lobodrocks
THINK · RESEARCH SPRINT

Where the money is. In five days.

The Research Sprint is five days of research for a brand in Spain or Europe that has to decide what to build next: market and category, five competitors measured in four AI assistants, the audience read from real reviews and conversations, and five short interviews with real customers to confirm what matters. It ends with a fifteen-page document: where the money is, three opportunities, the next step with a price. Fixed fee, quoted within 24 hours.

5days
5competitors measured
5customer interviews

The short version

The Research Sprint is five days between "we should probably do something" and "this is what we build, in this order, for this much". AI reads hundreds of sources in a day: the market, the category, five competitors, thousands of real reviews, and what ChatGPT, Gemini, Perplexity and Google AI answer when someone asks about your category. Real customers confirm what came out of it: five short interviews or a quick survey, the same week. On day five you get a document of twelve to fifteen pages that says where the money is, names three opportunities, and prices the next step.

It is the product most clients start with, because it is the cheapest way to find out whether the thing you are about to pay for is the right thing. Fixed fee, quoted within 24 hours.

What is in, what is not

In the fixed fee Not in it (and what we do instead)
A 45-minute brief and a list of the questions the sprint has to close A quantitative study with a representative sample: that is weeks and a different budget, and we recommend a partner if you need it
Market and category: size, direction, seasonality, what changed in the last year A brand book, a positioning workshop, creative: those come after, in Discovery or Build
Five competitors: positioning, prices, channels, and how often four AI assistants mention them across ten category questions A full AI Visibility Audit with 30 questions and confidence intervals: the sprint takes a light measurement, the audit is its own product
Your audience from real reviews, forums and conversations, plus synthetic segments to narrow the hypotheses Personas invented from a workshop: every segment in the document points at what real people wrote
Five short interviews or a quick survey with your customers on the three main hypotheses Recruiting a panel of strangers: we talk to your customers, you make the introductions
A document of twelve to fifteen pages: where the money is, three opportunities, the next step with a price, the list of sources A deck for a board meeting: the document is written to be read, and we present it once to your team
One presentation of the findings, and the raw material (reviews, competitor sheet, interview notes) in a folder you own Ongoing monitoring: the AI Visibility Program does that monthly, if the sprint says it is worth it

The five days, day by day

Day 0, the brief (45 minutes). What you are thinking of building, what you already know, what you are afraid of. We turn that into a list of questions the sprint has to close: usually six to ten. You name the five competitors, or we name them from what the assistants answer. You send what you have: analytics, sales by direction, the last campaign. The clock starts when the list is agreed.

Days 1 and 2, the AI research. The pipeline reads the market and the category: size, direction, seasonality, what changed in the last twelve months, with every figure tied to a source. It reads five competitors: positioning, price points, channels, what they claim and what their customers say back. It runs ten category questions through four AI assistants in clean sessions and records who gets mentioned, so you know your share of voice before anyone builds a page. It reads your audience where your audience actually talks: reviews, forums, comparison threads, support conversations, and structures what it finds into segments with the words people use. Then it builds synthetic segments from that material to narrow the field to three hypotheses worth a real conversation.

Days 3 and 4, real customers. Five short interviews, twenty minutes each, with customers you introduce, or a quick survey to your list if interviews are not possible that week. Three hypotheses, three questions each, the same for everyone so the answers can be compared. A person runs them. This is the part that turns a good guess into a decision.

Day 5, the document. Twelve to fifteen pages. Where the money is: which direction, which segment, which moment. Three opportunities, each with what it would take and what it would return in units: bookings, orders, signed contracts. The next step, with a price and a date, whether that is Discovery, a Launch Page, a mechanic, an AI Visibility Program, or nothing from us. The list of sources at the end. We present it once to your team and answer the questions.

What synthetic research can and can't do

The honest version. AI reading thousands of reviews finds the patterns a person would find in a month, in a day, and it does not get tired on review 800. Simulated segments are good at narrowing: they tell you which three of your ten hypotheses are worth a real conversation, and which are dead. Validation studies published in 2026 put the accuracy of synthetic respondents somewhere between 47% and 88% depending on the task, strong at the level of segments and aggregates, weak at predicting what one individual will do. Large brands run synthetic personas alongside real panels for exactly that reason, not instead of them.

So the sprint uses synthetic research for what it does well, the narrowing, and then talks to real customers about what is left. We do not skip the second part when the first one looks convincing, and we do not sell the first one as the truth. Where a number in the document comes from a simulation, it says so.

What stops the clock, and what happens then

  1. No customers to interview. If introductions are not possible that week, the interviews become a survey to your list or to your social audience, and the document says which. If neither exists, we say so on day zero and the sprint runs on the public material with a wider margin, stated in the document.
  2. The question changes mid-sprint. It happens once you see the competitor sheet. A change on day two is free; a change on day four moves the delivery by the days it costs, agreed in writing.
  3. A regulated category. Health, finance, alcohol, gambling: the market reading includes what can and cannot be said, and the opportunities are filtered by it. It does not slow the sprint, it is part of the brief.

What it costs

A fixed fee, quoted within 24 hours of your message, ex-VAT with an EU invoice. It covers the five days, the interviews, the document, the sources and one presentation. No hourly meter, no extras inside. The next step is priced in the document; taking it is your decision.

Who it is for

  • A brand about to build something and not sure it is the right something: a new direction, a new market, a campaign with real money behind it.
  • A brand whose marketing works "okay" and cannot say why. The reviews and the interviews usually can.
  • A founder before a round or a launch, who needs a document with sources rather than a feeling.
  • An agency that wants to walk into a pitch with numbers about the client's market, under its own name or with ours. Marketers say they want original thinking from partners first and help with transformation second; a sprint is both, in a week.

CMOs put 15.3% of their budgets into AI this year and only 30% say they are ready to scale it (Gartner, May 2026). The sprint is a way to get the research part of that without buying a platform: one week, one document, one decision.

Ownership and hosting

Everything the sprint produces is yours: the document, the competitor sheet, the review corpus, the interview notes, the AI answers we recorded, in a folder in your name. Nothing you share is used for anything but this sprint, and no personal data goes near a model's training set. If you never call us again, the document still says what to do first.

Where to go next

  • The sprint said "build the structure first"? That is Discovery, four weeks, a plan with a closed price on every block.
  • The sprint said "your team can run this in-house"? That is Marketing OS, two to four weeks.
  • The sprint said "you need a page by Thursday"? That is the Launch Page, 72 hours.
  • The competitor sheet showed the assistants never mention you? The AI Visibility Audit measures it properly, 5 days.
  • How the AI pipeline and the human decisions fit together: the method.

Tell me what you are thinking of building and what you are not sure about. You will have a fixed fee within 24 hours, and a document five days after the brief.

Questions

Is five days enough?

For the decision the sprint is built for, yes. Five days is not enough for a quantitative study with a representative sample, and we do not pretend it is. It is enough to read the market and five competitors properly, to read what your customers say when nobody from your brand is listening, and to test the three hypotheses that came out of that with real people. The sprint answers one question: what should we build next, and why. A brand that needs a full segmentation with confidence intervals needs a different project, and we say so on the first call.

Do we need to give you access to our data?

It helps, it is not required. Day one and two run on what is public: your category, your competitors, the reviews and conversations your customers already leave online, and what four AI assistants answer about you. If you can share analytics, CRM exports or sales by direction, the audience part gets sharper and the interviews get better questions. Everything you share stays in a folder you own, is used only for this sprint, and never goes near a model's training set.

What if the sprint says "don't build"?

Then the sprint paid for itself several times over, because the thing you were about to build would have cost more than five days of research. It happens. The document still names three opportunities, and one of them is usually smaller and closer than the original idea. Nobody at lobod.rocks is paid to sell you the next step; the next step is priced in the document, and you decide with a clear head.

What does synthetic research mean here, and how much do you trust it?

Synthetic means AI reads and structures what real people have already written: thousands of reviews, forum threads, comparisons, and the answers assistants give about your category. It also means simulated segments we use to narrow hypotheses before we talk to anyone. Validation studies published in 2026 put the accuracy of synthetic respondents between roughly 47% and 88% depending on the task: strong on segment-level patterns and elimination, weak on predicting what an individual will do. So we use it for what it is good at, narrowing, and we confirm with real customers, five interviews or a quick survey in the same week. We do not sell the synthetic part as the truth.

What does it cost?

A fixed fee, quoted within 24 hours of your message, ex-VAT with an EU invoice. It covers the five days, the interviews, the document, the list of sources and one presentation of the findings to your team. There is no hourly meter and no upsell inside the document: the next step is priced, and it is your call.

Who does the work?

The research runs on an AI pipeline built for this, the interviews are run by a person, and the reading of it all is done by Dmytro Lobod, ten years of agency work and 350 brands behind that reading. The document is written by a person, in your language, and every figure in it carries its source.

Sources

  1. lobod.rocks, llms.txt (product scope, process, ownership terms) 2026-09
  2. Gartner, 2026 CMO Spend Survey (15.3% of marketing budgets go to AI, only 30% of CMOs ready to scale AI capabilities) 2026-05
  3. CMO Barometer 2026, Serviceplan and University of St. Gallen (69% of CMOs want original thinking from agencies, 44% help with transformation) 2025-11
  4. Personia, What 2026 validation studies agree on about synthetic research (accuracy 47% to 88% by task, strong on aggregates, weak on individual prediction) 2026
  5. Rival Technologies, synthetic personas at Tropicana, Newell Brands and WBD 2026
  6. Funcas, III Encuesta sobre IA (regular ChatGPT use in Spain 4% in 2023 to 28% in 2025) 2025-01

Checked at the date shown; figures move.

Talk to Lobod

Start with the numbers. Then we build.

Tell me what you're launching, or what isn't working. Within 24 hours you get either the package that fits, with a price, or a five-day Research Sprint proposal.

Dima@lobods.comLinkedIn

Barcelona · Spain & EU · EN / ES · replies same day
Letters

Twice a month, something worth reading about AI visibility, launches and what actually moves a brand.

Double opt-in: one letter asks you to confirm, and nothing arrives until you do. One click at the bottom of any letter leaves for good.