.jpg)
Better Insights Start With Curiosity
- Lisa Ciancarelli
- 7 days ago
- 11 min read
How rank, trend, profile, context and curiosity uncover more from your data
Samantha, the Tent That Became a Blanket
The plan was straightforward: sew a proper pet tent for my 15-year-old cat, Samantha.
I had the IKEA-inspired pattern (for what that's worth!), the fleece, and a clear picture in my head of a cozy hideaway where she could curl up and disappear. I cut the pieces, inferred the instructions as best I could, and somewhere between the seams and the structure, the tent refused to become a tent.
It sagged. It slumped. It looked nothing like the shape I'd imagined. For a while, I wrote it off as a failed effort. I put the pieces away because they didn't match the original plan, and the fabric sat in my sewing bag for months, waiting for a day when I had more patience or a better vision.
Then, one day I pulled the leftover fleece out and, instead of wrestling with the tent again, I just laid it flat on the couch.
Shortly after, Samantha climbed on, settled her paws, and claimed it completely, not as a tent, but as a blanket. The thing I built to serve one purpose ended up serving a better one, simply because I stopped insisting on what I'd originally imagined.
That's the same permission I want to give you with your data today.
After 30 years working with numbers, campaigns, and an assortment of clients, I've come to believe some of the best insights aren't the ones you originally set out to find. They're the ones you stumble into when you stop forcing your first idea and give yourself the permission to see things from a different angle.
In a recent conversation, my colleague David Gustafson and I talked about how much a genuinely curious mind depends on that willingness to turn things around and look again. In analytics work, that second look isn't a luxury. It's a core part of how you move from reporting numbers to finding meaning under real-world constraints. This article is about taking that second look.
By the time you finish reading, you'll be able to:
Organize messy data quickly when you're under pressure
Use five simple criteria as angles instead of rigid steps
Spot useful, believable insights without complex tools or long hours
I'll move from the most important idea to the most specific: changing your point of view is not failure. Then we'll walk through five practical concepts you can use in your day-to-day work.
Rank and trend together
Context as your lens
Profiles that humanize the numbers
Curiosity-driven analysis
Hypotheses as starting points, not verdicts
Think of these as different ways of laying the data fabric on the couch. Sometimes what you thought should be a tent becomes a blanket. The real win is recognizing that moment and knowing what to do with it.
1. Rank and Trend, Together
Definition: Rank-and-trend analysis means looking at both where items sit in a current hierarchy and how their performance changes over time. Rank tells you who's on top right now. Trend tells you whether each item is gaining ground, holding steady, or slipping. It's really a two-part question: Who matters most today? And who's moving in a direction that should change our decisions?
Why it matters: You rarely make business decisions based only on the present or only on the future. Budgets, campaigns, and product changes depend on understanding both performance today and momentum over time.
If you only rank, you fixate on current winners and may miss rising stars or emerging problems. If you only track trend, you can get lost in every uptick and downtick without knowing which ones deserve real attention. Combining the two helps you prioritize under pressure and focus on items that are both important now and moving in a way you can act on.
A practical tactic: Build a simple two-column view for any list you care about, a spreadsheet in Excel or Google Sheets works fine.
Column A: Current rank for your key items (top 10 products, campaigns, or messages)
Column B: Trend signal over a clear time window (growing, flat, or declining versus last month or last quarter)
Then ask:
Which top-ranked items are declining and might need attention?
Which mid-ranked items are growing faster than the leaders?
Which lower-ranked items are quietly stable and supporting your baseline?
This doesn't require a complex model to build. You need the willingness to look past a single leaderboard.
A hypothetical scenario: A regional retailer runs Connected TV (CTV) campaigns across 12 publishers. At quarter's end, the team sorts performance by total attributed store visits: Publisher A is No. 1, Publisher B is No. 2, and Publisher C sits at No. 5.
Stop there, and the story is simple: A and B are the stars, C is secondary. But the team adds a trend view, comparing the last four weeks to the four weeks before. Publisher A is still No. 1, though visits have softened slightly. Publisher B is flat. Publisher C has a lower total but a steady week-over-week climb.
With rank and trend together, the recommendation shifts: protect Publisher A while running a controlled test that gives Publisher C more room to see whether that growth holds. The data didn't change. The angle did.
2. Context as Your Lens for Insights
Definition: Context is the set of conditions around a metric that shapes what it actually means: time frame, audience, creative, market environment, and business goal. A number without context is just a loose fact. A number with context becomes part of a story you can act on.
Why it matters: It's easy to drop numbers into a slide as if they explain themselves, "site visits dropped 10 percent," or "completion rate rose last week." Without context, statements like these can trigger a knee-jerk reaction. A drop might be entirely expected after a seasonal event. An increase might reflect one small audience rather than a broad win.
Context helps leaders decide whether to react, watch, or move on, and it protects your credibility. When you consistently frame numbers within their proper setting, it's harder for others to misread your work or pull conclusions the data doesn't support.
A practical tactic: Before presenting any headline number, run a quick checklist:
Time: What period is this from, and what's it being compared with?
Audience: Which segment, market, or placement does it represent?
Creative: Which message or format connects to it?
External factors: Any events, promotions, or market shifts likely affected it?
Objective: Which business goal does this metric support: awareness, consideration, conversion, retention?
Pick at least two of these to accompany any headline number. "Impressions dipped 15 percent compared with last week's sports final" lands very differently than "impressions dipped 15 percent."
A hypothetical scenario: A brand running a CTV awareness campaign sees a sharp drop in impressions and reach for one audience line, compared with the week before. Read alone, the number looks alarming, something might be broken.
Apply the context lens, though, and the picture shifts. The prior week included a marquee sports event that inflated inventory, so the current week just looks typical by comparison. This line also targets a narrower professional segment after early tests tightened it. A competitor paused a campaign, which shifted bidding slightly. And for this period, the brand's actual goal shifted from broad reach toward qualified engagement.
With that frame, the update reads differently: reach is lower than an inflated, event-driven week, and it reflects narrower targeting on purpose. Completion rates and qualified visits are stable, which matches the current goal. Nothing in the underlying data changed. The story changed once context stepped in, much like the fleece made sense as a blanket once I stopped insisting it be a tent.
3. Profiles That Humanize the Numbers
Definition: A profile is a short, human-centered description of a segment that turns aggregated data into a recognizable someone. It blends demographic clues, behavior patterns, and likely motivations into a few plain sentences, replacing abstract labels like "Segment 3" with a sense of who's actually behind the numbers.
Why it matters: Tables and charts often reduce people to variables, which might work for modeling but doesn't help a team design better experiences or campaigns. Business decisions hinge on plain questions: Who are we really reaching? What do they seem to care about? Why do they behave this way?
Profiles answer those questions in everyday language. Say "these are busy remote workers planning purchases midweek" instead of "Segment B has a higher conversion rate," and your recommendations feel grounded and easier to act on.
A practical tactic: For any segment that matters to a decision, write a short three-line profile.
Who: One line describing the person's situation or role
What: One line summarizing what they do in your data
Why: One line naming the likely motivation behind that behavior
Keep it specific and skip the jargon. For example: Who: remote professionals who watch CTV news in the early evening. What: they complete practical ads and visit stores on weekends after exposure. Why: they plan purchases during the workweek and act when they have free time. Consider these profiles alongside your charts. Over time, they become shorthand for your whole team.
A hypothetical scenario: A brand selling ergonomic office chairs sees one CTV segment, "Segment B," converting from ad exposure to online research at a notably higher rate. On its own, that's accurate but not very useful.
Dig into viewing behavior, though, and a pattern appears. This segment watches late-afternoon lifestyle and home improvement content, responds best to straightforward product demonstrations, and tends to read reviews within a couple of days of exposure before purchasing within two weeks.
The profile will reveal itself: home-based professionals trying to make their workspace more comfortable, who engage with clear demos and move quickly from exposure to research, because they're dealing with daily discomfort and want a credible, practical fix. In the next client meeting, that story lands far better than a spreadsheet cell labeled "Segment B."
4. Curiosity-Driven Analysis
Definition: Curiosity-driven analysis treats your tools, rank, trend, profiles, and context, as ways to explore your data rather than fixed steps to check off. It means giving yourself permission, even under a deadline, to ask "what else might be true here?" before locking into a single narrative. It's the analytical version of pulling the fleece out of the bag and seeing what happens when you lay it flat instead of forcing it upright.
Why it matters: Deadlines and dashboards push you toward quick certainty. You run your standard cuts, see a pattern that fits the brief, and stop. On paper, you have an answer. In practice, you might be missing the insight that actually changes a decision.
Curiosity is a small guardrail against that risk. It nudges you to question your first chart, zoom into a subset instead of the whole, swap one metric for another, and notice the stray pattern that doesn't fit your starting idea. For early-career professionals especially, you may not control the timeline or the tools, but you can control whether you give yourself one more look from a different angle.
A practical tactic: Build a curiosity block into every project. It doesn't need to be long. Ten minutes works.
Finish your main pass first, answer the brief, build the tables, draft the story. Then sit on it for 10 minutes and change one thing in your view: sort by a different column, isolate a different audience, adjust the time range, or swap the metric you're ranking on. During that window, write down at least one alternative explanation you didn't include in your first draft.
You might confirm your original story, or you might refine it. Either way, you've looked sideways at the data before presenting it.
A hypothetical scenario: A brand running a mixed-media campaign across CTV, online video, and display asks for a quick performance snapshot. The first pass shows CTV with the highest completion rates and strongest brand-lift signals, display delivering broad reach efficiently but with lower engagement, and online video landing somewhere in the middle.
The easy summary, where CTV is the hero, display supports reach, online video is fine, is tempting. Before finalizing it, a 10-minute curiosity block swaps the view to performance by age group and region. CTV turns out to be strongest specifically among older viewers in certain regions. Online video quietly outperforms for younger viewers on engagement and recall. Display has pockets of regional strength where other formats lag.
The final story gets sharper without adding hours of work: CTV leads overall, especially among older audiences in key regions; online video deserves a clearer role with younger viewers; display stays the broad-reach tool, particularly valuable where other formats underperform.
5. Hypotheses as Starting Points, Not End Games
Definition: A hypothesis is a focused, testable statement about what you expect your data to show. It frames your initial question and guides what you pull and compare. In practical analysis, a healthy hypothesis stays temporary, it's there to help you start, not to control how you finish.
Why it matters: Most projects begin with a belief, for example: younger viewers are the growth engine, retargeting drives most conversions, this creative concept is the main source of lift. Beliefs like these give you direction and help you decide what to measure first.
Trouble starts when you treat a belief like a verdict instead of an idea. Cling to the original hypothesis no matter what the numbers show, and you risk bending evidence, overlooking contradictions, and losing credibility. Treating hypotheses as starting points gives you structure and flexibility at once, you work quickly because you know what to look for, and you stay honest because you're willing to update your view.
A practical tactic: Write each hypothesis with two parts, side by side.
Expectation: What you think is happening
Permission: What you'll do if the data points somewhere else
For example: Expectation — we expect younger viewers to drive most conversions from CTV exposure. Permission — if older viewers show stronger conversion, we'll shift our focus and explore why. Use this format for your core assumptions at the start of a project, then revisit each one at the end and mark it supported, challenged, or refined.
A hypothetical scenario: A consumer electronics brand launches a CTV measurement project built on a clear internal hypothesis: younger viewers are the main engine of sales. The analyst sets up the permission clause up front, if older viewers appear more responsive, the team will adjust the narrative and dig into why.
The data comes back more nuanced than expected. Younger viewers show strong reach and healthy engagement, but only moderate conversion. Older viewers, ages 35 to 54, show somewhat lower reach but higher conversion and stronger in-store activity. Creative analysis shows why: older viewers saw messages about reliability and long-term value, while younger viewers saw aspirational content.
Instead of forcing the numbers to match the original story, the analyst leans into the permission clause: the hypothesis was challenged, younger viewers matter for reach and early engagement, but older viewers are currently driving more conversions in response to reliability-focused creative. The hypothesis didn't fail. It did its job, then stepped aside when the data pointed to a better narrative, just like the tent idea gave way to the blanket that actually worked for Samantha.
Let the Blanket Be a Win
The heart of this idea is simple: the thing you build to serve one purpose might serve a better one once you stop insisting on its original shape. In your data work, that translates into a handful of practical moves.
Use rank and trend together so you see both current leaders and meaningful movement
Wrap every key number in enough context that people understand what it really means
Turn segments into profiles so you remember the humans behind your charts
Give curiosity a small, protected window in every project to look sideways at your data
Treat hypotheses as guides, not verdicts, and show how your thinking changed as evidence came in
None of this demands complicated tools or a perfect data set. It asks for permission, permission to let your tent become a blanket when that's clearly what works.
Your next move
Don't rebuild your entire process. Pick one small shift instead.
Add a trend column to your ranked table
Write a three-line profile for a segment that keeps showing up
Block off 10 minutes for a curiosity pass before you send your next deck
Rewrite your main hypothesis with a permission clause built in
Try it next time, and see how your data starts to feel less rigid and more
responsive.
Some of the best insights you might ever find won't match the plan you had in your head. They'll show up like Samantha on the fleece, claiming a shape you didn't plan for but one that makes more sense the moment you see it. A big thanks to my friend and colleague David Gustafson for the reminder of how curiosity drives the best analytics and that granting yourself the freedom to look beyond what is immediately in front of you to satisfy your curiosity is incredibly rewarding in the insights that result!
