Data period: weeks 39/2025 (network entry) - 36/2026. Campaigns: weeks 15-36/2026.
Growth figures in the tiles compare the campaign-weeks average with the pre-campaign average, unless stated otherwise.
Rotation at DIA
+119%▲
from 0.79 (weeks 40/2025-14/2026) to 1.72 (campaign weeks 15-36/2026) units per store weekly, promotions excluded. Growth started before the flights; family share (×2.4) is shown separately in the shelf section
Brand searches
×3,8
peak of 247 monthly searches on Google (Spain) in July 2026 vs an average of 64 in the winter months before the campaigns (December-February)
Paid reach on IG
8,2 mln
total Instagram campaign reach per Meta (April-September 2026). Separately: four posts exceeded 1.1 M views each
Social investment
38,0 k €
Meta 28.9 and TikTok 9.2 k € (38.0 total after rounding) in weeks 14-36/2026; full accounts: 40.7 k €
Price test
0,91↔1,35 €
four levels, not a switch: 1.35 → 0.91 → 1.19-1.35 (sales 1.22) → 0.91 again (1.95, about +60%). A signal, not proof
From when to when: period comparison
There is no year-over-year comparison yet - the product entered DIA at the end of September 2025,
and the first full year closes in autumn 2026. Hence two cuts: before / during the campaigns, and by quarter.
Period
Avg rotation
Brand searches (monthly)
Avg price
Investment
Before the campaigns (wk 40/2025-14/2026, 28 wks)
0.79
77
€1.12
€0
During the campaigns (wk 15-36/2026, 22 wks)
1.72 (+119%)
172 (×2.2)
€1.02
€38.0k
Quarter
Avg rotation
Searches (monthly)
Avg price
Investment
Q4 2025 (wk 40-52)
0.63
78
€1.35
€0
Q1 2026 (wk 1-13)
0.90
69
€0.91
€0
Q2 2026 (wk 14-26)
1.50
183
€1.11
€21.7k
Q3 2026 (wk 27-36, 10 wks)
1.95
157
€0.91
€16.3k
Reading the sequence: the price cut from 1.35 to €0.91 (January) lifted rotation before any campaigns
(0.63 → 0.90); the campaigns (from April) coincided with further growth to 1.50 and 1.95, with a price rise
along the way in Q2 (average €1.11). Untangling price and campaigns - see the Price section and the closing conclusions.
What happened week by week
Four charts on a shared timeline: investment, reach, searches and sales.
Click the chapters below (arrow keys work too) - the cursor shows values.
Fig. 1. Weeks 39/2025 - 36/2026. Sales = units per store weekly, promotions excluded.
The two gaps in the sales line are the launch week and one promo week (details in the methodology).
Price (bottom panel) = implied price: weekly sales value in € divided by units sold,
across all stores in the data - a network-wide average, not the shelf price of a single store.
Where the money works: platforms
Meta campaign objectives vs Puro's share of the water family
Budget split by campaign objective (stacked, amber shades) and Puro's share
of total AGUA MINERAL family rotation, week by week. Grey bands: weeks with an active campaign.
Fig. 2. Tygodnie 39/2025 - 36/2026 (stack: Meta spend in the window, 28.9 k €).
Share = Puro rotation / total rotation of AGUA MINERAL family items, regular sales, SMD3 stores;
0.075% is a sliver of a very large all-waters family. On week-over-week changes the spend correlation
is weak (at most 0.16 for combined Meta+TikTok investment). This may mean campaigns build the level rather than spikes -
or that no weekly effect is simply visible in this data.
Rotation vs price, budget and searches
Pick one of the three variables and move the slider to a level.
On the right you'll see the average rotation in the weeks when that variable had that value.
How to read this: price, campaigns and season often changed at the same time.
So the result says "this is how much sold WHEN it was set like this", not "this is how much will sell IF we set it like this".
Treat levels based on 2-4 weeks as indicative only - the week count is always shown next to the result.
Top-reach posts
Best posts from @puro_exorcismo and collaborations, metrics for January-July 2026.
Clicking "Instagram preview" plays the video in place (when the file is opened locally or on hosting).
In the claude.ai preview embeds are blocked, so the click opens the post in a new tab.
Puro on the DIA shelf
The same data source we use to analyse the whole chain, narrowed to the brand context.
Hypothesis
-29% vs +31%
Growth against the season
Sparkling waters lose about 29% of sales from summer to winter. Puro grew about 31% that same winter
(autumn to winter; exact values depend on window definitions). The product has less than a full year of history,
so season cannot be fully separated from campaigns and rollout.
Observation from data
share ×2.4
Faster than the family
Puro's share of total AGUA MINERAL family rotation grew from 0.031% to 0.075% in under a year
from launch (48 weeks; series in the chart below). Still a small scale, but the direction is unambiguous.
Observation from data
4,0 vs 1,7
Gap to the functional-shelf leader
The best-rotating kombucha SKU (Komvida ginger-lemon) does ~4.0 units per store per week
vs ~1.7 for Puro (averages for weeks 15-36/2026). There is still headroom above.
Hypothesis
-2.0 per wk
Reshuffle in the 500 ml format
As Puro grows, its twin format loses: Font Vella sparkling 500 ml is down 2.0 units per week.
The Font Vella brand as a whole grows (+13.5%), so this is a hypothesis of taking one format's sales, not proof.
Investment split by campaign
Fig. 3. Campaign spend, full Meta export June 2025 - September 2026:
30.7 k € in total (Meta only), of which 28.9 k € falls in the chart window; TikTok is shown in Fig. 1 and the table.
What is fact, what is hypothesis
Facts from data
Rotation, spend, reach and search series: four independent sources
Growth started before the campaigns: the 0.53 low in week 52/2025
AGUA MINERAL family share ×2.4 in under a year: from 0.031% to 0.075%
The top kombucha SKU rotates ~2.3× faster than Puro (4.0 vs 1.7)
Hypotheses
Campaign contribution to growth (needs a control group)
Taking sales from Font Vella sparkling 500 ml (the brand overall grows)
Growth against the season (first year of the product, no summer 2025 baseline)
Price sensitivity: rotation 1.22 in the expensive window vs 1.95 after the cut (trend and season in the background)
To verify
Confirmation of the store count carrying the product (monthly file: 396→613 through July 2026)
Unit margin: cost per incremental unit and return on investment
What this picture does not prove
1
This is a consistent sequence, not hard attribution.
The campaigns had reach objectives, not sales ones - there is no measurable conversion path per platform.
2
The price changed in parallel with the campaigns.
1.35 → 0.91 → 1.19-1.35 → €0.91: rotation slowed in the expensive window (wk 16-22),
and the final stretch of growth was helped by the lower price; distribution grew in parallel.
3
No weekly effect is visible in the numbers.
Week-over-week changes in spend and rotation correlate weakly (at most 0.16 for combined Meta+TikTok investment) -
that fits reach objectives, but could equally mean no measurable effect.
Ahrefs data for August-September may be incomplete.
4
How much of this did the campaigns do? That cannot be settled with this data.
The cheapest possible step: a desk comparison against similar new products without campaigns
in DIA's own data - no in-store tests of any kind.
Unit conversions assuming 600 stores
One shared, explicit scale assumption: 600 stores. The actual count per DIA's monthly file
grew from 396 to 613 (through July 2026) and needs confirmation from the data owner.
These are conversions of observations, not forecasts.
Calculation
≈1 170 units/wk
Rotation 1.95 (wk 24-36/2026, price €0.91)
1.95 per store × 600 stores - an upper estimate. The level observed at the lower price during the campaigns.
Calculation
≈730 units/wk
Rotation 1.22 (wk 16-22/2026, price €1.19-1.35)
1.22 × 600. The difference between periods does not measure a clean price effect -
campaigns and season changed in parallel. Six weeks of observation (excluding promo week 17).
Reference point
≈470 units/wk
Rotation 0.79 (pre-campaign average)
0.79 × 600, the baseline from the +119% metric. A historical reference point at a different distribution -
not an estimate of no-campaign sales.
From DIA's monthly file
≈25,9 k €
Shelf sales value (24,627 units, Sep'25-Jul'26)
Monthly units from the file times the average shelf price; August-September 2026 missing.
This is store turnover, not brand revenue and not ROI: 38-40.7 k € is media spend, 25.9 k € is shelf value - without margin we do not divide one by the other. File to be confirmed.
One number is missing to convert units into euros: the unit margin.
Five takeaways in 60 seconds
Thesis: per-store sales doubled because three forces combined - the €0.91 price at the shelf,
campaigns building the brand above the shelf, and growing distribution. The exact contribution of the campaigns
to units cannot be settled with this data. Below are the proofs; each links to its section.
1
The result: per-store sales more than doubled.
From 0.79 to 1.72 units per store weekly (+119%), from a 0.53 low in December.
Rotation was rising before the campaigns started - we say this openly, because that growth lines up week-for-week
with the January price cut (takeaway 2). Proof: the timeline
2
Price is the strongest lever at the shelf - it works immediately.
The 1.35→€0.91 cut ignited growth in winter; the spring return to €1.19-1.35 cut sales to 1.22,
and €0.91 again lifted them to 1.95 (about +60%). A signal, not proof - but consistent in both directions.
Proof: the price slider
3
Campaigns buy the brand, not weekly sales: searches ×3.8.
After the flights started, Google queries grew from 64 to 247 per month. Rotation barely reacts
to weekly budget (correlation at most 0.16) - price works at the bottom of the funnel, campaigns at the top.
Proof: chapters 1-3
4
Puro grows faster than its surroundings - in winter too.
Family share ×2.4 in under a year; in winter Puro +31% while the sparkling segment lost about 29% (the winter comparison is a hypothesis - different windows).
This is not just a category wave. Proof: the DIA shelf
5
Two channels, two roles: Instagram builds reach, TikTok builds followers.
Instagram is 74.5% of the 40.7 k € spend and 8.2 M reach (€3.69 per 1,000);
TikTok campaigns (24.5%) report 8.9 k acquired followers. Proof: platforms
Business scale: at 600 stores the difference between the weeks 24-36 level (1.95) and the pre-campaign baseline (0.79)
is about 700 units weekly; only the unit margin is missing to convert this into euros (Conversions section).
Rotation: units sold per store per week, regular sales (no promotions), item 307333, SMD3 stores,
DIA contract base v2026-09-21_v1; the series lacks weeks 39/2025 (trace launch sales of 0.003 units,
excluded as atypical) and 17/2026 (promo week: sales in the promo row, outside the regular series).
The SMD3 label follows the project thesis register (thesis #23);
the store-population definition is to be confirmed with the data owner.
+119%: average regular rotation in active-campaign weeks (15-36/2026) = 1.72 vs 0.79 in the weeks before
(excluding weeks with no data). Other baselines give other multipliers (from the low teens when comparing final week 36 with week 15, to over +200% from the 0.53 low),
which is why the baseline is declared here and the chart shows the full series.
Searches: Ahrefs Keywords Explorer, Spain; monthly volume assigned to that month's weeks;
August and September 2026 may be incomplete. The ×3.8 multiplier = peak of 247 vs the winter-months average of 64
(December-February: 41, 54, 98).
IG reach: Meta Business Suite, weekly account reach. Investment: daily Meta Ads export aggregated to ISO weeks;
sourced in PLN, presented in € at the rate described below.
Cost of reach: Instagram spend of 30.3 k € / IG reach of 8.22 M from the Meta platform report
≈ 3.69 € per 1,000 (reach deduplicated within a campaign, summed across campaigns; full export).
Weekly reach sums (19.1 M) and mixing platforms in the denominator inflate or distort the result - not used.
Exchange rate: applies only to Meta spend incurred in PLN, converted at the average NBP table-A rate
for 01.04-22.09.2026: 4.2849 PLN/€ (122 quotations). Campaign bands on charts: weeks with combined (Meta+TikTok) spend above €100.
TikTok: Ads Manager export, amounts natively in €; daily spend aggregated to ISO weeks
(full account 10.0 k €, 9.2 k in the chart window). Account effects: 8.9 k paid follows per Ads
and 7,463 new followers per TikTok Analytics (Feb-Sep 2026); the gap is possible ad over-attribution - to be checked.
Price periods (implied from data: rot_eur / rot_ud weekly): €1.35 (wk 40-52/2025),
€0.91 (2-15/2026), €1.19-1.35 (16-22/2026, incl. full 1.35 in wk 19-22), €0.91 (from 24/2026);
transition weeks 1/2026 (0.97) and 23/2026 (0.96) assigned to adjacent periods; promo week 17/2026
(implied price ~1.13) outside the regular-series periods. Since September 2026 the dia.es price is €0.79 (outside the data window). Rotation: 1.71 in week 15 just before the window,
1.22 in window 16-22 (the first expensive week already at 1.39), 1.95 after the return to €0.91 (weeks 24-36).
The TikTok start (week 16) coincides with the start of the expensive window - despite the extra reach rotation still slowed, which gently strengthens the price signal but complicates campaign attribution. The post-cut jump coincides with high-spend weeks (23-24; the highest-budget week is 31) - effects inseparable;
a price-sensitivity signal, not proof (trend and season in the background).
Family share: Puro regular rotation / total regular rotation of all AGUA MINERAL family items, weekly, canonical base v2026-09-21_v1 (SMD3 stores).
Change correlations: Pearson on week-over-week differences, spend vs regular rotation and share,
lags of 0-4 weeks, combined Meta+TikTok investment. Rotation: [-0.10; 0.01; 0.16; -0.05; -0.35],
share: [0.03; 0.03; 0.00; -0.17; -0.13]; all values weak - no weekly effect is visible in this data.
Season (hypothesis): CON GAS segment summer (wk 28-36/2025) vs winter (44/2025-8/2026) = -29%;
Puro autumn (40-52/2025) vs winter (1-8/2026) = +31%; two different contrasts, hence the hypothesis label.
Benchmarks: Komvida = ginger-lemon kombucha SKU 250 ml, average for weeks 15-36/2026;
Font Vella = sparkling-water SKU 500 ml, year-over-year change (weeks 1-36); Font Vella brand = 8 family SKUs, +13.5% YoY.
Campaigns targeted Spain (account billed in PLN); audience geography is not available in the exports.
Simulator: average regular rotation across weeks matching the lever condition; price groups merge periods
(e.g. the €1.35 level covers autumn 2025 and weeks 19-22/2026), so the result shows co-occurrence, not effect. Budget levels are not monotonic (the middle one can be lower than its neighbours) - consistent with near-zero correlations: no budget dose is visible in the weeks.
Unit conversions: observed rotation × a shared assumption of 600 stores; the actual store count
per DIA's monthly file (396→613 through 07/2026, August-September missing) - the file's provenance is unconfirmed,
so the scale is an assumption and the cards are conversions of observations, not forecasts or attribution.
Cross-check: monthly units from the same file give ~890-990 units/wk for June-July;
the ~20% gap has a possible explanation in store-definition differences (selling in a week vs in a month) - to be checked. Meta objective mix: 85% of budget
on visits, follows and engagement, 15% on reach (full Meta export).
Shelf sales value: monthly sum (units × that month's average implied price)
for Sep/2025-Jul/2026 = 25,920 € at 24,627 units. Note: a different window than the spend (which runs to week 36, early September) - do not compare directly. Source: DIA's monthly file of unconfirmed provenance; an approximation.
Coincidence in time does not prove causation; the limits of inference are described in "What this picture does not prove".