Meta, Google, and TikTok metrics reward last-click and blend creative effect with budget effect — the creative that looks good in the report may only look good because it received more investment. Creative Dividend Index reads all creatives in a campaign together and returns a comparable ranking, with a forecast of what changes if you keep, cut, or scale each piece. Ready to defend the next production and budget round — with numbers, not gut feel.
You send media platform¹ exports (Meta, Google, TikTok, YouTube), the creative assets folder, and the campaign brief. We return a comparable ranking across all pieces, a forecast of what changes if you keep, cut, or scale each one, and a read on each creative's real role in the journey — separating piece effect from invested budget effect.
¹ assuming clean classification bases; otherwise we will assess whether Data Clean is needed
From agency to marketing leadership: when the question is which piece deserves to stay, which batch of creatives delivered real return, and how to prove to the CFO that creative investment is justified. Platform metrics answer click and view; this study answers relevance.
The next client meeting will ask: “which creative delivered the most return?” The answer today comes in CTR and CPM — metrics that do not say whether the piece worked or the budget was simply there. Here you leave with an independent benchmark to defend creative work, with a number that supports repurchase of the next flight.
When the CFO asks “what was the ROI of this batch of creatives?”, the answer today comes in CPM, CPA, or frequency — operational metrics that do not answer the question. The study delivers ranking by piece and incremental forecast in committee language, with the justification ready to defend production investment.
Without consolidated creative-level read, every campaign starts from zero — and what worked last quarter gets lost between spreadsheets and agency reports. The study closes the loop: it becomes direct input to decide production, audience, and mix in the next flight, with documented learning.
Each metric, one signal. Combined, they show which creative is most effective.
Five pieces in the study. Each one exists to answer a concrete question that will come up in the meeting with the agency, leadership, or client — not to fill slides.
Comparative ranking by creative, normalized by audience and objective. The piece that dominated because it had more budget may not be the one that contributed most — the ranking separates exposure effect from piece effect.
Incremental forecast with frequentist and Bayesian read, with confidence interval. Lets you simulate the next round before deciding.
Markov diagnosis (Brand Journey) with removal effect: each piece's contribution is calculated even when it was not the last click. Solves the classic platform metrics problem — which reward the last touch and ignore top-of-funnel work.
Temporal backtest and documented acceptance criteria. The model only enters the report if it passes the checks — and the report states where the read is robust and where it deserves a caveat.
Reads the asset, not just the number. Multimodal ingestion — we open the performance spreadsheet, read the brief PDF/PPT, and scan the creative assets folder to extract piece attributes. The ranking is not just “campaign X had more conversion”; it is “this type of creative, with this approach, performed better in this context”.
The research The Creative Dividend (Effie Worldwide × System1, 2026), with 1,265 campaigns, showed that creative and media together explain up to 98% of results — and that treating the two separately is the main reason campaigns underperform.
Even so, most of the market still evaluates creative by CTR and media by CPM. Creative Dividend Index closes that gap at the piece level: comparable, auditable read, in business decision language. explōrātiō proprietary methodology — Effie is a conceptual reference; what we validate is your data.
Another layer of read: relative importance of each creative in campaign results. It does not replace point tests, does not replace Media Mix Modeling, does not replace platform reporting — it completes what's missing.
Confirms a hypothesis between 2–4 variations. The Index reads the entire campaign and measures relative weight — it uses A/B as input, not as a substitute.
Reward last touch. Markov distributes credit by each creative's real role in the journey.
Qualitative narrative. The Index is the quantitative benchmark across brands, pieces, and channels.
MMM: how much each channel contributed. Index: which creative drove within the channel. They are complementary.
Four phases — features, validation, deliverables, workshop — in fixed order. What is in the signed scope is what gets done, and it is the benchmark for measuring “done” at the end.
Features → validation → deliverables → workshop
Revision per tier: 15 to 45 days after formal delivery, in one consolidated round within the signed scope. Requests outside the window become new scope.
Building the Creative Quality Index (CQI) — features normalized by creative, with multimodal blend of spreadsheet, brief text, and asset attributes.
Feature table by creative, comparable across batches.
Temporal backtest and documented acceptance criteria. Only outputs that pass the checks enter the report.
Criteria met table + explicit caveats.
Executive PDF report, PPTX deck for internal presentation, and auditable CSV tables (ranking, forecast, Markov journey, acceptance criteria).
Presentable material from the day after delivery.
Joint read session — interpretation of results, scenario simulations, and bridge to the next production and media decision.
Aligned team, with documented action plan.
Three study sizes, one-time delivery. Scope is sized by number of campaigns, creative volume, and brands involved — we align in the initial 30- to 45-minute conversation with a sample of your case.
R$ 3.947
R$ 8.547
R$ 16.947
For agencies that want to deliver the study under their own brand, and for marketing teams with structured BI that want the technical slice to integrate, specific formats by conversation. Pilot packages with discount for the first study — just ask.
Creative analysis is about campaign scope, and sometimes the scope changes — a new piece enters, an additional brand appears, an extra audience cut is requested. Instead of hiding the rule, we leave the table in plain sight: what triggers an add-on, what becomes a new study, and what each case costs on top of the contracted package.
+35%
Change up to 25% of signed scopeEx.: Essential (R$ 3.947) → add-on of R$ 1.381.
+50%
Change above 25% of scopeEx.: Essential (R$ 3.947) → add-on of R$ 1.974.
25%
Limit by added creativesEx.: Essential — 30 creatives in scope. +5 creatives (17%) → targeted. From the 8th extra creative (≥27%) → broad.
25%
Limit by audience/objective cutsEx.: Essential — 3 cuts in scope. +1 cut (33%) → broad. Cuts trigger full re-segmentation of the ranking, hence the more sensitive trigger.
15–45d
Per tier: 15d · 30d · 45dPer tier: 15d (Essential) · 30d (Complete) · 45d (Enterprise). One consolidated round, acceptance by email.
New
Outside add-on — separate quoteAnother brand outside scope, unplanned platform (TikTok added to a Meta+Google scope), new time window (campaign from another period), creative typology outside the closed dictionary in the diagnosis.
Criteria A and B measure % of scope changed · higher index prevails · targeted add-on +35% or broad +50% on the package · all within signed scope
The study only delivers a statistically reliable ranking when the campaign has volume and variety of creatives. Use the cards below to confirm it makes sense before the conversation.
From 20 creatives per campaign. The model needs sufficient variation between pieces to generate a statistically reliable ranking; below that volume, it becomes opinion with a nice chart.
The campaign period while it was active. Extended history is not required — unlike MMM, this study reads the campaign itself, not the brand's time series.
Today, exclusively digital creatives (Meta, Google, TikTok, YouTube, programmatic). No inference about offline vehicles.
CSV/XLSX exports from the media platform (impressions, clicks, conversions, spend), creative assets folder (video, image, copy), and brief in PDF/PPT when available. Column-by-column documented mapping.
Secure links and encrypted transfer. Your data is yours: everything you send is used only to run your model, and deleted afterward. Nothing is stored, nothing is trained, nothing is reused in another project. LGPD compliant, documentation at close-out.
The study reads a closed campaign with creative volume — it does not replace all marketing analysis. In some situations another explōrātiō product delivers a faster answer.
For 3 or 4 variations, A/B test answers better and faster.
Today the study reads digital only — for on/off mix, MMM Combined is the path.
The question "how much to put in each channel" is MMM, not Creative Dividend.
The study reads a closed campaign (or closed cycle within an always-on campaign). For real-time management, it is operational media reporting.
Study delivered within signed scope. Ranking, forecast, and journey in committee language, with presentable material for the next client or board meeting, and an auditable trail for anyone who wants to redo the read. No “almost ready”, no endless revision round.
It does not replace operational reporting — it complements it. Native metrics reward last touch and blend piece effect with budget effect. The study reads all creatives in the campaign together, with comparable ranking, incremental forecast, and Markov journey, to decide production and budget in the next round.
MMM answers how much each channel contributed to the KPI over time. The Index answers which creative drove within the digital campaign. They are complementary: one guides media mix; the other, which piece to keep, cut, or scale.
From 20 creatives per campaign, with real variation between pieces. With 3 or 4 variations, native A/B test is usually faster and more appropriate. Tier scope also limits campaigns, brands, and audience cuts — we align in the demo.
Generally, 3 to 4 weeks on the Essential tier and up to 6 weeks on larger scopes, after base and assets are locked. Unlike MMM, it does not require years of brand history — it reads the campaign while it was active.
We work with platform exports and journey data as available in scope (impressions, clicks, conversions, exposure sequence when available). The more granular the event trail, the better the MTA — but a data lake is not required; spreadsheets and treated exports enter the brief.
Secure transfer; use only to run your study. Nothing is trained for other clients or reused in another project. At close-out we document retention and LGPD alignment — coordinating with legal or DPO when needed.
Technical PDF, executive PPTX, CSV tables (ranking, forecast, Markov, acceptance criteria), read session and — per tier — workshop. Only outputs that pass documented acceptance criteria enter the report; caveats are explicit.
The technical terms that appear in the report, PPTX, and CSV. It is here because whoever operates data (BI, analytics, planning) will want to open and audit — whoever only receives the recommendation does not need to read it.
explōrātiō product that ranks creatives in a digital campaign, estimates increment, and distributes credit in the journey — for production, media, and budget decisions beyond platform CTR.
Concept from the research The Creative Dividend (Effie × System1): creative and media together explain most of the result. The Index is proprietary methodology applied to your data — Effie is a conceptual reference, not the model itself.
Feature index by creative — blend of performance spreadsheet, brief text, and asset attributes — normalized to compare pieces that ran on different audiences and objectives.
Relative importance order of creatives in campaign results, separating piece effect from exposure effect (budget, audience, objective).
Estimate of what changes in the KPI if you keep, cut, or scale a creative in the next round — with confidence interval, not just a point number.
Models calibrated to campaign history with point estimates and intervals derived from the fit — one of the two evidence lines in the study.
Models with priors defined in scope; produce posterior distribution and credibility interval. When they converge with the frequentist, the recommendation gains weight; when they diverge, the report explains why.
State chains representing the consumer journey between creative exposures — basis for distributing credit across the funnel, not just at the last click.
Attribution that considers multiple touches in the journey. In CDI, Markov with removal effect is the layer that corrects platform last-click bias.
Simulation of the "without this creative" scenario in the journey to isolate how much of the result disappears when the piece is removed — even if it was not the last click.
Platform rule that credits conversion to the ad seen last. The Index uses Markov and joint ranking so bottom-funnel pieces are not the only ones rewarded.
Validation in which the model is trained on one segment of the campaign and tested on another — answers whether the read would have been useful before the real decision.
Thresholds documented in scope (fit, error, stability). Outputs that do not pass the checks do not enter the main report — or enter with an explicit caveat.
Initial assumption about parameters before seeing campaign data — calibrated in the brief and influential when volume per creative is low.
Convergence: frequentist and Bayesian point in the same direction for a creative. Divergence: they disagree — the report documents causes (priors, volume, model form).
The initial conversation is to look at a sample of your creatives and campaign spreadsheet, and say frankly: there is a study here that solves it, or native A/B test can still answer. If yes, we design scope, number of creatives, and timeline on the spot. If not, you still leave with the diagnosis.