Marketing Analytics with Fexingo · 2026-07-02 · 7 min
Key moments - from our scoring
Substance score
71 / 100
Five dimensions, 20 points each
Marketing attribution models universally struggle with seasonal businesses because they treat every touchpoint identically regardless of timing or purchase cycle phase. Lucas shares a case study from Trailblazer Gear, an outdoor apparel brand, where standard attribution credited paid social with 40% of Q4 sales - but customer surveys revealed over half started research on Google in late summer, months outside the typical attribution window. Last-click and time-decay models simply dropped these early organic signals as noise. The solution involves segmenting attribution by product category seasonality, applying a seasonal baseline lift (1.5x multiplier for August organic searches that converted in November), extending lookback windows to 60-90 days during research phases, and validating with holdout tests. Trailblazer shifted 15% of Q4 paid budget into Q3 content marketing and SEO, actually improving overall ROI by feeding demand when intent was genuine. This approach applies directly to any seasonal vertical - winter fashion, summer gear, holiday retail - where research and purchase phases are separated by months.
Standard attribution models use a fixed 30-day lookback window and credit the last click, so organic search research from August falls outside the window and gets dropped entirely. The model only sees the final paid ad in December, creating a seasonality blind spot that misattributes credit to last-touch channels.
Segment your data by product category and season, then apply a seasonal coefficient (e.g., 1.5x multiplier) to early-stage touches based on historical correlation between those touches and eventual purchases. Extend your lookback window during research phases (60-90 days) and shorten it during peak season (30 days).
They moved 15% of Q4 paid social budget into Q3 content marketing and SEO, including blog posts and gear guides. This generated better overall ROI by feeding the top of the funnel during the research season when customers were actively seeking information.
A holdout test paused paid social for a segment of users in November, and the organic-only group converted at nearly the same rate, proving that early content nurture had already done the work before the paid ads ran.
Use product-level seasonality with separate attribution models per category, since winter gear and beach gear have completely different high and low seasons and research-to-purchase timelines.
Our reviewer’s read on each dimension, with quotes from the episode.
The episode delivers a concrete, non-obvious problem (seasonality blindness in attribution models) and walks through a specific fix (segmentation by season, seasonal multipliers, extended lookback windows). Most content is substantive rather than filler, though the conversation could have gone deeper into mathematical validation or edge cases. The core insight - that attribution models ignore calendar-based demand shifts - is genuinely useful for practitioners.
for seasonal purchases, that early research click is often the most important signal
organic search clicks in August got a 1.5 multiplier in the Q4 attribution window
The core argument - that seasonality breaks standard attribution - is relatively fresh and counterintuitive enough to surprise most practitioners who use off-the-shelf models. However, the fix itself (segmentation, weighting adjustments, longer lookback windows) is somewhat incremental rather than truly novel. The framing is stronger than the methodology.
attribution model sees the final sprint, not the start of the journey
standard attribution assumes your customer's purchase cycle is flat all year. But for seasonal businesses, it's anything but
Lucas appears to be a practitioner who has actually run attribution modeling work at scale (Trailblazer Gear case study), which is valuable. However, no clear credentials, company affiliation, or evidence of sustained expertise at a top-tier organization are provided. Luna functions as a competent interviewer rather than a peer expert. The guest is credible but not elite-caliber.
I worked with a mid-size outdoor brand - let's call them Trailblazer Gear, not their real name
We started manually, but later built a simple script in Python that reads the calendar week and applies the coefficient from a lookup table
The episode includes a named case study (Trailblazer Gear, anonymized), specific metrics (1.5 multiplier, 35% organic revenue lift, 20% paid social drop, 15% budget shift, 60-day vs. 30-day lookback windows), and a validation method (holdout test). The evidence is sufficiently concrete to allow a listener to replicate the approach, though some numbers (like the multiplier derivation) lack deeper granularity.
organic search's attributed revenue jumped by about 35 percent in Q4. Paid social dropped by nearly 20 percent
we extended the lookback window to sixty days in Q3 and kept it at thirty in Q4
Luna asks solid follow-up questions that push for implementation details and evidence ("Did it change the channel mix?", "Any pushback from the team?", "did you automate the seasonal coefficient?"). However, the conversation rarely challenges Lucas's assumptions or explores limitations - there's no pushback on statistical validity, sample size, or whether the multiplier approach could mask other confounds. The dialog is collaborative but not adversarial enough to expose soft spots.
And I'm guessing that changed budget recommendations?
That's the kind of evidence that changes minds.
Computed from the transcript - who did the talking, and the words that came up most.
Episode 88 of Marketing Analytics with Fexingo: Lucas and Luna dig into how standard attribution models flatten seasonal demand shifts, using data from a mid-size outdoor apparel brand. They walk through a real example where fourth-quarter attribution underplayed organic search while overvaluing paid social - not because the models were broken, but because seasonality wasn't factored into the window. Lucas explains why time decay models treat a November click and a February click identically, and how adding a seasonality coefficient changes the ROI picture. Luna brings up the concept of 'seasonal baseline lift' and why brands with predictable buying cycles need to adjust their lookback windows by quarter. The episode ends with a practical framework: run separate attribution models for high-season and low-season periods, then compare the channel mix. No hot takes - just a specific fix for a blind spot that affects every retailer on a calendar.
Transcribed and scored by The B2B Podcast Index.
Lucas: So you run an attribution model for an outdoor apparel brand. And every November, the model tells you paid social drove 40 percent of sales. But when you actually dig into the data, most of those customers first found you through organic search in August, when they were researching gear for fall hiking trips. Luna: Right - the model credits the last click, which is the December paid ad, but the real influence happened months earlier.
That's a seasonality blind spot. Lucas: Exactly. And it's not just last-click. Multi-touch, time decay, even algorithmic models treat every click the same regardless of when it happens in the calendar year.
They don't understand that a click in January has a different intent than a click in October. Luna: So the seasonal pattern is invisible to the attribution engine. That means you might overinvest in paid social in peak season and underinvest in organic content that builds demand earlier. Lucas: Precisely.
I worked with a mid-size outdoor brand - let's call them Trailblazer Gear, not their real name - and their attribution model consistently showed paid social as the top channel in Q4. But when we looked at customer surveys, over half the buyers said they started their research on Google, usually in late summer. Luna: So the organic search clicks from August - those are outside the standard thirty-day attribution window. They just get dropped.
Lucas: Gone. Time decay models put most weight on the last few touches, so a click from three months ago has near zero influence. But for seasonal purchases, that early research click is often the most important signal. Luna: It's like the classic 'traffic on the last mile' problem.
The attribution model sees the final sprint, not the start of the journey. Lucas: Right. So I want to walk through what we did to fix it. It's not about throwing out attribution - it's about layering seasonality on top.
Luna: Good. What was the first step? Lucas: First, we segmented the data by 'high season' and 'low season' for each product category. For Trailblazer Gear, winter jackets have a high season from October to January, but hiking boots peak in spring.
So we ran separate attribution models for each period. Luna: So you're not using one global model. You're splitting the calendar into windows that match the buying cycle. Lucas: Exactly.
Then we applied a 'seasonal baseline lift' - a percentage increase to the weight of any touchpoint that occurred during the research phase of the season. For example, organic search clicks in August got a 1.5 multiplier in the Q4 attribution window. Luna: So you're saying, 'Hey model, this click is more important than usual because it's the precursor to the seasonal peak.'
That makes sense. Lucas: Right. The multiplier was based on historical correlation - we looked at how often an August organic search was followed by a purchase in November, versus a click in November itself. The ratio gave us the coefficient.
Luna: Did it change the channel mix significantly? Lucas: Dramatically. In the adjusted model, organic search's attributed revenue jumped by about 35 percent in Q4. Paid social dropped by nearly 20 percent.
The total pie didn't change - we just redistributed credit to the channel that actually did the heavy lifting. Luna: And I'm guessing that changed budget recommendations? Lucas: It did. The brand shifted about 15 percent of its Q4 paid social budget into content marketing and SEO in Q3 - blog posts about winter hiking tips, gear comparison guides.
They actually got better overall ROI because they were feeding the top of the funnel when it mattered. Luna: So the key takeaway is: standard attribution assumes your customer's purchase cycle is flat all year. But for seasonal businesses, it's anything but. Lucas: Right.
And the fix isn't complicated - you just need to align your model with reality. Run separate models per season, apply a seasonality coefficient to early-stage touches, and re-validate with actual survey data or incrementality tests. Luna: One thing I'd add: lookback windows should also vary by season. A thirty-day window might be fine for fast-moving consumer goods, but for outdoor gear, you might need ninety days in the pre-season.
Lucas: Hundred percent. In fact, for Trailblazer Gear, we extended the lookback window to sixty days in Q3 and kept it at thirty in Q4. That alone captured more early organic traffic. Luna: Any pushback from the team?
Lucas: Yeah, the paid social manager was skeptical - her channel looked less effective. But we ran a holdout test: paused paid social for a subset of users in November. The organic-only group converted at nearly the same rate, because they'd already been nurtured by content. Luna: That's the kind of evidence that changes minds.
So for listeners who do this themselves, start by pulling your first-touch data and see if it correlates with seasonal trends. Lucas: Great advice. And if today's conversation felt useful - the kind of thing you might actually apply - these conversations stay ad-free because of listener support. If it's helped you, buy me a coffee dot com slash fexingo.
That's buy me a coffee dot com slash fexingo. Luna: Yeah, it really does keep the show independent. So back to the model - one more thing on implementation: did you automate the seasonal coefficient or just apply it manually? Lucas: We started manually, but later built a simple script in Python that reads the calendar week and applies the coefficient from a lookup table.
It's not perfect, but it's better than ignoring seasonality entirely. Luna: And what about brands with multiple seasons - like a retailer that sells both winter gear and beach gear? Lucas: Then you need product-level seasonality. For each category, define its own high and low periods.
The attribution model should be run per category, not as a single blended number. Luna: So the next time you see attribution data that seems off - like paid social suddenly crushing it - ask yourself: what season is it really? Lucas: Exactly. The model doesn't know the calendar.
But you do.
Other episodes covering the same guests and topics, from across The B2B Podcast Index.
How Idempotency-Key Design Prevents Payment DisastersThe Developer Tools Podcast with Fexingo · features Luna98 / 100
How Kubernetes Topology Spread Constraints Create Scheduling HotspotsDevOps Daily with Fexingo · features Luna95 / 100
Why Pipeline Velocity Trumps Deal Size Every TimeThe Growth Operator with Fexingo · features Luna95 / 100
Why Enterprise Software Deals Now Include a Vendor AI Model Explainability MandateB2B SaaS Talks with Fexingo · features Luna94 / 100
How B2B Brands Wreck Pipeline with Unsyncroned CRM DataThe Marketing Operator Podcast with Fexingo · features Luna92 / 100
How to Sell Against a Competitor Already in the BuildingSales Leadership with Fexingo · features Luna85 / 100
How Incrementality Reveals True Marketing Impact90 / 100How Marketing Attribution Fails on Subscription Models82 / 100Why Marketing Attribution Breaks on Subscription Models92 / 100Why Marketing Mix Models Beat Attribution for Long Sales Cycles92 / 100How Unified ID Replaces Broken Third-Party Cookie Attribution85 / 100