Start free trial
Trending New Genres Authors
EnglishEnglish
EspañolSpanish
简体中文Chinese
繁體中文Chinese (Traditional)
FrançaisFrench
DeutschGerman
日本語Japanese
PortuguêsPortuguese
ItalianoItalian
한국어Korean
РусскийRussian
NederlandsDutch
العربيةArabic
PolskiPolish
हिन्दीHindi
Tiếng ViệtVietnamese
SvenskaSwedish
ΕλληνικάGreek
TürkçeTurkish
ไทยThai
ČeštinaCzech
RomânăRomanian
MagyarHungarian
УкраїнськаUkrainian
IndonesiaIndonesian
DanskDanish
SuomiFinnish
БългарскиBulgarian
עבריתHebrew
NorskNorwegian
HrvatskiCroatian
CatalàCatalan
SlovenčinaSlovak
LietuviųLithuanian
SlovenščinaSlovenian
СрпскиSerbian
EestiEstonian
LatviešuLatvian
فارسیPersian
മലയാളംMalayalam
தமிழ்Tamil
اردوUrdu
Searching...
SoBrief
The Victory Lab

The Victory Lab

The Secret Science of Winning Campaigns
Campaigns can test what moves voters, but election night cannot tell them which tactic mattered.
by Sasha Issenberg 2012 321 pages
3.78
1k+ ratings
Listen
Immersive
V2.3
Amazon Kindle Audible
Try Full Access for 3 Days
Unlock listening & more!
Continue

Key Takeaways

Treat close elections as millions of small decisions, not one grand story.

Colorado’s narrow margin. In 2010, Democrat Michael Bennet’s Colorado Senate campaign sent a million plain letters to likely supporters. Each thanked the recipient for voting before and expressed hope they would vote again. The message avoided candidates and issues, leaning instead on the social discomfort of failing to follow through. Hal Malchow estimated that a 2.5 percentage-point turnout increase could mean 25,000 additional voters. Bennet ultimately won by about 15,000 votes, but that result cannot prove the letters decided the race.

Campaigns resist tidy explanations. A candidate’s appeal, the national mood, demographics, and turnout tactics can all shape an outcome. After George W. Bush’s 2000 win turned on 537 votes in Florida, strategists took small turnout gains more seriously. The book’s central argument is that elections are not explained by one brilliant speech or ad. They are assembled from individual decisions that campaigns can study, influence, and sometimes measure.

Analysis

What's striking is the gap between a compelling campaign story and a defensible causal claim. A close result makes marginal tactics consequential, but it also makes attribution harder: thousands of small events and broader conditions converge in the same count. This resembles risk analysis, where a small change in one component can alter an outcome without being its sole cause. The useful discipline is to distinguish a tactic's measured average effect from a claim that it decided a particular election. That distinction protects both strategic learning and public understanding from the comforting myth of a single decisive move.

Demand a controlled test before declaring a tactic responsible for victory.

Gosnell tests the guesswork. In 1924, University of Chicago researchers Harold Gosnell and Charles Merriam studied why eligible Chicagoans did not vote. Gosnell then sent registration postcards to comparable groups, some receiving reminders and others receiving none. Recipients were nine percentage points more likely to register. In the next aldermanic election, 57 percent of people who received a cartoon urging them to vote turned out, compared with 48 percent of those who did not. The finding showed that campaign contact could change behavior, and that its effect could be measured rather than inferred from election-night anecdotes.

A field experiment tests an intervention in the real world by randomly assigning people to receive it or serve as a control group. Comparing outcomes helps isolate the intervention’s effect from other influences. Gosnell’s approach was largely forgotten, but its logic returned decades later as political researchers began testing canvassing, phone calls, and mail. The method does not answer every question about an election, but it replaces confident folklore with evidence.

Analysis

Random assignment is powerful because it addresses selection bias: people who receive campaign contact often differ from those who do not, so a simple comparison can mislead. The same logic underpins clinical trials and modern program evaluation. Yet the method is not a magic stamp of truth. Results depend on who was studied, how the intervention was delivered, and whether the outcome was measured well. Gosnell’s lasting contribution is therefore not a universal recipe for turnout. It is a habit of asking what would have happened to similar people without the intervention.

Use social expectations, not civic lectures, to move habitual nonvoters.

A neighbor can be a nudge. In a 2006 Michigan experiment, consultant Mark Grebner sent some households lists showing which neighbors had voted and said an updated list would follow the election. Turnout reached 37.8 percent among recipients, against 29.7 percent in the control group. A standard civic-duty message raised turnout only to 31.5 percent. The intervention drew on social pressure: people may act to avoid being seen as out of step with those around them. It was also far more intrusive, and produced angry complaints.

Softening the pressure. Hal Malchow and researchers later tested gentler messages. A New Jersey letter thanking recipients for voting previously increased turnout by 2.5 percentage points. The core lesson was not that every voter should be shamed. It was that behavior can respond to social expectations more than to familiar reminders about civic virtue. The same psychological lever can motivate participation or provoke resentment, depending on how it is applied.

Analysis

The contrast between a generic reminder and a social cue reflects a basic feature of human behavior: norms often feel more immediate than abstract duties. The approach also exposes a moral tension. A tactic can be effective while exploiting embarrassment or uncertainty about privacy, and effectiveness alone does not settle whether it is acceptable. The gentler thank-you message suggests that social motivation need not require humiliation. Its success is a reminder that ethical design is not merely a constraint on persuasion. It can preserve trust, which aggressive pressure risks damaging.

Target individuals by likely response, not by familiar political categories.

From precincts to people. For decades, campaigns often treated neighborhoods as their basic unit: if an area usually favored a party, organizers assumed its residents did too. Microtargeting uses individual records, surveys, demographic information, and commercial data to predict which people may support a candidate, turn out, or respond to a particular message. In Oregon’s 1996 Senate race, Malchow surveyed 9,051 voters and found persuadable Democrats and independents in places conventional targeting might overlook. One small survey group of 67 people suggested a larger pool of roughly 8,000 similar voters across the state.

Prediction changes the map. Republican analyst Alex Gage later combined voter files with consumer data to build political segments, including people who lived in heavily Democratic areas but might support a Republican. In 2004, this helped Bush’s campaign contact voters beyond its customary strongholds. The method can find hidden opportunities and reduce wasted mail or calls. But a score is a probability, not a personal conversation, and a model can miss people who do not resemble its training data.

Analysis

The shift from geographic averages to individual predictions resembles precision medicine: treatment is directed toward a person’s estimated response rather than a broad population label. That can improve efficiency, but it also raises a question about what the model has actually learned. A predictive variable may be useful without revealing why it works, and a profile can conceal meaningful differences among people grouped together. A sound campaign treats scores as provisional guides for allocating attention, then checks them against real responses. The danger is not simply error. It is mistaking a useful forecast for a complete account of a voter.

Match each voter contact to the human attention it can earn.

Delivery changes the result. In a 1998 New Haven field experiment, volunteer canvassing raised turnout by 8.7 percentage points over the control group. Each mailed postcard produced a smaller increase of 0.6 points, while scripted calls from a distant call center had no measurable effect. The results challenged campaigns that spent heavily on calls simply because they were easy to scale. A conversation at the doorstep could feel personal; a rushed phone script could sound like noise.

Low-tech can be precise. The Obama campaign used voter data to identify supporters who rode particular bus routes, then bought ads inside buses in selected cities. It also tested different website buttons, finding that one version attracted nearly 20 percent more sign-ups than the standard wording. The larger lesson is not that bus ads or a particular button always work. It is that channel, context, and execution shape whether a message gets noticed. Sophisticated data can make neglected, inexpensive channels newly useful.

Analysis

The evidence puts a useful limit on the idea that communication scales cleanly. A phone call is not equivalent to a conversation, and exposure is not the same as attention. This distinction appears in education and public health, where information often fails unless it arrives through a trusted or usable channel. At the same time, the New Haven results do not establish that every door knock outperforms every call. Local context, caller quality, and the audience matter. The strongest practice is to test the actual delivery method rather than rely on the medium’s reputation.

Build permanent data capacity so each campaign inherits what its predecessor learned.

Elections reward short-term thinking. A campaign exists to win one contest, consultants have incentives to defend their own methods, and staff often discard records once the votes are counted. That structure makes learning difficult. After the 2000 election, Republican strategist Matthew Dowd and organizer Blaise Hazelwood created the 72-Hour Task Force to test turnout tactics and make results useful across campaigns. Its experiments found, among other things, that volunteer calls could outperform paid calls by five points in South Carolina. A test in Virginia’s Frederick County suggested that organized outreach to local conservative groups helped a Republican candidate run stronger there than in a comparable county.

Infrastructure compounds learning. After Democrats lost in 2004, operatives and advocacy groups built shared research and data operations, including the Analyst Institute and Catalist. These organizations helped pool voter information and test tactics beyond the life of one candidate campaign. The point is not centralization for its own sake. Durable institutions can preserve results, coordinate experiments, and reduce the odds that every new campaign repeats the same expensive mistakes.

Analysis

This is an institutional problem disguised as a technical one. Evidence accumulates only when organizations preserve data, reward candor about failure, and give people reason to cooperate across election cycles. That is familiar in aviation safety and medicine, where incident reporting and shared standards can matter as much as individual expertise. Political campaigns face a harder incentive structure because they are temporary and adversarial. A permanent research body can help, but it also concentrates influence over data and tactics. Its legitimacy depends on transparency about methods, clear safeguards, and meaningful access for the people doing the fieldwork.

Make analytics serve relationships, not replace them.

Obama joined data to organizing. In Iowa, Barack Obama’s campaign set a goal of bringing 180,000 people to the 2008 caucuses, far above the previous turnout. Its model used supporter cards, voter histories, and survey results to help organizers decide where to recruit, persuade, and mobilize. The final turnout reached 239,000, with many first-time caucus-goers. During the general election, the campaign built teams that translated scores into local plans, while volunteers gathered new information through calls and door-knocking.

The data needed people. The campaign tracked contacts in a system called the Matrix and used updated models to adjust its view of likely support. But its early operation also struggled to connect online sign-ups, donations, and field contacts across separate databases. Bringing analysts closer to the campaign’s daily work made it easier to turn projections into action. Data could show where a volunteer’s effort might matter; it could not create trust or do the conversation for them.

Analysis

The strongest version of analytics is organizational, not magical. A model has value when it changes a real decision: which door to knock on, whom to call again, or where to place volunteers. That requires feedback from the field, since voters may behave differently from their profiles. The campaign’s early data silos also show that collecting more information does not automatically produce more knowledge. Integration, interpretation, and human judgment are necessary. When measurement becomes the goal, teams can optimize what is easy to count while overlooking the harder work of building durable relationships.

Use voter models to allocate attention, not to declare people knowable.

A score is not a person. Barack Obama’s 2008 campaign faced uncertainty about whether voters would openly express support for a Black candidate. Analyst Ken Strasma developed an “openness” score using survey responses and other patterns to estimate which voters might be receptive. The purpose was to improve predictions and avoid treating every undecided voter as equally persuadable. The example shows how campaign models can incorporate sensitive attributes even when they are not directly recorded as simple facts.

False confidence has costs. A profile can misclassify people, and an inference about a voter’s likely attitude is not the same as knowing what that person believes. Social-pressure mail can likewise raise turnout while unsettling recipients who feel watched. These techniques put questions of privacy, fairness, and manipulation inside the campaign’s practical toolkit. Good measurement can reveal an intervention’s effect, but it cannot by itself determine whether the intervention respects voters or whether its classifications are just.

Analysis

The passage from prediction to treatment deserves particular scrutiny. A campaign that withholds contact from people assigned low scores may convert a statistical guess into a self-fulfilling exclusion. This resembles algorithmic decision systems in lending or employment, where proxies can reproduce inequities even when the model never explicitly asks about protected identity. The book’s account is valuable because it makes these mechanisms visible rather than pretending targeting is neutral. A responsible system needs audits for error and disparate impact, limits on sensitive inference, and room for people to defy their predicted category.

Expand who votes by making registration measurable, early, and easy.

The electorate is not fixed. Campaigns often focus on persuading people already likely to vote, but registration determines who can participate at all. Women’s Voices Women Vote used voter and consumer data to identify likely unregistered supporters, then tested mailed registration forms and improved the response process. In 2008, the group sent nearly 20 million applications and registered close to one million people, at an estimated cost of about seven dollars each. Experiments helped cut the cost of a registration from roughly eleven dollars to seven.

Small frictions matter. In a separate test, letters explaining that ballots are secret increased participation by two or three percentage points among registered people who had never voted. Some nonvoters misunderstood the privacy of voting. These examples broaden the book’s focus from choosing among candidates to shaping who enters the electorate. Registration and trust are not administrative details at the edge of politics. They affect which voices are present when the result is counted.

Analysis

This is the book’s most consequential shift in scale: campaigns do not merely compete for an existing electorate; their outreach can help determine who becomes part of it. That makes registration drives politically powerful even when they are formally nonpartisan. The work also shows why administrative clarity matters. If eligible citizens misunderstand ballot privacy or face avoidable paperwork, the resulting absence can look like apathy when it is partly a design failure. Yet efforts to expand participation require careful oversight, especially when targeting relies on inferred identity or party preference. Access and neutrality must be protected together.

Analysis

The book’s central achievement is to recast campaigning as a problem of learning under pressure. Elections are noisy environments: outcomes reflect candidates, events, and millions of individual choices, while campaigns have only one deadline and strong incentives to claim that their preferred tactic worked. Randomized field experiments offer a way to isolate some effects, but the story is not simply that science defeats intuition. It is that evidence becomes operational only when organizations can preserve it, interpret it, and change their routines around it.

The narrative traces three related revolutions. First, political researchers rediscover controlled experiments, from Gosnell’s early turnout tests to New Haven’s comparison of canvassing, mail, and calls. Second, analysts move from precinct averages toward individual-level prediction, using voter histories, surveys, and commercial data to find likely supporters and persuadable voters. Third, behavioral research reveals that turnout can respond to social pressure, planning, and trust, not only to information about candidates. The Obama campaign represents the fullest integration of these ideas, though its own data silos and operational failures show that scale is not the same as mastery.

A crucial distinction runs through all three developments: prediction is not causation. A model may identify who is likely to support a candidate without showing how to change that person’s behavior. An experiment may establish that a message raises turnout on average without proving it decided a particular race or will work in a different setting. The book is strongest when it makes these limits visible through the practitioners’ doubts, failed tests, and disagreements.

The resulting politics is more measurable, but not automatically more democratic. Social-pressure tactics can mobilize voters while intruding on privacy; microtargeting can reduce waste while making public argument less visible; and data-driven registration can widen participation while relying on delicate inferences about identity. The lasting lesson is therefore institutional and ethical as much as technical: test what campaigns do, retain what they learn, and treat every score as a fallible tool for engaging people, not as a substitute for their agency.

Summary in 30 Seconds
Campaign tactics are hard to credit in close elections, so randomized field tests compare contacted voters with control groups. In one study, canvassing raised turnout by 8.7 percentage points, while postcards added 0.6 and call-center calls showed no measurable effect. Social pressure can mobilize voters, but may provoke resentment. Individual data can sharpen targeting, yet predictions are not personal knowledge. Durable research teams help campaigns retain lessons, while ethical limits matter as much as measured results.
Contains spoilers

Last updated:

Report Issue

Review Summary

3.78 out of 5
Average of 1k+ ratings from Goodreads and Amazon.

The Victory Lab receives mixed reviews (3.78/5). Readers praise its insights into data-driven political campaigning and the evolution from mass advertising to micro-targeting, particularly Obama's 2008 campaign. However, critics cite issues with structure, too many characters to track, uneven writing style with unnecessary descriptions, and asymmetrical treatment of Democratic versus Republican sources. Many find it too focused on biographies rather than practical tactics, lacking actionable details for local campaigns. Political operatives appreciate the empirical approach to campaigning, while general readers find it dense and circular.

Your rating:
4.27
49 ratings
Want to read the full book?

Glossary

Randomized-control experiment

A test isolating an intervention

A study design that randomly assigns people to receive an intervention or to a control group that does not receive it. Researchers compare outcomes across the groups to estimate the intervention’s effect while reducing the influence of other differences. In campaign research, the intervention might be a postcard, phone call, or canvassing visit, and the outcome might be registration or turnout.

MicroTargeting

Individual-level voter prediction

A campaign method that combines voter records, surveys, demographic information, and sometimes commercial data to estimate individual voters’ likely preferences, turnout, or response to particular issues. It lets campaigns direct contact beyond broad party or neighborhood categories. Its scores describe probabilities, not certainties, and they do not by themselves prove that a message will change anyone’s behavior.

virtual IDs

Predicted voter preferences

Statistical estimates of how voters may behave when a campaign has not directly identified their preferences. Analysts build models from voters whose views are known, then use shared characteristics to predict the likely views of others. Virtual IDs help campaigns cover more people than direct calls or canvassing can reach, but remain predictions rather than confirmed statements.

social pressure

Influence from others’ expectations

A behavioral influence that encourages people to act in ways they believe their peers expect or commonly perform. In the book’s experiments, turnout increased when voters were reminded that voting records were public or that neighbors’ participation could be observed. The effect could be powerful, but more forceful versions also provoked complaints and raised concerns about privacy and coercion.

plan-making effect

Planning increases follow-through

The proposed effect that people become more likely to carry out an action after mentally rehearsing its practical details. In a voting context, a person might be asked when they plan to vote, how they will travel, and where they will start the trip. Such questions can prompt a concrete plan rather than merely remind someone that voting is important.

72-Hour Task Force

Republican turnout research effort

A Republican National Committee project launched after the 2000 election to improve voter contact and turnout operations, especially during the final three days of a campaign. It coordinated experiments on methods such as volunteer calls, mail, voter identification, and coalition outreach. Its results helped establish repeatable practices and made turnout research a party-wide priority.

The Matrix

Campaign contact data repository

A centralized Obama campaign data resource that recorded voter contacts, including calls, mail, canvassing, and exposure to broader campaign activity. It helped analysts connect interactions with voter records and examine patterns in support and turnout. The Matrix represented an effort to make data collection continuous and useful across campaign departments, though integration remained imperfect.

Bradley effect

Polls understate opposition to candidates

A term for a possible gap between what voters tell pollsters and how they vote, named after Los Angeles mayor Tom Bradley. In his 1982 California gubernatorial race, polls showed Bradley ahead, but he lost. The book discusses the concern that voters may conceal racial prejudice in surveys, making support for a Black candidate appear stronger than it is.

Analyst Institute

Progressive campaign research group

A research organization created by Democratic-aligned groups to coordinate experiments and share findings about campaign tactics. It brought political practitioners and behavioral researchers together to test voter contact, turnout messages, and other interventions. Its purpose was to build a lasting culture of evidence in a campaign industry where strategies were often repeated without reliable measures of their effects.

About the Author

Sasha Issenberg has authored four books covering diverse subjects including sushi business, medical tourism, and political campaign science. He extensively covered presidential elections from 2008 through 2020 for major publications including The Boston Globe, Slate, Bloomberg Politics, Businessweek, and The Recount. His journalism has appeared in New York magazine, The New York Times Magazine, and George, where he served as contributing editor. Currently, he works as Washington correspondent for Monocle and teaches in UCLA's political science department, bringing both journalistic experience and academic expertise to his work.

Download PDF

To save this The Victory Lab summary for later, download the free PDF. You can print it out, or read offline at your convenience.
Download PDF
File size: 0.34 MB     Pages: 16

Download EPUB

To read this The Victory Lab summary on your e-reader device or app, download the free EPUB. The .epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.
Download EPUB
File size: 1.41 MB     Pages: 21
Want to read the full book?
Follow
Listen
Home
Swipe
Library
Get App
Try Full Access for 3 Days
Listen, bookmark, and more
Compare Features Free Pro
📖 Read Summaries
Read unlimited summaries. Free users get 3 per month
🎧 Listen to Summaries
Listen to unlimited summaries in 40 languages
—
❤️ Unlimited Bookmarks
Free users are limited to 4
—
📜 Unlimited History
Free users are limited to 4
—
📥 Unlimited Downloads
Free users are limited to 1
—
Risk-Free Timeline
Today: Get Instant Access
Listen to full summaries of 26,000+ books. That's 12,000+ hours of audio!
Day 2: Trial Reminder
We'll send you a notification that your trial is ending soon.
Day 3: Your subscription begins
You'll be charged on Nov 24,
cancel anytime before.
Consume 2.8× More Books
2.8× more books Listening Reading
Our users love us
1M+ readers
Trustpilot Rating
TrustPilot
4.6 Excellent
This site is a total game-changer. I've been flying through book summaries like never before. Highly, highly recommend.
— Dave G
Worth my money and time, and really well made. I've never seen this quality of summaries on other websites. Very helpful!
— Em
Highly recommended!! Fantastic service. Perfect for those that want a little more than a teaser but not all the intricate details of a full audio book.
— Greg M
Save 62%
Yearly
$119.88 $44.99/year/yr
$3.75/mo
Monthly
$9.99/mo
Start a 3-Day Free Trial
3 days free, then $44.99/year. Cancel anytime.
Unlock a world of fiction & nonfiction books
26,000+ books for the price of 2 books
Read any book in 10 minutes
Discover new books like Tinder
Request any book if it's not summarized
Read more books than anyone you know
#1 app for book lovers
Lifelike & immersive summaries
30-day money-back guarantee
Download summaries in EPUBs or PDFs
Cancel anytime in a few clicks

Why one summary?

Scanner
Find a barcode to scan

⭐
⭐
⭐
We have a special gift for you
Open
38% off
DISCOUNT FOR YOU
⭐
⭐
⭐
$79.99
$49.99/year
Only $4.16 per month
Continue
2 taps to start, super easy to cancel
Settings
General
Widget
Loading...