An E-commerce brand doing at least 6 figures in annual sales collects, on average, between 5 and 50 TB of customer data each year. That’s 1st and 3rd party data: transactions, preferences, keywords, activity, audience targeting, on and on.
We’re drowning in that data. So why does it feel harder to just launch something?
85% of leaders feel their teams suffer from "decision-making distress" and overwhelm
and yet
86% of marketers say their data makes them feel less confident
What are we doing? Why are we gathering this data if it doesn’t inform decisions?
The ethics of user data collection, IP rights, LLM copyright infractions is a hot topic. So what right do we have to all this data, and do we even need more to make better business decisions?
No, friend. We don’t. Most marketers can’t even handle the data we already have.
Research suggests technology marketing teams only use a small percentage of the data they collect. Forrester revealed that 60% to 73% of all data collected by companies goes unused for analytics.
Like I tell my kids at supper: enjoy what’s already on your plate before you ask for more. It’s not fair and it’s expensive. (Also I want to eat the scraps after you go to bed.)
Information overwhelm creates analysis paralysis—overthinking available data so much that progress comes to a standstill. In a startup or enterprise team it looks exactly the same:
project status updates with no new real information
another pivot, or plan to pivot
standups feel repetitive and pointless rather than helpful
bringing smaller and smaller decisions to the founder or manager
adding more stakeholders after a project kick-off
‘looping in’ an exec across the org for ‘another POV’
when presenting: nerves you can see in body language and hear in their voice
capable teams will feel agitated and restless when milestones miss
burnout begins for disgruntled team members and can lead to departures
Paralysis doesn’t show a lack of effort. On the contrary, it’s usually a result of trying to get things right. With so much data available, people want an impossible level of certainty before acting.
Idleness is worse than imperfection. Waiting for 'enough' information leads to missed opportunity, stagnating growth, and frustrated teams. Funny enough, most perfectionists describe those three outcomes as their nightmare scenario. Ironic.
In this post I’ll show you how to break free of decision paralysis to get back to growing the business:
3 things you can say in meetings to reframe the problem
3 metrics that clarify activities for everyone
3 mindset changes you can model for your team today
3 experiments that produce conclusive evidence in less than a month
3 watchouts when acting on AI, 1st-party data and customer comments
But first — let’s burst the hypocrisy bubble. Marketers can collect data and be ethical using it.
Speaking of my kids, a paradox: I believe in parental-controlled, protected online privacy and I work everyday to provide seamless personalized shopping experiences for consumers. Huh? Hypocrite much? Like an eco-conscious lumberjack.
I used to work at a large enterprise software company and one of the chairmen of the board often opened meetings by reminding us that “intelligence is the ability to hold two truths at the same time”. I used to make a face when he said that but damn, I get it now.
His point is that we can’t wait for perfection; for one option to completely disappear so that we can dump all our energy into the other. There are always alternatives.
We have to choose. Choosing is an act.
Marketers have an almost limitless set of data options in this intermission we find ourselves in between the ‘big data’ track-everything era and this new the-LLM-knows-all one.
I want the ability to shield my kids from dangerous content online and I want to use data to reduce my clients’ CAC when they advertise. So,
I carefully select my cookie and app permissions on and off
I monitor my ad preferences and routinely clear caches
I decide which data layers will inform my ad sets and which I omit
I iterate on what’s working and drop what isn’t until CAC reaches growth levels
It was hard at first, but I made choices on what mattered most.
You can swim out of the data overwhelm too if you follow decision principles.
Asking your team “what information can help us reduce your uncertainty, right now?” is a bit blunt. These classic Design Thinking exercises can focus creativity into bite-size solutions:
"How might we..." is a great starting prompt for your teams to brainstorm with focus. It puts the problem in view and turns on solution-mode.
ie. "We need to limit overspending. How might we target the right audience for this campaign?"
"What needs to be true..." is a prompt to identify what we call tangential risks. These are risks that when isolated, become less scary and easier to resolve or avoid.
ie. "We need to standardize our CRM contact info to automate marketing. What needs to be true to validate emails without tripping global privacy regulations?”
"We can't go back..." is a way of understanding which parts of a decision are reversible and which are irreversible.
"To personalize our experiences we need to link CDP platforms. We can't go back once we buy an annual plan on Klaviyo - that decision has irreversible costs. Are there options that allow us to trial, or pay monthly?"
Isolate the elements of your decision holding you back and you may find it's much easier to make incremental progress on your big call.
Whatever you call it: North Star Metrics, OMM "one metric that matters", growth loop metric.
Teams should definitely ladder up to the same org-wide metric but it's paralyzing when you can't see the connection to your work. Teams (esp. founders) need to be precise and prescriptive when defining strategic growth metrics. Choosing is an act but a new metric isn't a fix onto itself. It's just another form of communicating strategy.
Additionally: these metrics will evolve. "Nightly bookings" at AirBnB was helpful during the growth stage because every team member could see their work moving the needle. As AirBnB scaled they adopted product and growth specific metrics so teams could focus their efforts on specific experiences.
The 3-goals-deep model is a great way to frame the metrics in your org for clarity and action.
Strategic metric - this is the team-rallying, deliberate one that drives growth. It should have several components: a quantifiable event, a defined “by who” customer segment, and a timeframe. ie. New users activating paid accounts within the last 60 days.
Input metric - for marketers this will likely be one of your performance measures like clickthrough rate, signup rate, retention etc. This will ladder into the strategic metric.
Decision metric - this project-specific measure is what decides the outcome for any experiment. For speed’s sake it must be adjacent to a change. This is what experimenters call the 'split'. It's often an action you're attempting to influence. ie. Rate of users who take the upgrade on the landing page.
These metrics ladder up into each other as performance gets tied to strategy. See how you can easily spot the chain from “my work” to “our goal”?
The way you talk to your team about decisions, risk and uncertainty will shape their own framing of decisions. Consider how you show up in meetings and one-on-ones with your decision-makers and try to role-model decisions the way you want them made. Wear your heart and brain on your sleeve. Ew. You get it though. Be repetitive.
Good data can reduce your uncertainty but won’t create a totally risk-free circumstance. No dataset or AI prompt can predict outcomes and no experiment can promise to stay within range of the mean. You’ll be surprised sometimes, so prepare how you’ll react before decisions are needed.
Quick story: my own business, Formentor Labs, is named for an accomplishment I'm very proud of. I love driving tricky roads especially in Europe, and this one in particular was one of the scariest. The cliff edges here were just immediate and the switchback roads had broken railings whenever the overgrown tree roots would cross it. There were no straights. It was turn after turn after turn. Up and down, but up more than down. I couldn't tell you at the bottom how I'd get to the top for the beautiful lighthouse view of the Mediterranean. I did know that rounding the next corner carefully, applying my handbrake on the slopes and anticipating oncoming traffic were the best ways I could mitigate danger while making progress.
A turn at a time. I put it into drive.
At no point were we danger-free. For real. With the right focus I was confident enough to tackle each new curve of the climb. If I ever hit a tricky spot I’d be ready.
Doug Hubbard inspired many of the thoughts in this post and none more than this one. You have more data than you think, and you need less data than you think.
On having more data:
If you’ve been in business more than a year then you have historical data and you have YoY improvement data.
If you sell in multiple channels then you have messaging that works elsewhere.
If you have competitors then you know what their customers preferred over yours, and vice versa.
If you sell in an established market then you have industry averages and benchmarks.
On needing less data:
Will you not launch if you haven’t reached a benchmark, or are you willing to launch and grow?
Will you never touch this campaign again or can price, message, features be edited later?
Are you doing something brand-new? “If you know almost nothing, then anything will tell you something”.
This always cheers my team up. When camping outdoors, remember the concept of “you don’t have to outrun the bear”. You don’t need to outrun the bear, you only need to outrun the other campers. 🐻
In business, you're rarely going to launch something in its final form. You shouldn't. MVPs, continuous improvement, Lean Startup, etc. You know the drill.
When deciding on whether you're ready to launch just consider what you're truly up against:
Are you creating something brand new? You're up against status quo and opportunity cost of doing something else. That has value, but it hardly requires perfection.
Are you up against a paid alternative like an existing campaign or channel? Then beat that. Evaluate over time and by segments.
Are you up against a competitor? Fun. Beat their traffic, their SERP position, their social sentiment, their audience size, their reviews, their share of market.
You aren't designing the perfect campaign or product feature -- you're setting a new high score until you can do it again.
One of the most straightforward ways to eliminate analysis paralysis is to run small, meaningful experiments that demonstrate evidence for your hypothesis. Tests allow you to quickly gather real-world feedback to refine your plan as you build.
Okay, say you want to launch a competitive offer. Instead of building everything all at once, iterate only the things you need to learn. An example experiment roadmap:
Test 10 ad units with a variety of value prop and positioning statements. Send to a "coming soon" landing page. Use the 3 winners to inform your landing page copy.
Create three landing pages around those value props. Interview customers for feedback to polish each offering. Launch these in a split test (A/B/n test) and keep the winner.
With the strongest converting landing page identified, now start to test pricing. You know what customers want, so what will they pay for it? Begin the willingness-to-pay framework and if you have the traffic, a Van Westondorp study. You’ll finish with a ‘range of appropriate prices’ you can count on.
These experiments don’t require perfect data to run. Instead, their results become new evidence that reduces uncertainty and informs your next steps.
Going back to how I opened this post, we need to be choiceful in using data to inform our marketing decisions. Not all data is the same. I talk a lot about “strength of evidence” and the duality between quant and qual data. They act together. Trust your data by order of proximity (levels of ownership):
Customer interviews - sitting down with customers and working through focused topics is the best way to uncover issues in your product or service, full stop. It’s an art, though, so use experts who know how to step around bias, baiting and assumption.
1st party data and in-market experiments - owning your own data, with your own definitions and configuration is stronger than trusting the reports and research from a 3rd party. Interpreting that data is an art, though, so make sure you’re asking experts to parse your databases for insights rather than declaring up is up.
LLMs and AI - this data is not yours. Unless you created the corpus yourself (some AIs offer this, like NotebookLM) then you don’t have clarity on the basis for the response. Be very, very careful acting on direction from a prompted bot without verifying it first. Even still, the “synthetic audience” approach of creating your own LLM for singular use rubs me the wrong way - like taking the humanity out of your doctor and then trusting her diagnosis. Are you… sure?
Analysis paralysis is easily one of the most common pitfalls for teams trying to create something new. You can get unstuck—by reducing uncertainty, starting with what you have, and prioritizing actionable decisions, you can turn data into a competitive advantage.
Master the art of running marketing experiments and making evidence-driven decisions by catching this Substack for frameworks, tips, and expert strategies to grow your business confidently.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.