What it does
Buyers ask questions in public that you cannot credibly answer, because you are the vendor. A customer answering the same question carries weight you do not have.
This feature finds those conversations and routes them to advocates as suggestions. It is a prompt, never an assignment. Advocates decide whether to engage and what to say.
Before you start
- Role. Owner and Admin configure sources. Owner, Admin, and Manager review runs. Viewers can look at the queue but cannot act on it. These rules are enforced.
- Configuration is required. Nothing can be prepared until your sources are saved.
- US English only in this release.
- You need active advocates. Publishing to an empty roster achieves nothing.
How runs get made
This part is not self-serve, and it is the thing most people assume wrongly.
Saving your sources does not start a run. Your Kindling contact prepares each one. You configure the inputs, they run it, and it appears in your queue for review. Each run produces 20 conversations.
So the sequence is: save your sources, tell your contact you are ready, review what comes back.
Configure your sources
Settings → Topics. Collected once, and they decide what gets surfaced.
| Input | What to put in it | Limits |
|---|---|---|
| Company domain | Your own domain without https://, for example acme.com |
Required |
| Product category | How your buyers describe what you sell, not your positioning | Required |
| Competitor domains | One per line. Coverage is measured against these | 3 to 10 |
| Buyer roles | One per line. Job titles of the people who evaluate and buy | 3 to 6 |
| Customer base context | Five to ten lines of free text. A soft signal, never a filter | Optional, up to 4,000 characters |
| Per-run spend cap | A run stops rather than exceed it | $1 to $500. Default $15 |
Product category is the field people get wrong. Use the buyer's words. If you call it advocacy-led growth and buyers call it customer marketing, write what they would type into a search box.
Customer base context wants specifics: what kinds of companies use the product, who owns it day to day, the two or three use cases that come up most, anything customers measurably improved, and what they commonly switched from. It nudges relevance rather than filtering, so a thin answer here produces a vaguer run.
Per-run spend cap bounds what a run costs. A run stops rather than exceed it, and cached requests are never re-billed, so a resumed run costs nothing extra.
Review and publish
Nothing in a run is visible to advocates until you publish it. Publishing is a separate step from approving, and it is one conversation at a time.
The queue groups each run's conversations into Awaiting review, Held, Approved, and Decided. From Awaiting review or Held you can:
- Approve. Clears the conversation for publishing. It is still invisible to advocates.
- Edit and approve. Rewrite the question in your own words, then approve. The edited question is what advocates see.
- Hold. Park it without deciding. Held conversations stay invisible.
- Reject. Requires one of seven reasons, plus an optional note.
From Approved, one more click publishes the conversation to your advocates. There is no bulk approve and no bulk publish. Every published conversation carries one deliberate click.
Bulk reject is available for clearing an off-target run: select the conversations, choose one reason, up to 100 at a time.
| Rejection reason | When to use it |
|---|---|
| Factually wrong | The premise of the question is inaccurate |
| Helps a competitor | Answering honestly advantages someone else |
| Legal or regulatory risk | Regulated claim, contract term, or open matter |
| PR risk | True but reputationally costly to discuss |
| Off strategy | Real topic, not one this company wants owned by customers |
| Nobody can speak to it | No customer in the base has this experience |
| Poor quality | Badly written or incoherent |
Reject anything where the honest answer is not your product. An advocate pushed into a conversation they do not believe in produces the one thing this model exists to avoid: a post that reads like an ad.
Run history and cost
Run history records what each run produced, what you did with it, and what it cost: conversations, published, approved, rejected, awaiting, keywords, SERP calls, and spend split between data and LLM calls.
Use it to tune your inputs. A run with a low approval rate means the sources are wrong, not the feature. Usually a product category that is too broad or competitor domains that attract a different buyer.
What advocates see
Advocates see published conversations under Ideas to talk about, as a plain list of questions. None of the review detail travels with them: no ranking, no demand figures, no evidence line, no competitor data. No obligation, no deadline, no assignment. They may engage in their own words or ignore the list.
An advocate sees the published conversations of every brand they are approved in, together in one list. No email or notification is sent when you publish.
If your program treats these as expectations, you get compliance-shaped content and lose the credibility you are paying for.
FAQ
Why can I not start a run myself?
Runs are prepared by your Kindling contact in this release. Save your sources and tell them you are ready.
Does this post on my advocates' behalf?
No. Nothing is posted automatically. Advocates write and publish themselves.
Do advocates have to act on them?
No. They are suggestions.
Can I edit a conversation before publishing?
Yes. Use edit and approve to rewrite the question, then publish it from the Approved group.
I approved a conversation. Why can advocates not see it?
Approving is the first step. Publish it from the Approved group to make it visible.
Does a run cost extra?
Runs have a cost, bounded by the per-run spend cap you set. Run history shows what each one cost.
Does engaging with one count as a submission?
Only if the advocate submits the post for review like any other. See Review and approve submissions.
To get a run prepared, or if your queue is empty after saving sources: support@kindlinghq.com.