When a topic hunt finishes, you get five topics. Each one has a title and an outline. When you approve one, the writer starts and 10 credits are spent. So the few minutes before you press approve are the cheapest editing you will ever do.
This post covers what to look at, how a solid topic differs from a shallow one, and a five-check routine to run on each topic before you approve it.
What you can actually read
First, a straight answer about the "log". RankMill does not show a live trail of what an agent was thinking. Each agent runs as one call out to a model, so there is no step-by-step reasoning to replay.
Here is what you do get:
- The topic itself: a title and an outline.
- Up to four questions: the facts the agent could not find on its own.
- The run in Activity: what kind of run it was, what it worked on, how long it took, and a plain sentence if it failed.
- The website profile that the hunt read.
That is the trail. The outline is the agent's reasoning, written down. Read it the way you would read a brief from a new freelancer. It tells you what they understood and what they missed.
The good part is that you have no log to decode. The catch is that some checks are yours to do, like whether the topic matches what people search for. They are quick.
Start with the website profile
A topic hunt reads your website profile. If the profile is wrong, all five topics will be wrong in the same way.
So before you judge the topics, spend a minute on the source. Open the profile and ask:
- Does it say what you sell, in words you would use?
- Does it name the right reader?
- Does it describe how you sound?
If something is off, fix it there. You can edit it by hand, or ask for an AI edit and read the diff before you apply it. The profile is only rebuilt when you ask, so a fix stays fixed. Then run a new hunt. You can start one whenever you like, and hunting is free.
Signs of a solid topic
The outline could only belong to you
A good outline uses the words in your profile. It names your reader, your kind of product and the problem you solve. Try this test: swap in a competitor's name. If the outline still reads fine, it is too generic.
It answers a question someone actually has
This is intent match. A strong title reads like something a person would type into a search box or ask a friend. The outline then answers that question in the order the reader needs it. It is not a tour of your features.
It has an angle the top results miss
This is gap logic. The best topics are not the same post everyone else has already written. The outline adds something: a routine, a checklist, a worked example, or a view only your business can give.
The questions ask for real facts
Some topics need a price, a date, a customer name, or whether you support some integration. A good run asks for those instead of guessing. Sharp questions show the agent knew where its knowledge ran out. If a topic clearly depends on a fact like that and arrives with no questions, read the outline more closely.
Signs of a shallow topic
- The title is broad. Something like "Why blogging matters", with an outline that could sit on any blog.
- The outline repeats itself. Three sections that say the title again in different words.
- It promises what you cannot back. Words like "results", "the numbers behind" or "case study" when you have none to give. The writer will not invent them, so the post will come out thinner than the outline promised.
- It drifts outside what you do. For RankMill itself, a topic about tracking search positions would be a miss, because RankMill does not do that.
- It overlaps a post you already have.
The five-check approval routine
Run these in order. Each one takes a minute or two. If a topic fails a check, do not approve it yet.
1. Does it fit the profile?
Read the title and outline next to your website profile. Check for the right product, the right reader and the right tone. If the fit is off, the fix is usually in the profile, not the topic.
2. Does it match what the searcher wants?
Search the title yourself and look at the first page of results. Are they how-to guides, comparisons, definitions or product pages? If every result is a how-to guide and the outline is an opinion piece, the post is aiming at the wrong reader.
RankMill does not report search volumes or positions. So this check is yours, and your own eyes on the results page are the best tool for it.
3. Does it add something?
While the results are open, skim the top two or three. Then look at the competitor sites you added to the project. Ask one question: what does this outline say that those posts do not?
If the answer is "nothing", the post will struggle to stand out. If there is a clear gap, that is a good sign.
Check your own posts too. Open Blogs, search by title, and make sure you have not covered it already. Two posts that say nearly the same thing compete with each other. If they later contradict each other, an audit will flag it, but it is easier to avoid the overlap now.
4. Are the facts in place?
Read the questions and answer every one you can. Your answers go to the writer as facts from the business.
You can approve a topic with questions left blank. The writer will then write the post so the missing fact is not needed, rather than invent one. That is fine for a side detail. It is not fine when the whole outline hangs on that fact. A pricing comparison with no prices makes a short post.
If you want a firm rule, switch on Human in the loop for the project. Then a post is never written until its questions are answered.
5. Is it worth the credits and a slot?
Writing a post uses 10 credits. During early access your team gets 100 a month, shared across everyone. That covers 10 posts.
So ask yourself: is this one of the topics you would pick this month? Then open the calendar. Does the post fill a gap, or does it pile up next to three posts on the same subject?
Remember that hunting is free. If none of the five topics pass, run another hunt rather than approving a weak one.
What to do with a topic that nearly passes
Not every topic is a clear yes or no. Here is what to do with the ones in between:
- The fit is off: fix the website profile and hunt again.
- Facts are missing: answer the questions, then approve.
- You like the subject but not the outline: write from an idea instead. Give your own title and a short brief, and the writer starts straight away. You can add rules for that one post on top of the project's own. This spends credits the same way approving does.
- It duplicates a post you have: leave it unapproved.
If a hunt failed, open Activity. The run will carry a plain sentence saying why.
Keeping the gate when you automate
A team setting can approve every new topic automatically. It has a spending floor, so automatic writing stops while you still have credits left for the posts you ask for by hand. That is handy once your profile is in good shape and the topics have been reliably good.
It also removes the moment where you run these checks. So if you switch it on, move the routine to after the post is written:
- Read each post as it lands.
- Fix small things in the block editor, or tell the editor agent what to change.
- Run an audit now and then to catch posts that contradict each other or drift from your profile.
A quick version for your desk
Before you approve a topic, check that it:
- Fits the website profile.
- Matches what searchers want.
- Adds something the top results and your own posts do not.
- Has the facts it needs, or does not need them.
- Is worth one of this month's writes.
Five checks, a few minutes each. The posts you approve will be better for it.