You already have the dashboard. It has the right numbers on it, it updates on its own, and nobody reads it. That isn’t a data problem or a design problem — it’s a structure problem, and the fix isn’t to build something else. It’s to give the report you’ve an order.
This covers the narrative frameworks that actually apply to marketing reports, which one to use in which situation, and how to restructure an existing Data Studio report so a story has somewhere to live.
Every narrative framework is doing one of three jobs
There are more narrative frameworks than anyone needs, and most articles list them without saying when to use which. Underneath, they only do three things:
- A change-shaped story. Something was true, something happened, something is now true. Three-act, Freytag, Dykes’ arc.
- An answer-first hierarchy. The conclusion, then the support, in descending order of importance. Minto, SCQA, BLUF, the inverted pyramid.
- An accountability ledger. What was agreed, what was done, what resulted. STAR and its relatives.
So the useful question isn’t “which framework is best”. It’s which of those three jobs your report is actually doing — because a monthly retainer report and a renewal pitch are doing different ones.
Three questions that settle which framework your report needs
Three questions decide the structure, and the first eliminates about half the options on its own.
- Are you in the room? If yes, you can withhold the answer and build to it. If no, the answer goes first — because a reader who abandons the page halfway should still leave correctly informed rather than half-informed.
- Is there one finding or many? One finding wants an arc. Nine want a pyramid.
- Are you reporting on the world, or on someone’s behaviour? The world wants a conflict. Behaviour wants a ledger.
Almost every mistake in report structure is answering question one wrong — building a slow, well-crafted narrative arc into a document somebody reads alone on a phone.
The seven narrative frameworks worth knowing for marketing reports
Setup, conflict, resolution — Scott Berinato
The one most people mean by “data storytelling”. Setup is the reality that exists; conflict is the event that changes it; resolution is the new reality, plus what to do about it. Without the conflict there’s no story — that’s the whole load-bearing idea.
Credit where it is due: the data version of the triad is Scott Berinato’s, set out in Good Charts and in his HBR Quick Study on telling stories with data in three steps. It’s routinely misattributed to Jim Stikeleather’s 2013 HBR article How to Tell a Story with Data, which is about Tufte’s definition of an excellent visualisation and five audience types, and contains no such structure. The underlying dramatic triad is far older and belongs to nobody — Aristotle, then Freytag, then Syd Field’s three acts.
Good for: a month where something genuinely changed, and any conversation you’re present for.
Bad for: documents read without you — the answer lands last. Also bad for a quiet month, because it tempts you to manufacture a conflict, which is the failure mode worth being most careful about.
SCQA and the Minto Pyramid — Barbara Minto
Situation → Complication → Question → Answer, from Barbara Minto’s Pyramid Principle — the structure McKinsey has run on since Minto joined the firm in 1963 as its first female MBA hire and was asked to fix its writing. One naming point, since precision is the whole business of this post: Minto’s own term is the SCQ framework, with the Answer sitting at the top of the pyramid rather than as a fourth step. “SCQA” is near-universal shorthand, but it isn’t hers.
It’s the same skeleton as setup/conflict/resolution with two changes that matter enormously for reporting. It makes you name the question your client is actually asking, rather than the one your data happened to answer. And it puts the answer first, with support beneath it in descending order of importance.
For a recurring retainer report this is the winner, for a mundane reason: the Situation is stable month to month. “You’re running a content programme aimed at non-brand organic revenue” doesn’t change in March. You refill one line rather than rewriting the story, which is what makes a structure survive contact with a monthly deadline.
The full Minto Pyramid — grouping findings into mutually exclusive, collectively exhaustive sets — is the one framework here that handles nine parallel findings without collapsing into a list. That makes it the right choice for a one-off audit, where the alternative is something that reads like a bug tracker.
The Data Storytelling Arc — Brent Dykes
Brent Dykes’ structure from Effective Data Storytelling: setting and hook → rising insights → aha moment → solution and next steps. Dykes is also the source of the data / narrative / visuals model that the guide to data storytelling for marketers opens with.
This is the one for a live presentation, and the reason is about pacing rather than content. Rising insights let the audience arrive at the conclusion a half-second before you say it — which is the difference between someone believing a finding and merely accepting it. That only works when you control the pace, so it fails completely in a document.
The Sparkline — Nancy Duarte
Nancy Duarte’s sparkline structure, from Resonate, alternates between what is and what could be, ending on what she calls a new bliss. She derived it by analysing great speeches — King’s “I Have a Dream” and the Gettysburg Address are her two worked examples — and it is engineered to move an audience to a position they don’t currently hold.
Which makes it the pitch and renewal structure, because the gap between what is and what could be is literally the thing you’re selling. It’s the wrong shape for routine reporting — used monthly it reads as perpetual dissatisfaction.
BLUF and the inverted pyramid — newsroom practice
Not really frameworks so much as governing rules for anything read without you. Bottom line up front: state the conclusion in the first sentence. The inverted pyramid, from journalism: everything after that in descending order of importance.
Together they govern the executive summary page of a report. The test is whether abandoning the page halfway leaves the reader correctly informed.
One aside, since it comes up constantly: the story that the inverted pyramid was invented because Civil War telegraph lines kept failing isn’t established fact. Historians have disputed its origin for decades — David Mindich dates it to the coverage of Lincoln’s assassination in 1865, while Horst Pöttker’s content analysis of the New York Herald and New York Times finds the form only became commonplace between 1880 and 1890, two decades after the war. Civil War reporting itself was still largely chronological. The telegraph story is repeated everywhere, which is a useful reminder that a good origin story spreads faster than a true one — a hazard this whole subject should stay alert to.
STAR — Development Dimensions International
Situation → Task → Action → Result. DDI dates the STAR method to 1974, when it introduced it alongside its Targeted Selection interviewing system — that’s the company’s own account of its history rather than an independent record, which is worth saying plainly.
One adaptation is ours and should be labelled as such: DDI treat Situation and Task as a single combined element. Splitting Task from Action into separate columns is the move that makes STAR work for a client report, because the gap between the two is the entire point.
That’s exactly what you want for the hardest conversation in agency reporting: the client hasn’t implemented what you recommended. Build it as a table — what was agreed, what was done, what resulted — and the empty Action cells make the gap a structural feature rather than an accusation you’ve to voice. The table starts the conversation for you.
Which narrative framework to use in which reporting situation
| Situation | Use | Why |
|---|---|---|
| Monthly or quarterly retainer report | SCQA | The Situation is stable, so you refill one line. And it forces you to name the client’s real question. |
| One-off audit or analysis | The Minto Pyramid | The only structure that handles many parallel findings without becoming a list. |
| Bad news — performance is down | Setup / conflict / resolution, but state the number in sentence one | The shared setup makes the drop an event rather than a verdict. Never make them hunt for the number. |
| Bad news — the client hasn’t implemented | STAR | Task and Action in separate columns makes the gap structural, not accusatory. |
| Pitch or renewal | Duarte’s Sparkline | The what-is / what-could-be gap is the thing you’re selling. |
| Executive summary page, read without you | BLUF + inverted pyramid | Abandoning it halfway should still leave them correctly informed. |
| Live presentation, you’re in the room | Dykes’ Arc | You control the pace, so they can arrive at the answer just before you say it. |
| The multi-page dashboard itself | The martini glass | See below. This is the one that changes how you build. |
The martini glass, the one structure designed for an interactive dashboard
Every framework above was designed for a document or a speech. Only one was designed for something interactive, which is what a Data Studio report actually is.
In 2010, Edward Segel and Jeffrey Heer’s paper on narrative visualisation analysed 58 examples and described three hybrid structures. The one that matters here is the martini glass. The metaphor isn’t theirs — they borrowed it from journalism training, crediting Steve Buttry — but mapping it onto interactive work is:
- The stem — a narrow, author-driven opening. You control what the reader sees and in what order. No exploration.
- The bowl — after the story is delivered, it opens out into free exploration.
That maps onto a Data Studio report almost exactly, and it resolves the tension that makes most dashboards fail:
- The stem is your summary page. One or two pages, placed first, with no filter controls. Deliberately. A control on this page is an invitation to wander off before you’ve said anything.
- The bowl is everything after it. Filters, date controls, drill-downs, pivot tables — all the exploration your client likes, available the moment they want it.
Segel and Heer frame the underlying tension as a tradeoff requiring judgement: messaging can clarify visual elements but produce clutter, while interactivity can engage a reader but detract from the author’s intended message. On a client report I would put it more bluntly, and this next claim is mine rather than theirs: hand someone a control and you’ve handed them the narrative. Whatever you meant to say now competes with whatever they clicked. Most dashboards fail because they try to do both things on every page and end up doing neither.
The martini glass is the structural answer: do them in sequence rather than simultaneously. And crucially it means you don’t have to abandon your dashboard — which is what most advice on this subject amounts to. You put a stem on it.
Restructuring a Data Studio report you already have, in five steps
This is a one-time build, not a monthly writing decision, and it’s what makes the frameworks stick — because a report with no stem has nowhere to put a narrative.
- Add a summary page and move it to position one. Three or four scorecards with comparisons, one chart, a written summary, the actions. The full build is in how to build an executive summary page in Data Studio.
- Take the filter controls off it. This will feel wrong. Do it anyway — that’s the stem.
- Leave the detail pages exactly as they’re. They’re the bowl and they were never the problem.
- Make the date control report-level so the reader’s selection carries through, rather than each page starting over.
- Write the summary page in SCQA, with the answer first. That’s the whole narrative; the rest is evidence.
What separates a report from a story, on a real page
Structure stays abstract until you see it applied. The worked example — one section of an SEO report rebuilt so that six page-section rows become two, pages published are mapped against sessions gained, two annotations explain the shape, and the action becomes the heading — is set out in full in the before-and-after section of the data storytelling guide.
The part worth carrying back here is what did the work. Not a new chart type: selecting fewer metrics on a stated rule, annotating the events that explain the shape, and naming the action in the heading. All three are structure decisions, which is why a report with no stem has nowhere to put them.
Three report sections that reliably provoke a client reaction
Once the structure exists, these are the report sections that consistently earn a reaction.
Pattern 1: performance against what was recommended but not done
Performance against what was recommended and not yet done. Keep the recommendation backlog somewhere visible and let it accumulate — the conversation it eventually forces (“should we pause this workstream?”) is a more useful one than another month of polite reporting.
Pattern 2: performance against your own internal events
The pattern in the before/after above. A migration, a redesign, a page deletion, a team change — annotate them, because otherwise every dip looks like your fault and every rise looks like luck.
Pattern 3: performance against competitor activity
Not a table of tracked positions. Keywords gained and lost against search volume, so the opportunity is visible. Competitor publishing frequency, which sets your production quota rather than just being interesting. And the move nobody writes about: using competitor analysis to stop tracking a competitor. When their content strategy has visibly shifted away from yours, “we stopped comparing you to them, and here’s the evidence” is a genuinely strong story beat.
If you learn only two of these narrative frameworks
Learn two frameworks and stop. SCQA for anything written, Dykes’ arc for anything spoken. Between them they cover most of agency reporting, and the difference between them is entirely the answer to “am I in the room”. Everything else is a specialisation of one of those two, or a device small enough to use inside either.
Then, once: restructure the report itself as a martini glass. It’s an afternoon’s work and it’s what stops the narrative having nowhere to go.
Frequently asked questions
Which storytelling framework should I use for a monthly client report?
SCQA — Situation, Complication, Question, Answer. The Situation stays stable month to month so you refill one line rather than rewriting, and it forces you to name the question your client is actually asking. Put the answer first, since they will read it without you.
Why does nobody read my dashboard?
Usually because it has no order. Every page is equally available, so nothing is being said and there is no obvious place to start. Adding a summary page with no filter controls and putting it first fixes more of this than any redesign.
How do I report that a client has not implemented our recommendations?
Use STAR as a table — what was agreed, what was done, what resulted. Separate columns for Task and Action make the gap a structural fact rather than an accusation, and the empty cells start the conversation without you having to.
Where to go next on narrative structure and reporting
- Data storytelling for marketers: the full guide
- How to build an executive summary page in Data Studio
- The “so what?” test for a dashboard
- Presenting a Data Studio report to a client
- Annotations and expert commentary in Data Studio
- Data Studio chart types, and how to choose one
- Dashboard design principles for readable reports