operating/product plan for Twitter
Twitter has immense potential. Here are my thoughts on how to get there.
We are deviating from normal startup musings to talk Twitter.
Today at a glance:
thoughts on the Twitter challenge
few things before we get into the growth plan
a simple 4-state tree of how I see this playing out
the plan: a set of ideas for how to get this done, focused on product & ops
what not to do right away
Thoughts on the Twitter challenge
Twitter is both the #1 western discourse platform and the tool for shaping culture/public opinion. Twitter is the multi-player information layer of today (YT/Snap/Insta are single-player and FB feed is a walking corpse).
Twitter has immense potential.
Whoever leads Twitter has one of the toughest (the toughest?) challenges in product/business for 2 reasons:
foster open dialogue and information sharing in a society that is more polarized than ever
achieve #1 while running a sustainable and growing business
#1 is hard. We may think we want this open network, but there is a % chance that we play out all the scenarios and we learn that we are unable to handle an open network given how we are wired as a species. Many said that #1 is not an engineering-first problem. I think they are right.
#2 is hard because Wall Street is insatiable, short-sighted, and tends to reason by analogy (i.e. why can’t Twitter circa 2015-2020 be more like FB).
Think Musk is right in taking the company private to solve #2.
All tough problems are worth solving. There are many hot takes on how to run Twitter so here is another one.
Few things before we get into the growth plan
Saying that Twitter has immense potential does not mean that Twitter is weak today. There is a lot of strength and there are positive trends (ex. shipped things in 2022).
You can make tough calls, build fast, AND be human. It’s more important than ever to be human and to treat people with kindness (i.e. fair landing for the current team).
To save time, I will skip the how behind each move. I’ll also skip culture, incentives, and team dynamics. Last, I’ll skip win conditions/what has to be true for us to go back to public markets (if ever) and if we do go public, how do we align with investors on what is reality for the business.
How we got here (i.e. act of taking Twitter private, growing political importance of twitter, 5 CEO shuffles since 2006, law suits) was “expensive”. We said that Twitter is a difficult problem and that Twitter is worth fighting for so expensive is implied. The point is that it cost us a lot to get here. Twitter is private. Now is not the time to be timid or reckless.
Poker analogy would be that we are 300+BB deep and we are running a 3-street bluff b/c we think our opponent is weak. We bet turn with the intention of betting most rivers. Once river hits (Twitter’s crucible moment) - it is not the time to be timid. We must follow through. That’s part of the reason why I recommend path D below.
Most of my ideas are around product and ops. Solving policy is probably the toughest question and beyond a few suggestions, I don’t know enough about this to have a strong POV.
A simple 4-state tree of how I see this playing out
I believe that leadership has to be bold and thoughtful (not reckless). Twitter has some time to get things right because there isn’t a clear alternative yet and everyone’s social graph is tied up on Twitter.
There are 4 paths:
A = torch it. Do not just fail to handle current moderation/policy/bot issues, but exacerbate them, leading to an exodus of users and advertisers. Fail to build the team. Fail to get the P&L under control while facing declining revenue from advertisers and dealing with interest payments on $20B of debt. A nuke would be a tragic bad act that spawns from Twitter hate mob and leads to further loss of trust. Twitter ends up like Tumblr.
B = high variance moves with a neglected core. Make bold product and policy moves without addressing the core experience. For example, this would be going after ideas #5-6 in the bigger idea list below head first. There is a X% chance some of them work, but there is a higher % chance that they fail, especially when layered on a subpar core - hence the spread.
C = upgrade the core. Address the core experience, find a reasonable solve to moderation/policy, take care of the bots, reduce reliance on advertising revenue, and straddle the middle in a way that is healthy for society. This will lead to higher EV vs. status quo.
D = moonshot turnaround. This is (C) + ship and find product market fit for all V2+ ideas around payments and social. Sequencing matters. Go for bold bets without the core and we end up in state B. This is a max EV scenario.
build lean team + get P&L under control
generate cash flow via cost cuts + blue v2 service to pay down debt
make nice with advertisers while diversifying away from ads with an on-platform service (blue v2)
handle the core (ex. policy, what behaviors get rewarded, bots) before going into bigger bets
ship a # of quick product wins
ship bolder bets (i.e. choose your own adventure, payments, social) and expand
let go of X% → listen to trusted top performers. Do it in one cut, do it now, and if on the fence - cut more than necessary. Those leaving - treat them with kindness. #1 reason for cutting is speed of execution. Nothing more.
fill leadership team → help needed
G&A is 46% of R&D → figure out why so high and cut to ~10%
Sales & Marketing is 94% of R&D → figure out why so high and cut to ~25%
make a plan to pay down $20B in debt and to reduce interest payments → renegotiate if X milestones hit
Product: quick wins + derivative ideas
unbury muted words from settings UX, rename to filters, make it accessible from timeline and grow adoption. Easy client change.
outright disable quote + reply → high % are hate/negative
beef up Twitter blue + raise price to $9.99: (1) better follower mgmt (search followers, mass actions), (2) better DM management, (3) auto-blue checkmark if verified ID, (4) analytics on followers, (5) better tools for writers - many ideas from hemingway clone to UX to auto-thread maker to preview etc., (6) post boost to followers, (7) ad free, (8) themes, (9) edit button and much more
allow to mark every item in #explore/#discover as not interested/spam + blacklist all fortune/hollywood reporter/empty media. High ROI real-estate and 50% of this is noise today.
dial in <content you might like> → ~20% seems random/spam
reward positive contributors with low follower count w/ profile icon or free Twitter blue for X months. Today a user with low count brings a meme or a trend forward → it gets picked up by popular accounts → user X fades. Need a better feedback loop.
Bigger/harder ideas (prioritized list)
re-eval core loop: today vitriol/dunking on others drives the discussion and gets rewarded by likes. Figure out how to flip the model and reward +EV behaviors/reasonable people. Maybe it’s equivalent of reddit karma that can be cashed out for $USD or spent on boosting your posts OR posts of others you think add value to discourse?
get a grip on bots
reward verified accounts: validate identity-> get checkmark and appear in algo & if NOT -> downrank in feed
build more resilient anti-bot system and tune it. There will be false positives and that is okay. Figure out ranking for each variable.
IF [(unverified) + (high degree of amplify vs mean) + (high degree of anon in profile like alphanumerical name) + (post frequency above mean w/ gradients >50/day, >100/day) + (followers have high degree of anon in profile) + (low follower count) + (liked/RT ratio is +/- 10%)]→ take action (ban, downrank, label as bot)
Use account correlation as heuristic to enable near real-time action vs bots (assuming bots act in tandem) → do not wait days to gather data on account X
if bot is benign account -> add bot flag in profile
choose your own adventure: users can select the experience they want among algo presets (pay to tune your own or get access to presets used by XYZ person you respect) + tune your timeline based on (geo, interests, live events etc.) This can be powered by developer-first marketplace.
experiment with discovery/engagement ideas: (1) top 5 tweets while you were AFK (may be missing from timeline but high EV content), (2) top 5 tweets liked by X where X is your followers, people you follow, people you follow + 1 degree removed, people from your location, (3) engage the lurkers in dialogue via preset polls or thoughts on X/link to current factual article
payments + wallet: add P2P payments and focus on fiat (crypto later). Allow tipping with 85/15 rev share, simple p2p payments, and withdraw to ACH. PayPal will eat itself plus we already have the social graph. Do 50% lower chargeback fees, no extra FX fees, and add micro-payments support (higher % trnx fee, 2-3x lower flat fee on transactions <$5). Creators will appreciate this and $TPV will grow.
social: tear down popular messaging/social apps, starting with KaKao, WeChat, Telegram (audio as input, gifting, commerce, calendaly/schedule sharerer, shopping, share location, calls, events, draw etc.) —> Musk alluded to this being the plan. Think it’s smart, but its much more than a copy-paste. Again - need to invest in the core loop before layering any of this.
Service + Developers
alerts on all activities/accounts outside the mean and if these alerts already exist - a strike team to take action 24/7 (ex. shirts that go hard, bot networks from region X = strike down in hours not weeks)
drop rate limits on API, but punish project IDs who tweet (bots?) and reward IDs who help improve the platform (i.e. fight bots, improve curation, help greater good, develop end-user feed algos)
document well (i.e. stripe)
council to decide on bans and guidelines -> seems to be WIP
EULA in plain English: at signup click through key terms and agree (not buried in ToS) + all existing users consent
actions on platform that incite violence/abuse lead to bans
appeals system for bans
80/20 revenue share for creators/developers sliding to 90/10 if earning on twtr in year 2+
this is bird in hand. Can’t let go even if we don’t love the biz model given long-term vision of the platform. So we must make it work
go on listening/inspire tour to damage control + instill confidence
evaluate CPMs -> twitter ad buy has been expensive vs others. Why?
invest in targeting and look-alike audiences
give brands control over context in which their ads show up: <IF X ad/product - never show my ad next to Y content/keyword>
~55% of total adv revenue is coming from US (not NA, but US)
twitter dominates in West among white liberals with a college degree
expand market to market -> this is both geos (ex. LATAM/SEA) and demographic (ex. hispanic/Rep leaning) -> diff playbook for each niche
What not to do right away:
pursue too much at once and/or attempt to clone Chinese apps (Vn) right away. Sequence product bets starting with core loop. What behaviors are rewarded? Get the core under control.
video/streaming → will drive up costs + more moderation/policy problems. Let TikTok/Insta/YT own video for now. Must focus.
disparaging anyone. Not a popularity contest. Be civil
go full boar into crypto on day 1→ majority do not care yet and web3 onboarding is a pain for 99% so $TPV and platform take-rate will be minimal. Better to focus on other areas. Can lean into web3 when have policy and P&L under control. Web3 will be here, but need to deal with the core.
stop listening/drop humility
add the edit button unless paid = immutable is core to twtr
Oh hire Rahul Ligma for PR/Comms!
That’s it! Go Twitter!