How to read a dumub report
A dumub report gives you a score in the first second and the reasons behind it if you keep reading. Most of the value is in the reasons. This guide walks through a report in the order that gets you to them, using a public example: Robert Pattinson as James Bond, a real dumub report on how six Southeast Asian markets took the casting, read in the two days after the news. It's a sample, not client work. To brief one of your own, see How to brief a dumub report.
The score is where you start, not where you stop
The Bond report scores 39 out of 100, leaning negative, where 50 is even. Look at what it's built on before you quote it: 1,760 of the 2,887 comments scored for tone are about Pattinson and take a side, and of those, 54% are negative, 13% mixed and 33% positive.
All 5,516 comments were read and sorted into themes. Tone is scored on a sample from every theme (all of a theme's comments when it has 30 or fewer, up to 95 from the biggest), so small themes are scored as well as big ones. The report's methodology says exactly how many were scored and how.
A score tells you which way the conversation leans. It doesn't tell you why, or what to do about it.
Topics show where the score comes from
Under the same 39, the topics pull in different directions:
| Topic | Share of mentions | Tone toward Pattinson |
|---|---|---|
| The casting itself | 41% | 54% negative |
| His acting | 6% | 60% positive |
| Twilight | 6% | 64% negative |
| The Henry Cavill comparison | 2% | 93% negative |
People who talked about his acting mostly liked it. People who talked about Twilight mostly didn't. Those need different answers, and one number would have mixed them together.
Narratives say what people actually argue
The narratives go a level deeper. They're named from what the comments say, ranked by how much of the discussion they hold. The Bond report found seven. The largest, at 40%, isn't "he can't act". It's that he doesn't fit the Bond archetype: commenters wrote "kurang macho" and "not masculine enough". The Twilight narrative carries its own line, "James Bond rasa vampire" (James Bond with a vampire flavour).
Look for the defence that's already in the comments, too. The report found fans pointing out that Daniel Craig was mocked in 2005 and acclaimed later. An argument your audience is already making usually lands better than one you write for them.
Read each market on its own
When a report covers more than one market, each market gets its own card in "Market voices": its tone, how many posts it rests on, and a short summary of how that market's conversation differs. The Bond report has six, and they don't agree.
Before you act on a card, check the post count. A market read from two posts can point you somewhere, but it isn't settled. A regional average hides both problems: it blends markets that disagree, and it gives a thin market the same weight as a deep one.
Check which comments people backed
The report ranks comments by likes. In the Bond report, five comments hold 18% of the 15,993 likes on every comment in the file. The most-liked, with 864 likes, is a joke that opens "the batman, the odyssey, dune 3, primetime…" and runs through his roles on the way to Bond.
Jokes travel further than reasoned criticism, and they can cut either way. If the comments people backed most don't match the finding you're about to present, find out why before you present it.
Open the comments file
Every report comes as a zip you keep: report.html, slides.html, a deep-dives/ folder, quotes.xlsx, comments.xlsx, a data/ folder and START-HERE.html. There's no login to open it and nothing expires.
The write-up shows a sample of comments under each narrative; the rest are in comments.xlsx. To check a finding, filter its theme, open a few permalinks, and read the originals beside their English lines. If the comments don't support the heading, don't use the heading.
You can also ask the report questions. The answers are worked out from the comments and come back with the comments behind them, translated and linked. Use it for follow-ups ("what do Malaysian commenters say about the price?"), and if it finds the report contradicted by the comments, it offers to rebuild that part for free.
Turn the reading into a decision
Four habits carry over to almost any report:
1. Answer the objection you actually have. If people praise the craft and reject the image, a defence of the craft misses. 2. Use the defence that's already in the comments before writing a new one. 3. Don't send one reply to every market when the cards disagree. 4. Check the most-liked comments and the thin markets before you brief anyone on the score.
Then take it where it has to go. slides.html is the same report as a three-slide deck (the read, the topics, the themes), and every quote in it is cited.
FAQ
Why shouldn't I stop at the overall score?
Because it mixes different objections together. In the Bond sample the score was 39 out of 100, but people talking about his acting were 60% positive while people talking about Twilight were 64% negative. The score said people weren't sold; the topics said why.
What's the difference between a topic and a narrative?
A topic is what a comment is about, like the casting or his acting, with its share and tone. A narrative is an argument people make across topics, named from their own words, like "he doesn't fit the archetype".
How do I check a finding?
Open comments.xlsx, filter the finding's theme, follow a few permalinks and read the original comments beside their English lines.
Can I trust a market card with only a couple of posts?
Treat it as a lead. Check how many posts the card rests on, and if that market matters to the decision, brief a deeper read of it.
Can I ask the report questions after it's delivered?
Yes. Every report has a box for questions; answers are worked out from the comments and cite them. If an answer shows the report got something wrong, you can rebuild that part for free.
Try it on the Bond walkthrough, or open the full Bond report. Other samples are at dumub.com/demo.
This page is published by dumub.