I tweeted for Hillary Clinton for a year and a half. I learned some things.

I moved to Brooklyn in May 2015 to start as the first social media staffer on Hillary Clinton’s presidential campaign, 273 days out from the Iowa caucuses and 554 days out from Election Day. I tweeted for Hillary through more than 50 primary contests, 40-something debates, two conventions, countless sky-is-falling press moments, and the most soul-crushing, high-stakes, completely batshit-insane general election ever. I worked alongside (literally, inches from) the small, scrappy, incredibly talented social media team: Alex Wall, Julie Whitaker, Diana Fakhouri, Alex Witt, John Buysse, and Hiroki Murakami (not to mention the amazing design, content, video, and digital communications teams).

A year since the best day and the worst night of my life, here’s what I learned about social in politics, working for the first woman ever to win a major party’s nomination and the popular vote, and running against Donald Trump.

Hillary and Trump were the good and evil sides of the same social media coin.

In some ways, the role social media played in 2016 was bigger than Hillary or Trump. Social media is democratic and ubiquitous — two powerful and dangerous things. It can be a tool for empowerment and connection, or, as we’ve seen with the snowballing revelations on fake news and Russian-sponsored ads, it can become a poisoned water supply. Trump represents the most serious threat of political social media: unfettered access to voters, with no accountability or fact-check. Total lack of constraint becomes “authenticity,” which supersedes sanity or competence. It’s an approach that, like the rest of Trump’s ideology and policy agenda, assumes voters are too stupid to realize they’re being fucked with. It bets that people will care more about shock and entertainment value than literally anything else. (As a jaded former Hillary staffer, I’m not sure he’s wrong about the second part. But.)

We did the opposite. Our strategy was fundamentally about respecting voters’ intelligence: We presented the evidence of why Hillary was the superior choice for president, first over another progressive and then over a lunatic. We offered the facts in a cycle driven — from the press to the opposition to the FB-damn-I — by misrepresentation, false equivalence, and lies. When Bernie Sanders said Hillary wasn’t qualified, we didn’t say, “What a jackass” — we laid out Hillary’s decades of experience, newspaper endorsements, and proof of her preparedness. To show voters why Trump couldn’t become president, we didn’t just say, “This guy is fucking nuts” — we offered up his own words and record and let voters decide for themselves. We showed who Hillary is and what she was up against, and we trusted that voters would make the right choice in the end. Sixty-five million voters proved us right.

Still, yes, we failed to reach the right voters. The next campaign will have to balance compelling, authentic content strategy with meticulous targeting. (But you can’t do only one or the other; digital ads alone will not win us an election. People know when they’re being advertised to. And I refuse to believe that voters will get burned twice; there will undoubtedly be more bots and sinister memes, but they’ll meet savvier readers.) In a cycle saturated with fake news, our content strategy was about presenting incontrovertible facts. Given who we’ll be facing again, this will be all the more important in 2020.

But more than our own strategy or Trump’s…whatever he had, 2016 was a warning about what social can do and will become. I’ve seen arguments that social will become less valuable than other outreach tools, like ads or video alone. Respectfully: bullshit. Social media will matter more and more in every election to come. More voters will look to social media as the gateway to everything else — their first stop in learning about candidates and policy and voting. Even more voters will seek out and consume their daily news on social media, maybe surpassing traditional media. Candidates’ distinctive voices and the size of their platforms will determine how effectively they can communicate with voters, even more than in 2016. But social media is dark and full of terrors: All of 2016’s trends toward polarization (driven by the Facebook algorithm, webs of who-follows-whom influence, and politics itself), fake news, fake accounts, and fake memes will only grow. That means the next campaigns’ social teams will be the front line. This work is not entry-level. In addition to surgically-crafted content strategy, those teams will need to figure out how to get around the obstacles between them and their audience: how to fight fake content, break out of echo chambers, and persuade and turn out voters.

Working for a “cool” candidate is easy. Working for the right candidate is harder.

Social media strategy in politics is about introducing your candidate to voters who will never meet them in person. Social media strategy for Hillary Clinton was about simultaneously introducing her to voters and methodically poking holes in 40+ years of stubborn misconceptions about her. Still, on every parameter from experience to temperament, Hillary was unquestionably the right person to be the 45th president.

But of course, she is a woman, and her astronomical approval ratings while in office meant nothing as soon as she sought a promotion. The next woman candidate for president is going to deal with the same shit. Her campaign will have to make an impeccable case for her candidacy while likely having to be as scrappy in building support for her as we had to be.

We were never the cool candidate. It was very, very hard to convince people to “come out” with their support of Hillary. It had to be a different strategy than Obama’s or Bernie’s, who could take for granted that people would like and be willing to share their ideas or likenesses. We started from scratch, deconstructing the things we loved about Hillary — smart, wonky, earnest, serious, badass, indomitable — and figuring out how to make them resonate online. Instead of someone else’s version of “cool,” we showed why knowing issues inside and out and never, ever quitting were worthy of the same respect. We amplified the voices of hundreds of other people who validated our case, from newspaper editorial boards to global leaders, past constituents to passionate supporters, and even the sitting president. If you weren’t going to trust Hillary, we’d show you a chorus of voices you had to trust. Again, incontrovertible evidence to back up our case for the presidency.

We didn’t follow anyone else’s playbook for social content. We didn’t do fluff. We tried, then quickly abandoned Twitter-native cutesy tactics. We went hard on policy, even when we got mocked for it (and we did), and even when the rest of the political media weren’t interested in policy. We gave Trump a taste of Hillary’s signature sass the deserved number of times, but mostly we didn’t mess around: We took things seriously, we called out the racism and misogyny coming out of his campaign, we kept it about the issues, and we made a thoughtful and thorough case for why Hillary should have been the president. In other words, we were true to Hillary’s own personality, occasionally with a social twist.

And we were smart about branding. 95% of social media design out there is branded within an inch of its life — two or three colors, a logo, almost always a photo of the candidate. We created products that intentionally didn’t have Hillary’s fingerprints anywhere on them, to show people that they agreed with, admired, or supported her, even if they didn’t think they did (we called it, semi-jokingly, “un-design”). We took a page out of Bernie’s book and wrote value statements that every progressive agrees with, whoever the messenger. Meg Vazquez, our brilliant social design lead, made pieces using every color, treatment, and type of photograph imaginable while still staying true to the brand. We made a whole series about other trailblazing women that you can’t help but admire — and we built affinity with and respect for Hillary without ever mentioning her name or showing her face. Steadily, we started to see heartening feedback that ranged from “I hate Hillary, but she’s onto something here” to “All right, she’s got my vote.”

Did our shrewd, methodical strategy work? When I started on the campaign, anything that came from Hillary on social media was toxic to all but a fierce band of (mostly female) supporters. By November 2016, we’d broken the internet a few times, went viral by any definition on more posts than not, and saw our most high-priority, messagey content perform best. We were able to tell the kinds of “Hillary is a badass” stories we couldn’t in 2015, and even reluctant voters had to agree. You can’t measure changed minds by vanity metrics, but you can measure shares, reach, and growth over time — and we got that. A lot of it.

Stop chasing Beltway headlines. It will kill you, strategically and mentally.

We didn’t do it all right. The biggest mistake we made as a social program was over-valuing the opinions and attention of political reporters, too often over setting our own narrative or reaching real voters.

Some (smart) people think social IS only useful for reaching reporters, not voters. Again, respectfully: bullshit.

Even if our accounts didn’t have millions of non-reporter followers who cared enough about the content we published to share it with their own friends, chasing the Beltway echo chamber every day is an exhausting exercise. We had enough rapid response just dealing with either defensive stories (good christ, the emails) or opposition (literally any time Trump opened his mouth). Most of the time, we couldn’t control the narrative, no matter what we did. I blame the press for following Trump’s empty stages and outbursts like the Pied Piper, and at the end of the day, we did what we had to do to stay on top of the news cycle. But we should have saved ourselves the head- and heartaches by telling our own story instead of swinging at every pitch.

The single most valuable thing on any campaign, along with money, is time. We spent time workshopping clever lines and developing flashy, flash-in-the-pan content when we could have been talking to voters about what mattered most: Hillary’s plans to make the country better for them. In the final GOTV weeks of the campaign, when the only thing we needed to do was get voters to the polls, we were still sparring with Trump. In 2020 (and, probably, 2018), it’s going to be tempting to take Trump’s bait and clap back at every provocation, but whoever does our jobs then should keep their eye on the prize.

Fighting for what’s right is worth it.

Working for Hillary Clinton for a year and a half was the greatest privilege of my life. Giving up every aspect of my life to do whatever it took to try to stop Trump was worth every second, whatever the outcome. I’m a better person and a better strategist precisely because every one of the circumstances we faced was utterly unusual. An incredibly controversial candidate. The first woman. An unhinged, quadruple-bankrupted, callous, racist sexual predator on the other side. We made some bad decisions, but we made a lot more good ones.

If you’re a social strategist (or a writer, designer, or videographer) — work on a campaign. Don’t just tweet. We need you. Work on a 2018 race in your state to take back the House and cripple Trump’s agenda (as of this writing, we are in extraordinarily good shape to do just that). Work on a 2020 race to make the Democratic Party stronger and then, for the love of god, fire Trump. Social media will be one of the most important battlegrounds. Figure out your plan and fight.