Navi is Reelistic's proprietary ideal-customer-profile (ICP) model. It decides what a brand should create before a single piece of content is produced.
Generating hundreds of content ideas is easy. Knowing which ones are actually worth making is a different problem, and that is the problem Navi solves. Tested against the choices of real readers, Navi picked the idea the audience preferred 87% of the time, three times better than chance.
Navi works inside Axel, Reelistic's decision and learning system. The Axel research paper describes how Navi's choices connect to human corrections and audience response.
Why is choosing the idea the real problem?
AI has made content cheap to produce. A model can write fifty ideas for tomorrow's post in seconds. But most of them should never be made, and they fail in ways that are easy to miss. An idea can be:
- Creative, but irrelevant to the audience.
- On-brand, but impossible to tell apart from a competitor.
- Trending, but completely wrong for the brand.
- Attention-grabbing, but disconnected from the business's goals.
Each of those ideas can be produced, polished and published. Each will look fine in a content calendar. And each wastes a slot that could have reached the right people with the right message.
The choice also decides what a brand can ever learn. Most ideas in a batch are never published, so nobody finds out how they would have performed. If the wrong idea is picked, the better one never reaches the audience, and the data never shows it was better.
Before Navi, the AI that wrote the ideas also scored them. Every idea landed between 88 and 96 out of 100. When everything scores the same, picking the top score is close to a coin toss. A writer grading its own work tends to approve what it just wrote.
The question Navi asks
Most content tools ask:
Can we make this idea?
Navi asks:
Is this the right idea for this brand, this audience and this moment?
An idea isn't valuable just because it is creative or likely to get attention. It needs to resonate with the right people and serve a purpose. Navi is the step that checks, for every idea, before any time or money goes into production.
How does Navi judge an idea?
It thinks like the brand's customer
Navi reads who the audience is from the brand's strategy: its ideal customer profiles and how they behave with content, the brand's positioning, its markets and what it is trying to achieve on each platform.
Then it reads each idea the way that customer would meet it in their feed: the hook, the first seconds, the visual (an image, each carousel slide, or a video's shots), the caption and the call to action.
What it deliberately ignores matters just as much. Navi never reads the writer's own argument for why an idea is good. A judge that reads the sales pitch ends up agreeing with it. Navi judges what the customer would actually see.
It knows what each post is for
A post meant to reach new people, a post meant to teach and a post meant to sell have different jobs. Judging them all by one standard rewards the wrong things, so Navi judges each idea by the job of its slot.
| Content goal | Who meets the post | What success looks like | What counts against it |
|---|---|---|---|
| Reach | Strangers who share the problem but don't know the brand | They watch it through and send it to someone | Pitching the product, needing context, a flat mood |
| Value | People actively trying to solve the problem | They finish it and save it | A platitude, a promise the post doesn't keep, a pitch dressed as a lesson |
| Sales | People who know the brand and are deciding | They act | The product buried under a setup, generic claims, vague proof, no clear next step |
Talking about the brand kills a reach post and makes a sales post. Navi knows the difference.
It predicts what the customer does, not a score
For each idea, Navi returns:
- A verdict: what a typical customer does when the post reaches them. Scroll past, notice, engage, save, send, or act.
- A reaction: who the idea lands with and why, in the brand's language.
- The objection: the doubt the idea leaves standing.
- A rank within the batch, with no ties.
Reviewers see the reaction and the objection next to every idea, so they can see why it was chosen, not just that it was. When Navi kept flagging sales ideas with "doesn't say what it includes or what it costs", that pointed at a missing fact the brand could supply, not a weak idea.
It double-checks itself
AI judges tend to favour ideas in certain positions of a list. Navi judges every batch twice, the second time in reverse order, and combines the two rankings. How much the two passes agree is recorded, so the system knows how much to trust each ranking.
How good is Navi?
Against real audiences
The real test of an audience model is real behaviour. Navi was tested on 200 published headline A/B tests from the Upworthy Research Archive. In each test, 2 to 5 headlines were shown to more than 1,000 real readers, and one was the clear winner.
| Ranker | Picked the headline real readers preferred |
|---|---|
| Chance | 29.3% |
| Navi | 87% |
Navi found the winner in nearly nine out of ten tests, about three times the rate of picking at random. It did so before seeing a single click, from the audience description and the ideas alone.
On real content ideas
Navi was then run on real idea batches from the ideation system, for brands in six different industries, across reach, value and sales slots:
- It catches generic ideas. When a batch's best idea was rewritten to sound generic, Navi ranked it lower 18 times out of 18.
- It respects the job of the post. Sales pressure added to a reach or value idea pushed it down 12 times out of 12.
- It listens to what the audience has already done. When a brand's record showed its audience rewarded one approach and ignored another, the rewarded idea rose 2.7 places on average and the ignored one fell.
- It is consistent. Two independent runs agreed strongly and chose the same first idea 15 times out of 18.
- It changes the decision. Without Navi, the batch keeps the order it was written in. That order had no relationship with Navi's ranking, and the same first choice appeared only 4 times out of 18.
Reading the batches shows its taste. In reach slots, Navi set aside calm scenes and "come inside our space" posts that only interest people who already know the brand. In value slots, it set aside advice for a niche case. In sales slots, it set aside services named without saying what they include. In their place it chose a moment the viewer would send to a friend, a checklist the main customer would keep, and the product with a concrete reason to get in touch.
Over months of publishing
A ten-month simulation of 40 brands, using Reelistic's real planner and calendar, measured how often published posts carried the creative approach the audience really rewarded.
| Scenario at month 10 | Without Navi | With a mediocre ranker | With a strong ranker |
|---|---|---|---|
| Strong audience preference, daily posting | 59% | 89% | 100% |
| Moderate preference, daily posting | 31% | 50% | 70% |
Even a mediocre ranker helped in every scenario and hurt in none. A strong one put the winning approach into nearly every post. That is what picking right before production buys: the feed starts doing what works months earlier.
What this changes for a brand
- Less wasted production. Time, design and video budget go into the idea most likely to land, not the first plausible one.
- Posts with a purpose. Every post is judged by the job it has to do, so reach posts travel, value posts get saved and sales posts sell.
- Decisions you can read. Every choice comes with the customer's reaction and the objection, so a reviewer can agree, disagree or fix the gap.
- Faster learning. Better choices put better ideas in front of the audience, and Axel learns from what the audience does with them.
Why the audience still has the final say
Navi is very good at predicting what a customer will do. The audience is still the one who actually does it. So inside Axel, real results always outrank Navi's predictions:
- Navi's verdicts are recorded separately from real outcomes and count for less.
- They stay with the brand. Navi judges one brand's audience, so its verdicts are never shared with other clients.
- The audience overrules it. When enough real results for an approach disagree with Navi, Navi's view on that approach is dropped.
- Its picks are tested. Some slots go to ideas Navi ranked lower, so its judgment is checked against real response instead of assumed.
This matters more than it sounds. In the simulation, a ranker left unchecked kept publishing its own favourite, and the approach the audience really rewarded fell to 21% of posts, below the 32% it reached with no ranker at all. With the audience in charge, Navi makes the system faster without letting it fool itself.
Not just social posts
The same model ranks ideas for email and blog articles, judged by what each channel's audience does. For email, it predicts whether subscribers ignore, open, read, keep, forward or click. For blog topics, it judges a reader who arrived from a search and didn't ask to hear from the brand. A topic whose honest answer sends the reader to someone else's business is ranked last.
Knowing what to create is the real advantage
Content generation is becoming cheap. Every brand will soon have access to more ideas than it could ever publish. What will separate them is not how much they can make, but how well they choose.
At Reelistic, our AI systems don't just produce more content. They make better decisions about what deserves to be created in the first place. Navi is where that decision happens.
Related Reading
- What Is Axel? The AI Marketing System That Learns From Feedback
- How AI Should Learn From Marketing Decisions and Results
- How to Build a Growth System Across SEO, Social, Email and Paid Ads
- What Is Radar? How AI Finds the Growth Channels That Fit Your Business
Frequently Asked Questions
What is Navi?
Navi is Reelistic's proprietary ideal-customer-profile (ICP) model. It ranks content ideas for the customer described in a brand's strategy and the goal of each post, and decides which idea is produced before any content is made.
How accurate is Navi?
On 200 real headline A/B tests, each shown to more than 1,000 readers, Navi picked the headline the audience preferred 87% of the time, against 29.3% by chance.
Does Navi replace human review?
No. Navi chooses and explains: it shows the customer's likely reaction and the objection each idea leaves. People still approve, edit or reject the work, and their corrections feed Axel.
Why doesn't the AI that writes the ideas also rank them?
When the writer scored its own ideas, every idea landed between 88 and 96, so the top score was close to a random pick. A separate judge that never sees the writer's justification produces a ranking that actually separates good ideas from weak ones.
What is the difference between Navi and Axel?
Navi decides which idea is worth making for a brand's customer. Axel is the wider system that connects that choice with human corrections and measured audience response. The Axel research paper describes both.