Paid social — week of August 16

What $915 bought in four days

Paid social influenced $14,725 of revenue and cleared $2,919 of contribution after media — $4.19 back for every dollar spent. And if you throw away every assisted order and count only the ones where social took the final click, it still cleared $1,282.

That second number is the point of this report, and it is the only thing last click is good for. It is the floor — the worst-case, deliberately unfair reading in which every ad that opened a customer's consideration gets zero credit unless it also closed them. Even there, the week is profitable at 2.4× break-even. The profit is not in question. What is in question is how much of it we are choosing to see.

Everything above the floor is the actual job. Meta is a demand-creation channel — it interrupts people who were not shopping and puts BTO in front of them. Judging it on last click is judging a channel on the one thing it is structurally worst at, and it is the reason prospecting budgets get cut in exactly the weeks they are working hardest.

Four readings of the same week

Paid social touched $14,725. It closed $8,495. The difference — $6,230 — is not a rounding artefact or a measurement error. It is the work Meta is bought to do.

Why last click understates a social channel specifically. Someone scrolling Instagram is not shopping. They see a Bell Moto-10 at 40% off, register it, and come back two days later through a Google search for the model name or by typing btosports.com. Google gets the click; social created the demand. On a last-click scorecard the ad that made the sale happen reads as a wasted impression, and the search campaign that harvested it looks like a genius.

The same logic runs the other way for Meta's own 37.1×. That figure counts anyone who merely saw the ad within a day, and measures against the pre-return value at the pixel. It is the platform marking its own homework and it should not leave the ad account. Any-click is the honest middle: an order Shopify can actually point to, where a paid-social click sits somewhere in the path.

From revenue to contribution — the working number, then the floor

Applying BTO's own measured return rates and cost structure by category, this week's mix clears 27.2%. Run it on the revenue paid social influenced and on the revenue it closed, and both land well above zero. The floor is the proof; the ceiling is the ask.

Now the same walk-down on the floor. Same cost model, same categories, same axis — only the revenue definition is tightened to orders where paid social took the final click. This is the sanity check, not the scorecard.

This is not a new margin assumption. It is the same category-level model already in the profit diagnostic — measured return rates from BTO's own Shopify data, markup held flat at 40%, with freight, processing, return handling and markdown recovery as stated estimates. Applied to this week's actual product mix rather than a catalogue average.

Returns are priced in, not ignored. The clearable rate is applied to gross revenue and already carries each category's own return rate, so the $333 of returns visible in the window is not double-counted — and the returns still to come on these four days' orders are already accounted for. Returns land weeks after the click; a four-day report showing only booked returns would flatter itself badly.

The honest sentence for the board: "Four days of paid social cleared between $1,282 and $2,919 of contribution on $915 of media. The lower figure assumes every assisted order would have happened anyway."

Where the influenced revenue came from — and what each category clears

Helmets and boots are 68% of the week's influenced revenue — and they are the two categories that clear the least, because they carry the highest return rates in the catalogue.

This is the whole markup-versus-contribution argument showing up in a single week. Helmets influenced $6,957 of gross — half the week — at a 25.6% clearable rate and a 3.91 break-even ROAS. Goggles, gloves and consumables clear 28–32% and break even below 3.30. The blended break-even lands at 3.68 precisely because the mix is helmet-heavy.

It is not a reason to sell fewer helmets this week. They cleared comfortably. It is the reason the same headline ROAS means different things in different weeks, and the reason a mix shift toward goggles, apparel and consumables would raise contribution without raising spend.

Note where the assist layer is heaviest. Pants went from $794 closed to $1,767 influenced and jerseys from $324 to $641 — apparel roughly doubles once assists are counted, because it is what goes in the basket after social put the helmet in someone's head. On a last-click scorecard, apparel looks like a category paid social barely touches.

Spend by creative type against what it returned

UGC took 22% of the budget and produced 30% of Meta-attributed value. The catalogue feeds took 43% of spend and did the volume. Neither is redundant — and UGC is the half that creates demand rather than harvesting it.

UGC is the efficiency lever; the catalogue is the volume engine. ASC_Gear_Catalog alone took a third of the entire budget ($304) and drove 41 purchases at a $7.42 cost per purchase — that ad is the account's workhorse and it is not close. But per dollar, UGC returned roughly 1.7× the catalogue's rate, and it is the only creative type that can be manufactured on demand. A catalogue feed can only sell to someone already in market; a UGC video puts someone in market.

The video read supports it. UGC_Flash_Sale_Bell_Moto10_40%_FH held 24.3% three-second play rate and carried 282 viewers to 95% — 12.6% of everyone who started it finished it. Sale_UGC_Youth_Bell_Moto9_FH hooked 33.3%. These are the assets to reproduce.

Every ad over $3, calibrated to the Shopify basis

Meta's ROAS is scaled down by the account's own 0.43 calibration factor — the ratio of influenced Shopify revenue to Meta-reported value — and set against each category's real break-even. Six ads, $144 of spend, are underwater even on the generous reading.

These six fail on every measure — floor, influenced and platform alike, which is what makes them safe to act on without an attribution argument. $143.64 — 15.7% of the week's budget — produced about $58 of contribution against $144 of cost. A net drag of roughly $86 over four days, about $7,800 a year if left running. Redeployed at the account's own average it would have produced closer to $345, so the real cost of leaving it alone is the $287 gap, not the $86 loss.

The biggest single offender is Sidi_X-Power_Boots_50%_v1 at $93.56 — the second-largest spend in the account. It generated 13,131 impressions and 319 link clicks but only three purchases, at a 2.43% CTR and a 0.94% click-to-purchase rate against the account's 2.64%. The creative is getting clicks; the page is not closing them. Worth checking size availability on that SKU before killing the ad — it is the exact signature of the depleted-size problem already documented in the Channable work.

Ten products, 57% of the week's influenced revenue

Ranked by revenue paid social influenced. Two of the top ten got zero last-click credit — on a strict scorecard they are products social apparently had nothing to do with.

Bell Moto-10 Fasthouse Raven leads at $2,821 influenced against $1,166 closed — the single most assisted product of the week, and the subject of the account's best-performing video. Alpinestars Tech 7 Boots ($530) and the Fox Proframe MTB helmet ($428) sold with a paid-social click in the path and zero last-click credit — $958 of revenue a last-click report files under "not social."

Order Protection was attached 28 times across influenced orders — the only order-count signal in this dataset. It is opt-in, so it sets a floor rather than a count: at least 28 distinct orders had a paid-social click somewhere in the path.

The case for more budget — and the honest caveats on it

Efficiency can degrade 76% on the working number, or 58% on the floor, before this week's spend stops clearing. That is the room available, and it argues for stepping up — carefully.

• Change what we report on, permanently. Any-click becomes the working number for paid social; last click travels beside it as the floor, labelled as such. Reporting the floor as the headline has been costing this channel budget in the weeks it was working hardest — $1,636 of contribution was invisible this week alone. One column added to the weekly report ends the argument for good.
• Cut the $144 first. Six ads are underwater on every measure. Pausing them costs nothing and frees 16% of the budget immediately — before asking for a dollar more.
• Then step spend up 40–50%, not double. Current run rate is roughly $230–260/day, about $7,000–7,900 a month. Paid social ran $29,129 in December — so this is a fraction of proven capacity and the audience is nowhere near saturated at a 1.0–2.2 frequency. Standard practice is 40–50% increments with a fortnight per learning cycle, watching cost per purchase rather than ROAS, because CPA moves first.
• Put the increment behind UGC and ASC, not the catalogue. The catalogue feeds are already scaled and will absorb more budget at declining efficiency. UGC returned about 1.7× per dollar and there are only five such assets live. ASC is the campaign generating the assist layer. The constraint on paid social right now is creative volume, not audience.
• Reactivate the inactive winner. Holiday_UGC_Bell_Moto9_WIOWIG_Video_HAS ran at an 84.9 Meta ROAS on $10.38 — a 36.9× calibrated return — and is switched off. It is a trivial amount of money to test at a proper budget.
• Watch the mix, not just the number. Break-even this week was 3.68 because helmets were half the revenue. A week that skews further into helmets needs a higher ROAS to clear the same contribution — and a week that pulls goggles, apparel and consumables into the basket clears more on identical spend.
• Do not extrapolate this week to the quarter. Four days, 125 influenced items, an active flash sale, and August is not December. The stable forecast inputs are the contribution rate and the break-even. The 16.1× is not one of them.