How it works
One Domain In, One Share Out
The AI visibility score is the share of about 16 searches where a site appears in search results. Here is how one call gets there.
1. You send a domain
Send a website address such as example.com and, if you want, a country. A full address like https://www.example.com/page is reduced to its domain.
2. We run about 16 searches
From public information about the domain we work out the searches its audience would type: 14 general topic searches that do not name the brand, and 2 that do. Each runs for your country.
3. We count where the site appears
For every search we check whether the site itself shows up in the results, and at which position. score is the share of the 14 topic searches where it does, from 0 to 1.
4. You get two numbers, kept apart
Brand searches nearly always find the brand, so that share sits close to 1 for every site. It comes back as brandScore, never mixed into score, so it cannot inflate the headline.
What the number is, and is not. It is a share of search appearances for searches about a site’s own topics. It is not a rank position and not a traffic estimate.
Pricing
Five Price Tiers, Per Site Checked
One call is one site, however many searches run. The rate falls as your deposit tier rises; there is no monthly plan.
Site check
GET /api/visibility/score—domainrequired- About 16 searches per site by default;
keywordssets 4 to 30 - The price is per site, not per search
- Same countries as web search
Per 1,000 sites: $200 on Default, $120 on Growth, $70 on Scale, $40 on Enterprise and $20 on Enterprise Plus. Growth takes one qualifying $100 deposit, Scale $500, Enterprise $1,000 and Enterprise Plus $1,250 for new credit. The deposit is spendable balance rather than a fee; existing credit keeps its original schedule. Refunded or charged-back amounts don’t count toward your tier. There is no monthly plan.
Quick Start
A Visibility Score in One Request
One header, one GET. Every key is always present, so an absent value arrives as null and never as a missing key.
# Score otto.de for Germany. One call, one site. curl "https://api.apiserpent.com/api/visibility/score?domain=otto.de&country=de" \ -H "X-API-Key: YOUR_API_KEY" # Score your own list of searches instead of the ones we work out. curl -G "https://api.apiserpent.com/api/visibility/score" \ --data-urlencode "domain=example.com" \ --data-urlencode "topics=garden hose,raised garden bed,compost bin" \ -H "X-API-Key: YOUR_API_KEY"
import csv, os, requests h = {"X-API-Key": os.environ["SERPENT_API_KEY"]} sites = ["otto.de", "heise.de", "t3n.de"] with open("scores.csv", "w", newline="") as f: w = csv.writer(f) w.writerow(["domain", "score", "brandScore", "answered", "run"]) for d in sites: r = requests.get("https://api.apiserpent.com/api/visibility/score", params={"domain": d, "country": "de"}, headers=h, timeout=60).json() # score uses the searches that were answered; read both counts before you store it. w.writerow([d, r["score"], r["brandScore"], r["searches"]["answered"], r["searches"]["run"]])
{
"success": true,
"domain": "otto.de",
"landedDomain": null,
"topicSource": "derived",
"score": 0.64,
"brandScore": 1.0,
"keywords": [
{
"q": "uhren online kaufen",
"type": "topic",
"answered": true,
"appeared": false,
"bestPosition": null
},
{
"q": "armbanduhr damen",
"type": "topic",
"answered": true,
"appeared": true,
"bestPosition": 1
},
{
"q": "herrenuhr sale",
"type": "topic",
"answered": true,
"appeared": true,
"bestPosition": 2
},
{
"q": "automatikuhr kaufen",
"type": "topic",
"answered": true,
"appeared": true,
"bestPosition": 4
},
{
"q": "chronograph herren",
"type": "topic",
"answered": true,
"appeared": false,
"bestPosition": null
},
{
"q": "mode online shop",
"type": "topic",
"answered": true,
"appeared": false,
"bestPosition": null
},
{
"q": "möbel online kaufen",
"type": "topic",
"answered": true,
"appeared": true,
"bestPosition": 5
},
{
"q": "wohnlandschaft günstig",
"type": "topic",
"answered": true,
"appeared": true,
"bestPosition": 1
},
{
"q": "matratze kaufen",
"type": "topic",
"answered": true,
"appeared": true,
"bestPosition": 1
},
{
"q": "handtücher set",
"type": "topic",
"answered": true,
"appeared": true,
"bestPosition": 1
},
{
"q": "weißwein gläser",
"type": "topic",
"answered": true,
"appeared": true,
"bestPosition": 4
},
{
"q": "heißluftfritteuse test",
"type": "topic",
"answered": true,
"appeared": false,
"bestPosition": null
},
{
"q": "e scooter mit abe",
"type": "topic",
"answered": true,
"appeared": false,
"bestPosition": null
},
{
"q": "qled fernseher 55 zoll",
"type": "topic",
"answered": true,
"appeared": true,
"bestPosition": 1
},
{
"q": "otto uhren",
"type": "brand",
"answered": true,
"appeared": true,
"bestPosition": 1
},
{
"q": "otto mode sale",
"type": "brand",
"answered": true,
"appeared": true,
"bestPosition": 1
}
],
"searches": {
"run": 16,
"answered": 16
},
"meta": {
"elapsed": "17021ms",
"timestamp": "2026-11-18T15:00:00.000Z"
}
}
This is a real response for otto.de, 18 Nov 2026. score is 0.64: the site appeared in 9 of the 14 topic searches. brandScore is 1.0, as it nearly always is. bestPosition is the best place the site held in that search, counted from 1, and null when it did not appear. The 20-site table further down is a separate, earlier test, so its numbers differ.
Parameters & Fields
What You Send, What Comes Back
One required parameter and four optional ones. The response has a fixed shape, so you can parse it without checks for missing keys.
| Parameter | Required | What it does |
|---|---|---|
domain | Yes | A website address such as example.com. A full address like https://www.example.com/page is accepted and reduced to the domain. |
topics | No | Comma-separated search phrases, 2 to 100 characters each, up to 30. When sent, we score exactly these and topicSource is supplied. |
country | No | The country the searches run for. Default us. Same countries as web search. |
language | No | Two letters. Defaults to de for Germany, Austria and Switzerland, and to fr, es and it for France, Spain and Italy. Otherwise not set. |
keywords | No | How many searches to run, a whole number from 4 to 30. Default 16. The price per site does not change. |
Send your key in the X-API-Key header.
| Response field | What it holds |
|---|---|
success | true when the call returned a score. |
domain | The domain that was scored. |
landedDomain | null unless the domain you sent redirects to a different domain. Then the domain we landed on is the one scored, and it is reported here. |
topicSource | derived when we worked the searches out from public information about the domain; supplied when you sent topics. |
score | The headline number, 0 to 1: the share of topic searches where the site appears in search results. |
brandScore | The same share for the brand searches. Close to 1 for almost every site. |
keywords[] | One row per search: q (the search), type (topic or brand), answered, appeared and bestPosition (from 1, or null when not seen). |
searches | run and answered. If some searches could not be answered, score uses the answered ones and answered is lower than run. |
meta | elapsed and timestamp for the call. |
What a score looks like
A Small Test of 20 German Sites
We ran 20 German publisher and shop sites on 18 Nov 2026. It is a small test, not a benchmark, and it used 20 searches per site where the default is now 16.
| Site | Topic score (0 to 1) |
|---|---|
| zalando.de | 0.69 |
| reisereporter.de | 0.69 |
| schoener-wohnen.de | 0.63 |
| otto.de | 0.56 |
| finanztip.de | 0.56 |
| heise.de | 0.44 |
| utopia.de | 0.44 |
| eatsmarter.de | 0.38 |
| nettolohn.de | 0.38 |
| netzwelt.de | 0.31 |
| t3n.de | 0.19 |
| Four small sites | 0.06 each |
| One blog | 0.00 |
How to read it. In that test 3 of the 20 sites could not be scored and one had moved to another domain; the average across the other 16 was 0.34. The bigger sites scored 0.44 to 0.69. Four small sites sat at 0.06 and one blog at 0.00. That is the pattern you would expect: broad head terms favour big sites. Twenty sites say little about any one site, so use the table to get a feel for the scale, not to set a threshold.
Honest limits
What the Score Can and Cannot Tell You
It is a good way to sort and compare sites. It is a poor way to judge any single site on its own. Four limits to know before you build on it.
It measures one thing
The score counts the topic searches in which the site appears in search results, and nothing else.
It is not a rank position and not a traffic estimate. bestPosition is shown per search, but the headline is a share.
Big sites score higher
Large sites appear for broad head terms, so they score higher than small niche sites that are excellent at one thing. In our test the large sites scored 0.44 to 0.69 and the small ones near 0.06.
Compare sites of a similar size, or compare one site with itself over time.
It is a sample of about 16 searches
Each search is worth about 0.07 of the topic score, so a site can move by a step or two between runs without anything changing. Send topics to keep the same searches from run to run.
Raise keywords, up to 30, for a finer reading.
Country, and one site per call
Results differ by country, so a score belongs to the country you asked for. A site can look strong in one and absent in another.
One call checks one site. To rate a list, call once per site and store the results as you go.
Use cases
Where a Visibility Score Earns Its Keep
Four jobs where one comparable number per site is more useful than a judgement made by hand.
Rating thousands of publisher or client sites
Agencies, ad networks and SEO tools often hold a list too long to open site by site. One score per site lets you sort the list, set a cut-off, and spend review time on the sites near it.
Call once per site and write each result to your own table.
Pre-screening domains before outreach
Before you write to a site, check whether it shows up for the topics it claims. A site at 0.00 for its own subject is a different conversation from one at 0.60.
Keep brand and topic apart: brandScore will be high for almost everyone.
Tracking a site’s search visibility over time
Re-run the same site on a schedule and keep the score with its date. Send the same topics each time, so a change in the number is a change in the site and not in the searches.
Read searches.answered beside each score before you plot it.
Sanity-checking a niche
Score a handful of sites in one niche with the same topics and see who appears for the searches you care about. It is a quick read on who already owns the space.
Use one country per comparison; the same site can score differently elsewhere.
FAQ
AI Visibility Score Questions
score counts those. The other 2 name the brand and are reported separately as brandScore.
brandScore is the same share for the searches that name the brand. It sits close to 1 for almost every site, because a search for a brand finds that brand, so it says little about visibility. We keep it in its own field so it can never inflate the headline score, which counts topic searches only.
keywords changes the total, from 4 to 30. To score your own list, send topics as comma-separated phrases, up to 30. We then score exactly those and topicSource comes back as supplied instead of derived. Sending topics is also the way to keep the same searches from one month to the next.
searches tells you how many that was: run against answered. Each row in keywords also carries its own answered flag. If you need every search answered, check that the two numbers match before you store the score.
Score your first site
10 free API calls, shared across every free-eligible endpoint. No card, no subscription, no monthly plan. From $0.02 per site checked.
Get your free API key
