Recipe
The weekly recipe for ten minutes of link data
Four calls, one notebook file, one named weekday. This is the plain way to check backlink movement for your site and a few rivals without opening a big audit tool.

What you are cooking
Think of it as a small meal, not a banquet. Each week you collect the same few numbers in the same order. Then you add one line to a file, and that is the whole job. Nothing more.
A single reading means little. After four to six entries you can see whether your referring domains are creeping up, standing still or sliding, which is exactly the question a ten-minute routine exists to answer. Direction is enough.
Our own crawler and our own link index feed every call. The index refreshes about once a month, so a link someone added last week may not appear yet. Hence the short recipe.
Rivals matter too. If you want to see how they compare, the rival gap table shows how to read the numbers next to yours.
Ingredients and what each costs
| Ingredient | Action | Cost per call |
|---|---|---|
| Your DR and referring domains | overview | Free, no key |
| A few rivals side by side | compare (up to 20 domains) | 1 credit |
| A quick scan of a long list | bulk_dr (up to 2,000 domains) | 1 credit |
| Who links to you | refdomains | 2 credits |
| Which phrases point at you | anchors | 2 credits |
The method, five small moves
Warm up with the free call
Run overview for your own domain. It needs no key, so it is a gentle start and a quick test that the command works.
Add the rivals
Use compare with up to 20 domains in one call. Keep the same three or four rivals every week, because swapping them ruins the trend.
Scan the long list
If you manage a bundle of sites or a list of prospects, bulk_dr takes up to 2,000 domains per call for 1 credit. One call, one charge.
Once a month, pull the names
Fetch refdomains with per=100 and order=desc. Save the file, then compare it with last month's to see which domains arrived and which left.
Write one line
Date, your DR, your referring domains, and a three-word mood such as flat, rising or odd. Close the file. You are done.
The requests, ready to paste
Replace the placeholder key and the domains. Every authenticated response includes an ok flag and a _credits field with the cost and your remaining balance.
# 1. free, no key
curl "https://rankdose.com/api.php?action=overview&q=yourdomain.com"
# 2. one credit: many domains at once
curl "https://rankdose.com/api.php?action=bulk_dr&domains=yourdomain.com,rival-one.com,rival-two.com&key=sk_live_..."
# 3. two credits, monthly: who links to you
curl "https://rankdose.com/api.php?action=refdomains&q=yourdomain.com&per=100&order=desc&key=sk_live_..."
# header style also works
curl -H "Authorization: Bearer sk_live_..." "https://rankdose.com/api.php?action=anchors&q=yourdomain.com"
A worked example, clearly invented
Here is a made-up month for a made-up flower shop. Every number below is invented. Do not read them as data about any real site, because the shop exists only in this example.
In week one the shop writes down 41 referring domains. Weeks two and three show 41 again, which is normal when the index has not refreshed. Then week four. It jumps to 47, and the monthly pull shows six new names, two of them local bloggers.
That is a good dose: six names to say thank you to, nothing to panic about. Because DR here is round(100 · log(ref_domains+1) / log(20,000,000)), 41 domains gives 22 and 47 gives 23, so the score barely twitched while the people behind it grew.
Week five will show whether 47 was a blip or the start of a new level, which is why the recipe looks at a run of entries rather than one.
Questions that show up in week four
Week four is when the notebook starts to talk back, and that is also when the doubts arrive. Most have simple answers. Read them before you change the recipe.
Does a flat week mean the routine broke?
Usually it means the index has not moved yet, even when you check backlink changes every week. Our link index refreshes about once a month, so two readings a week apart can match while people really are adding links elsewhere on the web. Say you run the overview on one Sunday and again the next, and both show the same count. That is a refresh gap.
Flat is a result too. Write the date and the count, then add the word flat, because a dated flat line is evidence that the routine ran at all.
Compare a flat week with the week before it, not with the best week you ever logged. Ordinary months are mostly ordinary.
How many rows does the monthly pull need?
The refdomains call costs 2 credits whatever the number of rows it returns, up to 1,000 per call. Size is not the cost. Most small sites act on a few dozen names. A page of 100 is plenty, and deep pagination, which reaches 50,000 rows, only matters when a real decision depends on it.
Save each pull with its date in the file name, and keep last month's copy beside it. Naming is half the job. The comparison then takes minutes rather than an afternoon. For the usual slips, see the mistakes and fixes page.
Can a weekly check show a penalty?
No. A referring-domain count cannot tell you whether Google has taken manual action against your site. That answer lives in Search Console, under manual actions and security issues, and it is worth a look once a month beside your notebook.
If the page shows nothing, the link numbers answer a different question. Two questions, two answers. Keep the two questions apart, and the notebook stays useful.
Can I run the same routine on rivals I do not know well?
You can, because the calls accept any domain. Permission is not evidence. Be careful about what the numbers mean for a site you do not run, though. A rival's referring-domain count tells you how many sites linked to it, not whether any of those links bring visitors.
Keep the same rivals each week, as the method says, and read the table as a rough map. It shows direction. It does not prove why a page ranks.
When a rival's count jumps, resist copying its whole list. Patterns beat copies. Read two or three of the new pages and ask what they give their readers.
Should I add a new rival after a few months?
Only when the niche itself has changed, such as a new competitor taking the top spot in your searches. Change is the trigger. Adding names mid-trend makes the earlier readings hard to compare.
Write the new name down with its start date, and leave the older rows exactly as they were. Start dates matter.
Is the monthly pull worth it when time is short?
It can be skipped for a month without harm, as long as the weekly lines continue. Skipping is allowed. The monthly pull is where the new names live, so it is the first thing to restore when you have the time again.
Write skipped in the notebook, with the date. Then return to the normal slot.
Before you trust a monthly pull
- The date on the notebook line matches the day you ran the call.
- The same three or four rivals appear as they did last month.
- You noted whether the monthly refresh has landed since your last pull.
- You counted domains, not links, when you compared the two files.
- You read twenty of the new names by eye before you call them growth.
- Search Console was checked for manual actions in the same month.
- The index refresh note matches the date in your notebook.

A rhythm for the quiet weeks
Not every week will produce a story. Some weeks the notebook looks dull, and that is normal for a habit built to be dull. Boring is the goal.
When the calendar slips, the answer is rarely more effort. Smaller works. If a spike does arrive, the alarm-line page says what to do next.
What do I write in a week with nothing new?
Write the line anyway, even if it reads exactly like the one before. A gap in the notebook is harder to read later than an entry that says nothing changed. Two minutes is enough.
If you skip a week on purpose, leave a short note such as away this week. Honest gaps are fine, and they keep the trend readable when you look back.
What if the calendar slot keeps slipping?
A slot that slips twice in a row is a sign that the routine is too big for your week. Shrink it. Run only the free overview and write the line, which keeps the thread alive without the full set of calls.
A smaller routine kept every week beats a full one kept twice. That is the whole trick.
Do I need a spreadsheet for the notebook?
No. A plain text file with one line per week does the job, exactly as the recipe describes. A spreadsheet adds columns you will rarely read, while a single file is easy to back up, search and share with a colleague.
Name the file after your domain and keep it in one place you open anyway. Simple beats clever.
If you ever switch to a spreadsheet, keep the same columns, so the old lines still line up. Same columns, same order.
Say the weekly part is one compare plus one bulk_dr call, and the monthly part adds two 2-credit pulls. That is 2 credits most weeks and 6 in the heavy week, well under a tenth of a cent each time, against a $100 free balance.