Laboratory Data
The power behind the screen: segmentation or manipulation?
By Violeta Ayala — Tecnóloga
Rodrigo Paz's campaign won spending almost nothing. Behind it is a lesson that works for you: how a campaign divides and reaches people, which free tools let you see it, and how to notice when you are the one being segmented.
Open data, method and sources on GitHub: github.com/violetajaguar/mapa-medios-bolivia
Segmenting means dividing people into groups that share something, to speak to each group through the right channel and with the right message. Every campaign, every brand and every small business does it. You do not speak the same way to a 19-year-old in El Alto as to a 50-year-old shopkeeper in Santa Cruz. It splits them, and reaches each one differently. This is Bolivia's clearest and most recent case, told so you understand it and, above all, so you can audit it yourself.
It works across four dimensions: who they are, where they are, what they do online, and where they get their news. The Paz and Lara campaign used all four. Tap them to see how.
The lesson: segmenting is not about spending more
Here is the heart of the case. In the first round, more than Bs 2.3 million was spent on political ads on Facebook, according to Bolivia Verifica. Look at who spent and who won.
Declared spending on Facebook ads · 2025 first round
The one who spent the most lost. The one who spent almost nothing won. The advantage was not the budget. It was understanding where people were and what format they consumed.
Source: Bolivia Verifica, Meta Ad Library (first round, August 2025). Bars to scale.
The polls did not see it coming
The favorites were not only the ones who spent the most. They were the ones the polls called winners. And they missed badly. Before the first round, almost every poll put Samuel Doria Medina first and Quiroga second, with Paz third or lower, in single digits.
2025 first round · what the polls said vs what happened
Pre-election poll · Ipsos-CIESMORI for Unitel
Actual result · TSE count
The poll favorite was knocked out. The one running third, in single digits, won the election.
Source: Ipsos-CIESMORI for Unitel (pre-election poll) and TSE count (result). El Post, Infobae, La Nación.
The gap was brutal. One pollster projected Paz between 3% and 9%, and he got 32%, an error of more than 23 points. None of the eleven published polls anticipated the result. Doria Medina, the poll favorite, finished third and out of the runoff, and ended up asking his voters to back Paz. Even the runoff polls were wrong: they gave Quiroga the lead with 44% to 47%, and Paz won with almost 55%.
The explanation connects to everything above. Paz captured the vote the polls did not see: undecided, blank and null votes added up to nearly 33%, more than any favorite, plus the orphaned progressive vote the MAS left without a candidate. The lesson for reading any poll: it is a snapshot of a moment and does not guarantee the future, and in a fragmented country it can miss by a lot. Add that fake polls were one of the campaign's disinformation weapons. Before believing a number, check who ran the poll and with what methodology.
The tools, step by step
You can verify all of this yourself, free. Here we do not just name the tool. We show you what to tap, what you will see, and how to read it. No expertise needed.
Meta Ad Library
FreeIt is Facebook's official search where you see who paid for political ads and how much they spent. That is how we learned Quiroga spent more than a million and Paz almost nothing.
- Open the page. You do not need a Facebook account to look.
- Under "Ad category", choose
Ads about social issues, elections or politics. - Under "Country", choose
Bolivia. - In the search box type a name: a party, a candidate or a page. Try
Rodrigo Paz. - Tap each ad and open "See spending details". The figures are there.
Active ads: 34
This is roughly how it looks. Each ad shows spend, reach and who paid.
How to read it: if a page that looks "neutral" actually pays for ads in someone's favor, it shows here. Spending comes in ranges, not to the cent. Compare pages to see who really invests.
Google Trends
FreeIt shows how many people search a name over time and in which region. During the campaign, "Edman Lara" led interest, above everyone.
- Open the page and tap "Explore".
- Type a name. At the top, change the country to
Boliviaand the time to12 months. - Tap "Compare" and add another name. Try
Edman LaraagainstTuto Quiroga. - Look at the curve: who rises, when, and in what spikes.
- Scroll to "Interest by subregion" to see which departments search each one most.
The numbers run 0 to 100: relative interest, not an exact count of people.
How to read it: it tells you who sparks real curiosity and where. A candidate with little ad spend but heavy search has organic strength. That is what happened with Lara.
TikTok: search and trends
FreeTikTok's own search lets you trace how an audio or a hashtag goes massive, and who pushes it. That is how "Que Lara más que vuele" traveled.
- Tap the magnifying glass and type a name or hashtag. Try
#EdmanLara. - Switch between the top tabs: Top, Videos, Users, Sounds.
- Open a sound or audio: TikTok tells you how many videos use it.
- Notice if many different videos repeat the same audio or the same phrase.
- Look at the accounts: are they real people with history, or new empty accounts?
If the same sound appears in thousands of videos, it is a live trend.
How to read it: many real accounts using the same audio is genuine virality. Many new empty accounts saying the same thing is a paid operation. Learn to tell them apart.
Bolivia Verifica
FreeA team that checks what is false and follows the ad money. Almost all the data in this case comes from there. Use it before believing or sharing anything.
- Open the page. There is a search box at the top.
- Type the topic, phrase or name of what reached you. Try a WhatsApp chain you doubt.
- Open the article: it tells you if it is False, Misleading or True, and why.
- Check the date and the explanation: where the claim came from and how they checked it.
- If it is not there, do not share it yet. Absence is also a signal.
No methodology, no pollster.
Each check carries the verdict, the date and the evidence.
How to read it: a poll with no methodology is worthless. A video with no source, the same. If something was already debunked, here you find out before falling for it.
Electoral Body data (OEP)
FreeThe electoral body publishes results by department and municipality. It is the base for understanding the geography of the vote: who won where.
- Find the official results of the election by department.
- Note each candidate's percentage in each region.
- Open
data/ciudades.csvin this repository: it has each city's population (2024 Census). - Cross the two: where each one won and how many people live there.
- That reveals the real coalition: Paz strong in the highlands, another candidate strong in the east.
Crossing vote with population reveals where each one's real strength is.
How to read it: the vote map is the map of geographic segmentation. It shows which region responded to which message, and where a campaign chose to fight.
DataReportal
FreeA free annual report that says how many people use each network in Bolivia. It sizes each platform before you decide where to be.
- Go to the page and search
Digital 2025 Bolivia. - Open the report. It is long but you skim it by charts.
- Find the users by platform slide: Facebook, TikTok, Instagram, YouTube.
- Note each one's size and the daily time on social.
- With that you choose where your audience is before spending a peso.
Each network's size tells you where it is worth being.
How to read it: Facebook gives volume and adults. TikTok gives youth and organic reach, the one that decided 2025. Choosing the platform is the first step of segmenting well.
How to notice when you are being segmented
The same machinery that reached voters in 2025 reaches you every day. These are the signs.
"Why am I seeing this ad?" On Facebook and Instagram every ad has that option. It tells you what criteria they used to pick you: your age, your area, your interests. There you see your own segment.
The paid label. By law, paid propaganda cannot be anonymous. If political content does not say who paid for it, be suspicious.
The same message across many new accounts. Dozens of accounts with no photo or history saying the same thing on the same day are an operation, not people with opinions. In 2025 a firm charged about Bs 535 for 500 comments.
AI-made content. In 2025 there were 534 disinformation cases, most on TikTok (44%). Deepfakes of Arce, of Milei, with stolen CNN or Univisión logos. If a video looks off, verify it.
Emotion as bait. Content built to go viral wants to enrage you in seconds. When something makes you suddenly furious, breathe before sharing. That reaction is exactly what the algorithm rewards.
The ethical line
Legitimate segmentation: using public, aggregated data (how many people use TikTok, how old a city is, which region searches what) to speak better to a group. That is what this class teaches.
Manipulative microtargeting: using personal data without consent, from leaks, from the voter roll or bought from brokers, to aim individual messages, often with deception.
Bolivia has no personal data protection law. Without it, no one protects you if a party or a company uses your data without permission. The sala.red rule: segment with public, aggregated data, never with personal data without consent.
Paz's campaign did not win by spending more. It won by understanding who its people were, where they were, what app they used and where they got their news. That is segmenting.
And the tools to see it are free and within anyone's reach. The same machinery that seeks your vote seeks your attention. Knowing where information comes from, and why it reaches you, is the first right.
Epilogue: what they are doing now
Segmentation does not end on election day. The same machine keeps running from government, and the audience you won can be turned. That is exactly what happened in Bolivia, and through the same channel that united them.
Paz and Lara took office on November 8, 2025. Within two weeks, the alliance began to break in public. Lara claimed on TikTok that he was being sidelined from power. In December he declared himself in opposition to his own government, accused Paz of governing for the rich and surrounding himself with corrupt people, and promised to expose it from inside. Paz answered by distancing himself from the format: he said he does not do TikToks, he acts.
- 8 Nov 2025 Paz and Lara take office. The ticket that won on social media reaches power.
- Nov 2025 Within two weeks, Lara claims on TikTok that they want to sideline him and are not returning his calls.
- Dec 2025 Lara declares himself constructive opposition and accuses Paz of governing for the rich. Paz replies: "I don't do TikToks, I act".
- Jan 2026 From the Assembly, Lara announces a bill to block Paz's decrees. The Constitutional Court receives challenges.
- May – Jun 2026 A month of blockades chokes La Paz. Analysts do not rule out that Paz may have to resign before year's end.
Lara was left with no real power in the cabinet. His only trusted minister, at Justice, was dismissed within days, and Paz closed that ministry. The wear reached both of them. Paz began with 65% approval in November 2025 and fell steadily: 52% in April, 46% in May, and 38% in June 2026, when disapproval passed approval for the first time. Lara fared even worse, at 81% disapproval, according to Ipsos-CIESMORI in the four main cities. His spontaneity, the very thing that made him viral as a creator, now works against him as a governor.
How this was made
This case rests on verified reporting: Infobae, CNN, La Nación and AFP on the campaign; La Razón and the analysis of political scientist Natalia Aparicio; Bolivia Verifica and its Electoral Disinformation Observatory (the 534 incidents, the deepfakes, the ad spending); and Infobae on the role of Catalina Paz and strategist Fernando Cerimedo, whose influence in government is debated. The epilogue on the government draws on reporting from December 2025 to July 2026: Infobae, CNN, AFP, El Diario, Opinión, Urgente.bo and Panamericana on the Paz-Lara rupture, an Ipsos-CIESMORI poll, and analysis from CSIS and Nueva Sociedad on the 2026 crisis. As this is developing information, check it against the consultation date. The first-round polling failure and the 2026 approval figures draw on Ipsos-CIESMORI, CB Consultora, El Post, Infobae, La Nación and CNN. Context: INE 2024 Census and DataReportal 2025. All audience and spending figures are estimates with source and date. License CC BY-SA 4.0.