domains above and below you |
| How current is it? | Roughly two days of delay | Continuous, with a trend over 28 days |
| What do you use it for? | Return on and prioritising of editorial work | Positioning and sizing up the competition |
| Where does it break down? | Low-volume queries are missing | No relationship with revenue at all |
The two-day delay is not a shortcoming of the panel but a property of how Google processes data. It still causes more false alarms than anything else. Look at the weekend on Monday morning and you see a sharp drop that exists only because those days are not fully computed. Two days later the graph looks normal.
Three working agreements follow. The last forty-eight hours are not interpreted. Reports keep three days of distance from the end date. The weekly check sits late in the week, so the preceding week is complete. The 28-day and 90-day presets already account for the delay.
A weekly routine of thirty-five minutes
The routine below costs roughly thirty-five minutes per week per domain. What matters is not the steps but the numbers behind them: each carries a boundary that decides whether anything happens. Without it, the routine is a guided tour.
- Departures over seven days — ten minutes. Threshold: four or more leavers from the Top 10 on one URL or within one theme. Individual cases you note down without acting on them.
- Arrivals over seven days — five minutes. Boundary: every newcomer on the first page clearing 150 impressions gets a new title and description.
- Keywords per language — ten minutes. Boundary: above 250 impressions, a click-through rate under one and a half per cent and still a place in the Top 10. Then the search result is at fault, not the position.
- Position bands month on month — five minutes. Boundary: if the Top 10 group shrinks by more than fifteen per cent, an investigation follows. Measure per language, never across the total.
- Competitors, monthly — three minutes. Boundary: a domain climbing from fifteen to forty shared keywords within a month deserves a closer look.
- Device split, monthly — two minutes. Boundary: if mobile trails desktop by more than two and a half places on the same keyword group, a technical check follows.
Run the routine consistently and you arrive at the question of who does the tasks. For companies without SEO capacity of their own, the panel couples analysis to execution at two levels; the views described here stay available in both.
AutoSEO and FullSEO
Two levels for companies that would rather not carry out the tasks from the weekly routine themselves.
- AutoSEO, 149 USD per month. Automatic keyword discovery and prioritisation, automatic link building, recommendations for your own pages and full access to the analytical views described here.
- FullSEO, 500 USD per month. On top of that, manual keyword selection with an automatic fallback, targeted link placement at authority level and a review mode in which a human releases changes to the site.
- Keywords remain a decision. The candidate list comes from Search Console, the live results page and your own suggestions; each candidate is approved, rejected or deferred individually.
- What to expect on timing. The first measurable movement usually sits four to eight weeks after the start. Concluding anything earlier means measuring noise.
These thresholds are not constants of nature. A domain with seventy ranking keywords needs different boundaries than a portfolio of four thousand, and a bilingual site calibrates per language. The quickest way: run the assessment once against your own Search Console data from last quarter and count how often each boundary would have fired.
Frequently asked questions
Why does the position in Search Console differ from the position in rank tracking?
Because two different things are measured. Search Console averages real impressions across locations, devices and personalised results; rank tracking runs a standardised query without personalisation. Both can be right at once: use Search Console for return, rank tracking for your place among competitors.
My average position is getting worse while traffic is growing. What is going wrong?
Probably nothing. On a growing site, keywords keep arriving at low positions and press the average down while visibility rises. Look at the position bands: if the Top 3 and Top 10 counts are higher, the movement is positive.
How do you keep Dutch and English apart on one domain?
Not with the country filter, since both languages are searched mainly from within the Netherlands. The separation runs through the URL structure: with each language version in a directory of its own, the Pages view becomes a language filter and you judge keywords per market. If the languages run together on the same URLs, you have a structural problem first and an analytical one only afterwards.
Should Belgian traffic count towards the targets?
Only if you genuinely serve Belgium. The language is shared, the market is not: Dutch-language results differ per country, and Flemish visitors rarely request a quote from an Amsterdam firm. Set the Belgian share aside in the country view, or your headline figure will look better than your diary.
Why do the clicks per keyword not add up to the total?
Queries with very little volume stay out of the detail rows so individual searchers cannot be identified. Those clicks sit in the total but in no row. On a bilingual domain the share rises, because volume fragments across two languages. A property of the source, not a counting error.
What tips the balance
None of these limits makes the analysis worthless. They only determine which statements survive. Cause analysis stays manual work and assumes your own changes were recorded with a date. Keeping that habit around technical and editorial adjustments saves hours of argument later; related analyses sit on our blog.
Whether search data yields anything rarely depends on its availability. It depends on whether somebody wrote down which observation leads to which action, and whether that agreement contains a number rather than an impression. In Amsterdam one further question belongs to it: which of the two markets is this figure actually about?
To see your own numbers in this arrangement, connect your property to the Semalt dashboard with your own domain and let a quarter run through it. What helps first is rarely the average position; it is the list of keywords that have vanished from the first page — and on a bilingual domain that list almost always comes from one language.
Hulp Nodig met Uw SEO?
Onze Amsterdam SEO-experts kunnen u helpen deze strategieën te implementeren en uw rankings te verbeteren.
Gratis Consult Aanvragen