9 Lessons From Politics for Data
Lessons from my time in politics that I still use in data.
At the end of 2007 I knew more than most on how to plan and run effective election campaigns, quite a lot on communication, a reasonable amount about motivating volunteers and fundraising and almost nothing about data management, analysis, reporting and beyond.
So it was a brave when The Labour Party trusted me to build the UKs first national database of voters and their opinions. Fast-forward to 2024 and Contact Creator is still helping Labour to win elections and I’m VP of Data with Taxfix in Berlin. I still benefit every day from the lessons I learnt from working in politics.
Here are 9 of them.
1. Focus on what matters
Politics, in the UK at least, is about margins and where you make a difference. There’s no proportional representation so elections are a collection of individual contests. Some contests will be won or lost by a lot; Labour has held Liverpool Walton at every election since 1961. Whereas it won Hendon by 15 votes out of over 41,000. Other seats make the difference; Dartford has been won by the winning party in a UK General Election since 1964.
I learnt this lesson in the first election campaign I organized in Hammersmith & Fulham, London, in 2002. Working for Labour, we won 28 (61%) of the seats with 41% of the vote whilst the Conservatives gained 18 (39%) of the seats with 42% of the vote. How? There were just over 100,000 votes cast in that election - the six smallest labour majorities added up to 161 - under 0.2% of the vote. We focused our efforts and beat the odds.
Data teams often spend too much time trying to support everyone; whether it is new product features, marketing measurement, operational efficiency, sales performance, FP&A support or investing in AI or ML. Everything goes on the backlog and we never say no. The result - many of our best opportunities are left untouched.
Learn to prioritize and don’t keep it a secret. My advice is to focus on those team and opportunities that bring the best value. This could change over time. It could be getting the experiment right on a big feature bet, it could be predicting expansion accounts, it could be giving the right support to FP&A during budget planning. But have a framework.
Personally I like RICE :
It is a great, quick, framework to start to reason about choices when it comes to priorities. Use it to start a discussion - not as an absolute and infallible scoring system. If you only work with a limited range of stakeholders, then you can simplify it further by dropping reach and making it ICE.
2. Mistakes happen, get over it, learn from them
Ever been asked to explain a mistake you learnt from in an interview? I have no problem with this question. For the 2009 European Election we prepared a highly personalized mailing for millions of households, taking account of known or likely voting preferences, the latest in demographic targeting and local information. We hired one of the highest volume direct mail printing works in the country. And we got our selection wrong and we didn’t notice. Until it was too late.
The result? Over a million people who we thought were “NOT” likely to vote for Labour got a letter intended for people who were, a massive missed opportunity and a lot of money. Once we realized there was nothing to be done, I think the post-mortem lasted about 20 minutes. We were simply too busy to get bogged down in what had already taken place.
Going forward we improved our quality checks and broke down complex targeting into simpler chunks. We fixed the problem and moved on.
I’m a believer in retrospectives and in continual improvement. Too often though in data and tech, the retrospective lasts a long time, with an even longer list of actions, feeding into an ever extending backlog and reporting up to the Director, VP, CTO and CEO one-by-one (well depending on just how big the problem was). All the time, the poor person who made a mistake - often under massive pressure - is left reliving it time and time again.
My advice, keep the learning, the sharing of the learning and the resulting actions quick. Make sure your actions are things you will actually do and make sure they improve the situation going forward. And move on.
3. Know when deadlines are real
Election day is election day; it doesn’t get delayed because you aren’t ready. Get ready as best you can and go for it. This tends to lead to two things in politics:
- Everyone works incredibly hard and oftentimes hours that cannot, or should not, be sustained.
- People get used to making hard choices about what will be done because you can’t move the deadline to get everything done.
An early example of this for me is the daily campaign reporting we did on Contact Creator, the election database we developed. The dream was full automated reporting landing in the inbox of every MP, campaign organizer and political campaign leader by 8am each day.
The reality was me running reports on my 1 hour long train commute from Ipswich to London, using the 3G connection of my phone, Crystal Reports on my laptop and watching out for the dead signal areas. But we got the most important information out, to the people making the biggest decisions and kept our part of the campaign moving.
Now we all talk about agile development and many of us evangelize about having a skateboard level minimal viable product. But how often do you really keep it simple, how good are we are really getting the early learning in, how often do we either agree to just a few more features or underestimate the effort? And what is the consequence? Either delay - and the cost, disappointment and lost opportunity associated with that. Or rushing to chop and change scope - or more often quality - at the last minute and then being disappointed when a poorly executed solution fails to meet expectations.
If you are a seasonal business and you know that your deadlines are real - that you really can’t move Christmas - then respect that. Then you can focus on making sure that what you can deliver is the best it can be.
4. Leave your dogma at the door
Most political campaigners and a good number of successful politicians are not fixated on the fine details of their political philosophy; you shouldn’t be fixated on purity of your data strategy.
Outside of politics, a lot of people think of those working in it as the most committed believers in some ideological viewpoint, be it socialism, free market libertarianism or something else. In reality, elections are won by getting normal people to change their mind - either to switch who they vote for or to start voting. And, by and large, normal people are not extreme in their views. They want a job, some security, a good future for themselves, friends and family and their health. And ideally to spend not too much of their life thinking about the government.
So people working in politics tend to focus on winning over people in the center. For a start they are often somewhat willing to change their minds, and secondly there are a lot of them. Kier Starmer just became Prime Minister of the UK with a huge landslide and he did it by winning the center. And winning it well.
There’s a lot of ideology in data. Are you an advocate for Data Mesh or Data Fabric? Do you believe in centralization or decentralization? Are you a fan of data warehouses, data lakes or lakehouses?
These decisions matter; although the right decision is rarely purity of one idea or another. What matters more is the outcome and the results you have on the business. Many colleagues outside of data and tech will have limited interest in your technology and design choices. They will have much more interest in the insights and results you deliver. My most entrenched opinion is to start by understanding your business and customers.
5. Decisions don’t have to be perfect but do have to be made
And the British people sometimes will forgive a wrong decision. But, you know something, they won’t forgive not deciding. They know there’s not some fantasy government where nothing difficult ever happens. Tony Blair, Labour Party Conference, 2006
Tony Blair remains one of my political heroes and his advice on making decisions applies just as much in business as it does to running a country. Leading Britain into the Iraq War was a big decision, based on imperfect information, where no consensus could be achieved. More than 20 years after it was made, the rights, wrongs and consequences still are not clear.
The decisions your teams make won’t be so consequential or so controversial and yet we find ourselves putting them off. Or we find ourselves giving options but not opinions.
When I worked at Contentful, we were running two tracking solutions - Snowplow and Segment. Having two tools for the same job introduced confusion in analysis, duplication in our technology stack, overhead in maintenance and increased costs. Choosing which to keep was hard, and we kept Segment. Looking back I often think we chose the wrong tool based on short-term marketing needs instead of long term flexibility, quality and ownership. But we ended up in a better place, with higher quality data, lower costs, clear ownership and easier analysis.
I’m not advocating that you take all the decisions as a leader; in fact I believe in empowering teams. But ensure decisions are made. One really useful tool for this is the RAPID framework.
6. Understand the motivations of your team
Volunteers are the lifeblood of most political parties and campaigns. In the UK they put up posters, make telephone calls, knock on doors, deliver leaflets, stuff mailings in envelopes, build websites, run your local social media, run your local campaigns, cook food for your fundraisers and then buy tickets for the same event and much, much more. And they do it all for free.
Back when I was organizing election campaigns in Croydon, one couple would regularly volunteer to fold 4,000 to 5,000 letters a weekend into envelopes - and then give me tea and cake when I turned up to collect them.
Volunteers help for many reasons but the three most important are that they believe in your cause, they feel valued and they believe they are making an impact. Motivating them is fantastic training for motivating a team that actually gets paid.
Take the time to understand what the goals of your team members are and don’t be afraid of them. You might not have that manager job they want, you might not need more Staff Engineers. But you can help them grow towards these goals. You can help them gain these skills - and when the time comes maybe you’ll have an opportune vacancy or maybe they will become an advocate for you elsewhere. In the meantime, they can enjoy growing with you.
Have a look at The Alliance Framework.
7. Don’t be too in awe of anyone
Work in politics long enough and you will meet a lot of important people: Members of Parliament, Lords, maybe the Prime Minister. One thing you eventually realize is that they are actually quite normal. And in my experience, the very best of them treat everyone with great respect and truly appreciate their contributions.
As I first started working with Members of Parliament, particularly trying to get them elected, I was somewhat in awe and would spend great effort trying to achieve anything they asked for. Oftentimes, over delivering something that really didn’t need it.
To be successful, and to make them successful, you actually need to see them as a colleague. Be prepared to challenge their views, work with them to improve their ideas and, when something is a five minute job actually just do what needs doing in five minutes.
It can be the same with CEOs or other senior colleagues. When they ask for the results of an A/B test, maybe they do just need pointing to the result dashboard and don’t need two analysts to spend a day compiling a deck and deep analysis. That can always come next if it becomes apparent it is needed. And when your VP asks whether you have considered some new tool, maybe it is OK to say that what you’ve got is working well and not setup a head-to-head two month trial.
Always be proud of what you bring to the table, of your ideas and the worth of your contributions and theirs. Bring this to the conversation whether you are talking with an analyst intern or the CEO.
8. Get (really) bored of saying it
How often has your team presented the result of an analysis, spoken about the importance of data quality or the risks associated with an prediction…and a few days later no-one seems to remember. You are not alone.
When You’re tired of saying it, they’re only just starting to hear it. Jeff Weiner, ex-CEO LinkedIn
Politicians and marketeers, the two groups are a lot alike, have known this lesson for a long time. To get your message across, keep it clear, keep it consistent and repeat it frequently. Don’t make the mistake of thinking that everyone else hangs on every word you say - and then reviews their notes afterwards.
9. Pick yourself up after a setback
Winning and losing are at the heart of politics. Every contest has a winner and sometimes it isn’t you. In 2005 I organized the election in Croydon, South London. Our candidate lost by 75 votes, or 0.2% of the vote. I remember being phoned up by the press at 10:30am after about an hour in bed, the election count having ended up 8am, and being asked why I thought we lost. “Because the other candidate got more votes” was my answer.
At the time, the candidate, myself and our volunteers were devastated. This had been the main goal of two years work and I personally had averaged 85+ hours a week on the campaign for at least six months, probably 70+ hours a week for the 18 months before that. How do you motivate yourself and your volunteers after that?
During the campaign a long standing local councillor, Mary Walker, had died and the rules said we had up to six months to call an election for a new councillor. I chose to call it immediately. Why? Because people deserve representation and we needed a new goal to get behind.
The weekend after losing by 75 votes, we were out delivering leaflets, knocking on door and trying to win another election. It worked. We won and we moved on from our losses.
The point here is that you can’t change your losses, you can’t make the failed A/B test a winner, you can’t rerun that presentation that went wrong, you can’t stop the fact that you lacked the data needed to build a good predictive model. What you can do is learn, and ask what’s next. Have a retrospective, decide what to do different next time and start to act on those learnings as soon as possible. And do it together.