LinkedIn Post Ideas for Heads of Growth

47 experiments shipped by my team last quarter. 6 clear wins by our pre-registered success criteria. That's a 13% win rate, and I want to walk through why that's actually a good quarter, not a bad one. The portfolio logic: at a 13% hit rate, if your average win moves the target metric by even 4-6%, and you're running enough volume, the compounding effect outpaces a team that runs 8 highly-vetted experiments a quarter at a 40% hit rate but smaller individual effect sizes. We modeled this explicitly — expected value per experiment slot, not just win rate. The failures broke down three ways: 24 were flat, genuinely inconclusive, useful mainly for ruling out a hypothesis. 12 were negative — the experiment made the metric worse, which is data, not waste. 11 had implementation issues that invalidated the read entirely, and that's the number I actually want to shrink, because that's process failure, not learning. The six wins, combined: an 8% lift in trial-to-paid conversion. That's the number that goes to the board. The other 41 experiments don't get a highlight reel, but they're why we knew where to point resources for the six that worked. Growth culture that only celebrates wins trains people to stop reporting the 87% that didn't work, which is exactly the data you need to get better at picking bets. What's your team's real win rate, honestly?

I've stopped using the phrase "growth hacking" internally, and I want to explain the actual shift, not just the rebrand. A hack, by definition, decays. The Instagram contest funnel that worked in 2017 doesn't work now. The cold email sequence that got 40% open rates two years ago gets 8% now. Growth hacking optimized for tactics with a shelf life measured in months. What we build instead, and what I'm calling distribution engineering: channels treated like products, with their own roadmap, their own maintenance budget, their own depreciation curve we plan for instead of get surprised by. Concretely: our SEO content engine isn't a series of one-off articles anymore. It's a system — a content production pipeline, an internal linking structure that compounds, a refresh cadence for decaying pages, owned by someone whose job is that system, not "growth marketing" broadly. Same with referral. Same with our partner integration channel. Each one has an owner, a quarterly roadmap, and a build-versus-maintain budget split, the same way an engineering team plans a product. The mindset shift this forces: you stop asking "what's the next hack" and start asking "which of our three channels needs investment this quarter to compound, and which is showing decay signals that need a rebuild." Distribution engineering is slower to start and much harder to kill. That's the entire point — durability was never what hacks were built for.

Our funnel data said all channels performed roughly the same on activation. It was lying to us, in a specific way I didn't catch for two quarters. I pulled every churned user from the last twelve months and cohorted them by acquisition source, then looked at time-to-churn, not just churn rate. What that revealed: one channel — a comparison-content SEO cluster — brought users who converted to paid at a perfectly respectable rate, but churned at double the speed of every other channel, median 47 days versus 94 days elsewhere. Aggregate churn numbers hid this because the channel was a small enough share of volume not to move the blended average much. What that channel was actually doing: bringing tourists. People comparing us against three competitors, converting on a trial-ending push, then churning the moment the comparison research was "done." High intent to evaluate, low intent to commit. Our best channel, by contrast — a partner integration referral source — had the highest 90-day retention of any channel we had, by a wide margin, despite a lower raw conversion rate. That channel brings citizens: people already inside a workflow where we're the natural next tool, not comparison shoppers. We reallocated budget away from the comparison-content cluster and doubled down on partner integrations. LTV:CAC on the reallocated budget improved by roughly 60% over two quarters. Churned users tell you which channel brought people who were never going to stay. Studying them beats studying your winners half the time.

Our referral program launched with a flat $20-for-both incentive. Six months in: referral share of new signups stuck at 2%, and the ones we did get skewed toward low-intent accounts that churned faster than average. The diagnosis: the flat incentive attracted the wrong inviters. It was just large enough to motivate people optimizing for the $20 itself, not people who actually loved the product enough to vouch for it. We were paying for referral volume from exactly the segment least likely to bring good customers. The fix: we restructured the reward around usage-based tiers instead of a flat cash incentive. Inviting a friend who reaches an activation milestone within 30 days earns a meaningfully larger reward than one who merely signs up. We also changed who's eligible to send invites at all — only accounts past their own 60-day activation milestone can refer, on the theory that people who haven't gotten value themselves can't credibly vouch for it. Result: referral share of new signups went from 2% to roughly 8% within a quarter — a 4x lift — and the churn rate on referred accounts dropped below our blended average for the first time since launch. The mechanism matters more than the headline incentive size. A referral program is a selection mechanism for who refers, not just a reward mechanism for how much. Get the selection wrong and the reward size won't save it.

Our blended CAC doubled over two quarters. Here's the exact diagnosis tree my team used to find out why, because "CAC went up" is a symptom, not a diagnosis. Branch one: channel saturation. Did our top channel's audience pool shrink relative to our spend? We checked impression share and frequency — frequency had crept from 2.1 to 4.8 per user, meaning we were re-showing ads to a saturated pool instead of reaching new people. Branch two: creative fatigue. Did response rates decay on the same creative over time, independent of audience size? Click-through on our top-performing ad had dropped 40% over eight weeks with no creative refresh, a classic fatigue signature we'd been slow to catch. Branch three: mix shift. Had our channel mix shifted toward inherently more expensive channels without anyone deciding that deliberately? Yes — paid social spend share had crept up as SEO's relative output slowed, and paid social was 3x our SEO-driven CAC. Branch four: attribution drift. Had a tracking or attribution change quietly reassigned credit in a way that made CAC look worse without underlying reality changing? We ruled this out after an audit, but it cost real time to rule out. The actual answer: 70% mix shift, 20% creative fatigue, 10% saturation. We refreshed creative, capped frequency, and rebalanced spend back toward SEO. CAC came down 35% within six weeks. What's your diagnosis tree look like for a CAC spike?

I blocked my calendar for a full week and did onboarding calls myself, no delegation, five months ago. As head of growth, this isn't supposed to be my job day to day. It should have been. What our funnel data showed before that week: a clean, evenly-distributed drop-off across onboarding steps. Nothing alarming, nothing actionable — the kind of data that lets you assume things are basically fine. What I heard in the first five minutes of nearly every call: people didn't understand what to do after connecting their first data source. Not a UI problem exactly — a sequencing problem. We'd built onboarding assuming people would explore; most people, it turned out, needed to be told the next concrete action, explicitly, every single step. Two changes came out of that week: an explicit "do this next" prompt after every major onboarding action, replacing an open-ended empty state, and a change to our welcome email that named one specific first task instead of three options. Activation jumped from 41% to 53% within three weeks of shipping both changes — the largest single activation gain we'd seen in over a year of smaller experiments. Nobody on my team was doing anything wrong. The funnel dashboard just can't hear confusion the way a live call can. If your activation's been flat, block a week and do the calls yourself before running another dashboard-driven experiment.

Five growth metrics I've caught lying to my own team, and how each one does it. 1. Blended CAC. It hides channel-level decay behind a healthy-looking average, the same way one great channel can mask two dying ones for a full quarter before the blend finally turns. 2. Activation defined too early. If your activation event fires on account creation or first login, you're measuring signup completion, not value realization. We moved ours three steps later and our "activation rate" dropped 20 points overnight — a more honest number, not a worse one. 3. NPS as a retention proxy. Our NPS sat at a healthy 42 for two quarters while churn rose. NPS measures sentiment among people who bothered to respond, which skews toward engaged users, not the quiet churners who never fill out a survey. 4. Total signups as a growth headline. Signups with no activation floor reward top-of-funnel volume regardless of quality, and I've watched teams optimize this number straight into a worse customer base. 5. Experiment win count without effect size. Ten "wins" that each move a metric by 0.5% look impressive on a slide and mean almost nothing compared to two wins that move it by 8% each. Every one of these looked fine in isolation and lied by omission. Pair each with a second metric that would catch it decaying, or it will fool you exactly when you can least afford it.

Cheap content arbitrage is dead, and I mean the specific thing, not content marketing broadly: churn out volume, rank on long-tail terms nobody else bothered writing, harvest the traffic before anyone notices. AI made producing that content free for everyone, which means the arbitrage — the gap between production cost and traffic value — collapsed to zero. Where we're reallocating the budget that used to fund that: First, proprietary data content — reports and benchmarks built from data only we have, which AI can't replicate because it doesn't have our dataset. This is now our highest-performing content category by lead quality. Second, distribution partnerships — co-marketing and integration placements where the content lives on someone else's audience, not competing in an increasingly AI-saturated search results page. Third, community and owned channels — a newsletter and community where the relationship, not the search ranking, is the asset, because that's the layer AI-generated content can't route around. What we cut entirely: the long-tail SEO content team that used to produce 40 articles a month optimized purely for search volume with no differentiated angle. That content still ranks today; it just doesn't convert anymore because five AI-generated competitors rank next to it saying the same thing. Budget reallocation is the real signal of where growth actually believes the next decade of acquisition lives. Where's yours moving?

I report our failed experiments to the board every quarter, by name, with the numbers, on purpose. Most growth leaders I know only bring the board their wins. I think that's a mistake I used to make too. What changed my mind: two years ago I presented only wins for three consecutive quarters. The board's read on our growth function wasn't confidence — it was suspicion that we were cherry-picking, and one director asked directly what we weren't telling them. Now every board deck has a section: experiments run, win rate, and two specific failures with the actual hypothesis and why it didn't hold. Last quarter's included a paid channel test that lost us roughly $30,000 in wasted spend before we killed it at the two-week mark based on our pre-set kill criteria. What changed in the board conversation since: risk tolerance discussions got easier, not harder. The board now understands our experimentation velocity requires an expected loss budget, the same way they'd expect R&D spending to include dead ends. They approved a larger testing budget for this year specifically because they trust the reporting is honest, not curated. The governance-level honesty did something I didn't expect: it made the board partners in learning velocity instead of auditors of a highlight reel. Failed experiments reported deliberately build more long-term trust than a quarter of unbroken wins ever will.

What's one growth tactic that worked embarrassingly well for you — the scrappy thing you almost didn't try because it felt too small or too manual to matter? Mine: I personally DM'd 200 users who'd used a specific underused feature, just asking what they were trying to accomplish, no automation, no sequence, just me typing messages for an afternoon. Fourteen of those conversations turned into case study content that became our best-converting landing page copy for the next two quarters — better than anything our content team had written from a brief. It worked embarrassingly well because it didn't scale, and I almost didn't do it for that exact reason — everything in growth is supposed to be about what scales. But the insight density from 200 real conversations in an afternoon beat months of survey data we'd been sitting on. I'll admit the part that's actually embarrassing: I resisted doing this for two quarters because manual outreach felt beneath a "growth strategy," like something a scrappier, earlier-stage company should be doing, not us. That resistance cost us two quarters of a tactic that ended up outperforming most of our paid experiments that year. What's yours — the thing that worked too well for how little effort or sophistication it took?