Answers to the questions people actually ask

HumanCulture reads the social content a talent has already published, works out what is driving performance, and turns that into direction you can brief against.

It does three things: it analyses published content and tells you which themes, formats and posting patterns are working; it lets you test a caption, image or video before it goes live; and it gives you Jasper, an AI creative strategist you can ask questions of in plain English, grounded in that talent's own data.

Yes. There is a See demo option on the landing page that opens a fully populated example profile, so you can walk through a real report without connecting anything.

New workspaces also get a demo talent by default, so there is something to look at while your own first report is still generating. You can delete it once you no longer need it.

They are the three levels everything sits inside.

• Workspace — your shared environment. It holds your members, your subscription and billing, your talent and your content. One person can belong to several workspaces.
• Folder — an optional grouping inside a workspace, for organising your roster however you like. Folders can be shared as a unit.
• Talent — one person or brand you are analysing. Reports, content tests and Jasper conversations all sit at talent level.

There are two kinds of talent: ones you manage, where the socials are properly connected and you get the full report, and ones you simply follow, which only need a username.

1. Go to Roster and choose New talent.
2. Give it a name and follow the prompts to activate it.
3. Connect the socials using the Instagram and TikTok buttons at the top right of the talent.
4. Leave it to generate. You do not need to stay on the page.

Instagram and TikTok. Both connect through the platform's own official authorisation screen — you sign in with the account as normal and approve access there.

No, and you should not ask for them. Send an invite instead: the talent gets an email with a link, opens it on their phone, signs into Instagram or TikTok themselves and approves access. Nothing is shared with you at any point.

They do not need an existing HumanCulture account, and they never see your billing.

Around three days. After that it expires and you will need to send a fresh one — which takes a couple of seconds from the talent page.

Expiring links are deliberate: they keep an unopened invite from sitting live in someone's inbox indefinitely.

Instagram and TikTok both expire the permission they give us after a set period. This is their policy, not a fault on our side, and it will happen periodically on every connected account.

When it does, open the talent, click the Instagram or TikTok button at the top right, and reconnect from there. The talent signs in and approves. Historic data is not lost — collection simply resumes.

Up to twelve months of published content. Everything within that window is analysed together, and each post is stored with the exact time it went out rather than just the date, so posting-time analysis is accurate to the hour.

A first full profile analysis usually takes one to two hours, depending on how much content the talent has. Content tests in the Content Hub are much quicker — typically five to ten minutes.

Treat both as guides rather than promises. The work depends on how quickly Instagram, TikTok and the AI respond on the day, and that varies.

You do not need to wait on the page. Leave it, do something else, and come back.

• Performance — how the talent is trending, volume of content over time, and where the gaps are.
• Formats — how Reels, carousels and static posts compare.
• Themes — what the content is about and how it looks, ranked by engagement.
• Timing — when this audience actually responds.
• Audience — what people are saying in the comments, grouped by topic.
• Recommendations — what to do next, tied back to the evidence above.

Two things drive a post's performance independently: what it is about and how it looks. Collapsing those into a single label like "authentic content" sounds analytical but tells nobody what to shoot. So we classify both, then combine them.

The formula table is the headline of a report. Most talent settle into roughly fifteen to twenty combinations, and that set is specific to them — it is the closest thing a talent has to a content fingerprint. Narrative and visual are also shown as separate tables if you want to dig in.

Engagement rate is measured against follower count, and worked out the same way for every post, every format and every talent. That consistency is the point: it is what lets a theme, a format or a time slot be compared against another honestly.

Where a report compares groups of posts, it uses the median rather than the average, so one runaway hit does not make an entire theme look stronger than it is.

Almost always because no interactions came through for that post in the data we receive. The usual causes are trial Reels and posts that were later taken off the main grid.

Trial Reels carry a second quirk worth knowing about. If the Reel is deleted afterwards, its link stops resolving to the original post — so following it from a report can land you on something unrelated, and sometimes on content that has nothing to do with that talent at all. The row itself is not wrong; the post it pointed at is simply gone.

If you can see engagement on a post in the app but the report shows zero, that is worth flagging to us — send the post link.

Because there is not yet enough content to say anything honest. Timing recommendations only appear once a meaningful share of a talent's posts fall into the relevant slots; below that threshold a finding is either marked as directional or withheld.

We would rather show you less than have you brief a shoot against noise.

No. Everything in a talent's report and recommendations is derived from that talent's own content and audience. Nothing from another workspace, another client or a wider pool is blended in.

The AI does the reading. It watches the video and looks at the images to work out what is actually happening in them, sorts captions and comments into themes, and writes the narrative you see in a report. Jasper then answers your questions against that same material. The analysis runs on Anthropic's models.

We do not build or train a model of our own on your content. We do keep the data we have collected and analysed — that is what makes month-on-month comparison possible, and what Jasper reads when you ask it something.

If you need the specifics for a client or a legal review, ask us and we will set them out properly.

Jasper is the AI creative strategist built into the platform, opened from the top bar on any talent. Think of it as the analyst who has already read the whole report and will answer follow-up questions in plain English.

Anything you would ask a strategist looking at the same numbers. For example:

• “Give me a performance summary for the last month — what went up, what went down, what worked.”
• “Which posts over-performed this month, and which underperformed? Include the links.”
• “Draft me a month of posting ideas based on what is working.”
• “Why is engagement down on carousels?”

Jasper will only answer from evidence in the report in front of it. If the analysis it needs has not finished — or the talent does not have enough content for that particular question — it says so rather than inventing an answer.

Usually it will also tell you what is missing. If a report has fully generated and you still hit this on a straightforward question, let us know what you asked.

No. Jasper answers from the report for the talent you have open. It cannot see other workspaces, other clients, or accounts you have not connected.

Testing content before it goes out. You bring in a post you are considering, and the Hub reads it against everything that talent has already published — where it sits against the formulas that work, and what to change.

It keeps a list of recent files so you can pick up where you left off, and global search covers both files and talent names.

Captions, images and video. You can test a caption on its own if the visual is not cut yet — the narrative and visual sides are assessed separately, so a partial post still returns something useful.

No. On a re-analysis only genuinely new material — a changed caption, a new image, a new video — is processed. Anything already analysed is reused, so a second pass should come back faster than the first, though how much faster depends on how the platforms and the AI are performing that day.

• Admin — everything in the workspace, including every talent, plus members and billing.
• Editor — can work on the talent and folders they have been given, and nothing else.
• Viewer — can read the talent and folders they have been given, and nothing else.

Editors and viewers see only what has been explicitly given to them. Adding someone to the workspace alone does not expose your roster.

Yes, and this is the normal case. If you manage twenty talent and a member only works on two, invite them to those two — individually or as a folder — and the other eighteen stay invisible to them.

Pricing is per talent, per month, and the rate comes down as the roster grows. Your first three months are charged at an introductory rate; the settled rate applies from month four.

Rosters of a hundred or more are priced individually — get in touch and we will put something together.

Yes. A one-time onboarding fee is charged per talent when it is first set up — once, not monthly. It uses the same roster bands as the monthly rate. Founding member rates:

Click your workspace name in the top left, then Manage workspace → Plan & billing. Card details, invoices, plan changes and any discount code are all handled there.

No. You are billed for the talent in your workspace, not the people looking at them. Members you invite as admins, editors or viewers cost nothing, and a talent connecting their own socials never enters payment details.

Tell us — this is the feedback we most want. Send the talent's name, the section, what you are seeing and what you expected, with a link to the post if it relates to one.

Data accuracy is the thing the whole platform rests on, so a report that disagrees with what you can see in the app is treated as a priority, not a niggle.

Email hello@humanculture.com and we will come back to you. Include a screenshot where you can — it usually saves a round trip.