How to brief a dumub report
A report is only as good as the search behind it. Most of that search is planned for you: before anything is bought, dumub's planner researches the brand and decides where to look. Your part of the brief is smaller, and it's the part only you know: which markets matter, over what period, what the report is for, and anything too new or too local for the planner to find.
This guide covers what the planner does, what to give it, and how to choose what gets read. Once the report is back, How to read a dumub report picks up from there.
Start with the question
Before you type anything, write down what you need to know, in a sentence or two. A good brief is a question about what people said and why, not a request for a mention count. For the public Robert Pattinson as Bond sample, it might read:
> How are people in Southeast Asia reacting to Robert Pattinson as James Bond? What share are for him, against him or joking, and why? Do the markets differ?
There's no box for the question itself. It decides everything you do enter: which markets, how far back, which extra words, and how you answer the quick check. Two tests help. Ask for reasons, not only the split ("what are they objecting to, and is it the same in every market?" gets you a finding; "are people positive or negative?" gets you a score). And know what the report is for: a decision, a campaign readout and a slide for Monday's meeting need different depth, even when the comments are the same.
What the planner does before you pay
When you search a brand, the planner reads the web and the platforms first, then writes the plan from what it found:
- the names people actually use, including product names that stand on their own and the consumer brands behind a parent company
- the hashtags local posts carry in each market, and the queries people there would type, in their own languages
- the brand's own local accounts and the creators who post about it, with the posts that point to them
- campaigns, launches and controversies in your window that press coverage dates
- what counts as the brand for this run, and where it knows it can't reach
You see all of it before you pay: the campaigns in the quick check, the rest under "What we looked for" on the proposal. So you don't need to list every product name, hashtag or spelling yourself. Give it what it can't find, and check what it assumed.
1. Search the brand
Type the brand as you'd name it. The planner works out the names people use around it, so a search for a parent company still finds the brands people actually write about.
Search a single product only when the report is about that product alone. A search for the brand already covers its products.
Already know the posts you want read? Paste the links instead, one per line, or add them from a .csv or .xlsx file. dumub reads those and skips the search.
2. Pick your markets
dumub covers Singapore, Malaysia, the Philippines, Thailand, Indonesia, Vietnam and India. "Southeast Asia" picks the six Southeast Asian markets in one tap. Picking none means worldwide.
Each market you add is searched in its own languages, with the queries people there would type. A post counts as being in your market only when its content shows it (the language, the place, the account behind it), not because a search for that market happened to find it. So adding a market doesn't pad the report with posts from somewhere else.
Multi-market reads are good for comparing how the same launch or campaign landed in each country, and for finding where the conversation actually lives. A narrow read helps when you need depth: one report holds up to 100 sources, so with fewer markets more of them go to each one, you hear more local voices per market, and the proposal is ready sooner (finding takes about 12 minutes per market).
Pick the markets you'll act on. A market added out of curiosity still takes time to search and room in the report. And expect them to disagree: the Bond sample's six markets each got their own card in the report, because a regional average would have hidden what made each one different.
3. Choose the window
The window runs from today back: 24 hours, 3 days, 7 or 14 days, 30 or 60 days, 3 months, 6 months or a year. You can also start from a date.
Match it to the question:
| What you want to know | Window that fits |
|---|---|
| How a launch, ad or announcement landed | 7 or 14 days from the day it went out |
| How a campaign is doing | 30 or 60 days |
| What people think of the brand overall | 3 or 6 months |
| What changed over a season | 6 months or a year |
Every window ends today because the platforms can be asked how recent a post is, not which dates it falls between. If the moment you care about was months ago, start the window just before it.
A tight window keeps the conversation you care about apart from older talk that only shares a name. The Bond sample read the two days after the casting news, which is why its comments are about the casting and not a decade of arguments about his earlier films.
4. Add the words your audience uses
"Also look for" takes up to eight words or short phrases. The planner finds what's public, so use these for what it can't:
- a campaign or launch too new to have press coverage yet
- a nickname or spelling you've seen your customers use
- a rival you want compared, if the comparison is the point
- the angle you care about ("delivery", "price", "after-sales")
Your words are searched ahead of the planner's own, and it turns each one into what people in each market would type, so you don't need to translate them. Words that only describe what you want, like "complaints", still shape the plan even when they aren't searched on their own.
Leave out offer words like "sale", "discount" or "promo". They find shops and bank promotions with a handful of comments, not people talking.
5. Leave out what shares the name
"Leave out" takes up to five words. The planner already separates the obvious things that share your brand's name, and asks when it isn't sure. Use "Leave out" for the ones you know cause trouble, or topics you don't want in the report. A kitchen brand called Ninja shares its name with a motorcycle, a cartoon and a video game; "Kawasaki" or "Turtles" keeps those away.
Posts that carry a word you left out aren't deleted. They move below the cut, tagged with the word that put them there, so you can see what was left out and tick any back.
6. Check the planner's work
Sometimes the planner stops to show you what it found and ask a few things before it searches. Each is optional, and the skip button runs the plan with its defaults.
- Campaigns and moments it found for the brand, with dates where press coverage stated one. Untick any you don't want searched; conversation about a campaign usually lives under the campaign's name, not the brand's.
- Your words and how they'll be searched, market by market. Check that each one turned into something a local would type.
- One or two questions about what the report is for. For a brand that puts out shows, products or collections under their own names, it asks whether to read the brand itself, the things it puts out, or both. For a brand whose own accounts post a lot, it asks whether you want people talking among themselves, reactions under the brand's own posts, or both.
- "Something else we should look for?" Add a campaign, a collab, a recall, anything you know about that the list missed.
These answers matter beyond the search. dumub writes a short brief from them (who the report is for, what they own, what they want to know), and every later step reads it: the screen that decides what counts, the reading of each comment, and the write-up.
7. Read the proposal before you pay
Finding and pricing the conversations takes about 12 minutes for one market, longer with more. The proposal shows what was found before anything is bought.
Start with "What we looked for", the planner's own notes. It says what counted as your brand, what was searched, what dumub assumed without asking, and where it knows it can't reach. If an assumption is wrong, that's the moment to search again, not after the report.
The overview tab shows what each market held and the early themes in the sources selected so far. The early themes come from titles and captions, so read them as a preview of what the report will cover, not as findings.
If the run comes back thin in your markets, you'll be offered two ways forward: add the biggest worldwide conversations, or search again with a longer window and the names, hashtags or spellings the first search missed.
8. Choose the sources
The four cards at the top set how much to read. Narrow, Balanced and Comprehensive cover about half, three quarters and nine tenths of the comments dumub recommends reading; Everything takes the whole shortlist, up to 100 sources in one report. Each card shows the comments, conversations and markets it would cover.
Then go through the list. The tabs split it by platform, and each row has a box you can untick. The tags under a title tell you what kind of conversation it is:
| Tag | What it means | When to keep it |
|---|---|---|
| Brand-owned | posted by the brand's own account | when you want reactions to the brand's own posts |
| Paid creator | a sponsored or affiliate post | when the campaign's paid reach is part of the question |
| Retailer | a shop or reseller selling the product | when price and stock talk matters |
| Your word | found by one of your keywords | usually; it's what you asked for |
| No brand mention | the text never names the brand | only if you know the post is about it |
| You left out | carries a word from your Leave out list | only if the match was a false alarm |
"In-market only" unticks every worldwide and wider-field source in one go, so the report reads only conversations from your markets. Below the list, "The rest of the field" holds relevant conversations that fell under the cut; add any of them by ticking. You can also add your own links to the run.
A good basket for most briefs: the biggest conversations in each of your markets, organic posts first, brand-owned posts only if their comments are part of the question.
Watch the comment counts as much as the number of posts. A news article with no comments can show how the press framed a story, but it can't tell you what audiences said. If a market comes down to two or three small posts, keep them if that market matters to the decision, and read its card as a lead, not a verdict.
9. After you approve
A typical report takes about 30 minutes to read and write. Before it's delivered, it goes through a final check of its own labels, titles and method; if something doesn't pass, the order shows "final check" while it's fixed.
Each report comes with a chat. Ask it anything about the comments; How to read a dumub report covers what to look at first. If it finds that something in the report is contradicted by the comments themselves, it offers "Rebuild the report with this fixed". That's free, and it changes only the passages that were wrong.
A brief on one page
| Field | What to put | Common mistake |
|---|---|---|
| Your question | what people said and why, and what the report is for | asking only for a positive/negative split |
| Search | the brand as you'd name it | listing every product and hashtag yourself (the planner does that) |
| Markets | the markets you'll act on | adding every market "just in case" |
| Window | sized to the question, ending today | a year for a launch-week question |
| Also look for (up to 8) | what's too new or too local for press, the angle you care about | offer words like "sale" |
| Leave out (up to 5) | names you know cause trouble | leaving it empty for a common word |
| Quick check | what the report is for; untick wrong campaigns | skipping it when the brand has many products |
| Sources | the biggest conversations in your markets | ticking everything without reading the tags |
FAQ
Do I need to list every product name, hashtag and spelling?
No. The planner researches those before it searches and shows you what it found. Add only what it couldn't know, like a campaign that launched this week.
Do I need to translate my keywords into each market's language?
No. The planner turns each word into what people in each of your markets would type, in their language, and shows you the result in the quick check.
What happens to posts that match a word I left out?
They aren't deleted. They sit below the cut, tagged with the word, and you can tick any of them back into the report.
Why can't I pick a window that ended months ago?
The platforms can be asked how recent a post is, not which dates it falls between, so every window ends today. Start the window just before the moment you care about.
How many sources can one report hold?
Up to 100. The Everything card stops there, and the proposal tells you if you've ticked more.
Can I change the brief after I order?
The search and the sources are fixed once you approve. If the report gets something wrong that the comments contradict, ask the report chat and rebuild that part for free.
What if I already have the posts I want read?
Paste the links, one per line, or add them from a .csv or .xlsx file. dumub reads them without searching.
This page is published by dumub.