Neodrop
Create
FeedDiscoverMy ChannelsAuto-publishJoin Our Discord
FeedDiscoverCreateMy Channels
2026-11-01

Deep Reads: When the Number Is Not the Thing

Deep Reads Weekly
Deep Reads Weekly@Ana Askari
Deep Reads: When the Number Is Not the Thing

A pharmacy chain in the United States ran promotions across nine billion dollars of front-store sales, and when its analysts added up the profit, the total came back about twenty-five million dollars negative.

0:00 / 10:12
This week's three reads were published between 27 October and 1 November 2026, and each one is about a number that arrived without the thing it was supposed to describe: a sales lift with no profit in it, a barrier percentage with no mechanism underneath it, a completed research task with nothing expressed in it. The question they share is what a figure is actually a reading of.

A price cut that sold more and earned less

Marco Vriens writes about pricing for Research World, and his piece on 28 October is a synthesis of published retail-pricing studies rather than the report of a single one. 1
His anchor case is the promotional programme of a large American pharmacy chain, which was measured promotion by promotion. Across roughly $9 billion of front-store sales, the net profit effect of the whole programme was about minus $25 million. More than half of the promotions lost money. The ones that paid were mostly manufacturer-funded buy-one-get-one offers, promotions on the chain's own brand, and promotions on high-priced brands that are on deal all year anyway. 1
The arithmetic explains why. A $2.50 item at a 30 per cent gross margin earns 75 cents. Cut the price 10 per cent and the margin falls to 50 cents, a third of it gone, so the retailer needs to sell half again as many units to stand still. Vriens then uses the average short-term promotional elasticity from a meta-analysis of 1,851 published price elasticities, about −3.63, which says a 10 per cent cut lifts sales by roughly 36 per cent. At 10,000 units the promotion earns about $6,800 instead of $7,500, a gross profit about 9 per cent lower than doing nothing. About a third of a promotion's lift, he adds, is shoppers switching from a competing brand on the same shelf. 1
The turn worth carrying into a meeting is about the benchmark. Studies like these treat the regular price as given, but a chain that promotes constantly has no real regular price: its shelf price is inflated precisely because the promotions exist. With a more realistic benchmark the pharmacy chain's promotions look less unprofitable, though they remain the weakest thing it does. 1
The other half of the piece is about price increases. A European DIY retailer, bauMax, modelled price changes across 60,000 products, tested the predictions on a handful of items in a few stores, then repriced about 20,000 items, deliberately leaving around 100 competitor-tracked products untouched to avoid a price reaction and tracking whether shoppers began to see the store as expensive. Gross profit rose about 8 per cent. For the pharmacy chain the same reasoning pointed at the promotions that lost money: eliminating them chain-wide would cut sales by about $7.8 million and add about $52.6 million of profit a year. 1
The boundary to keep. Vriens is summarising other people's work, mostly American and European grocery and pharmacy, so the elasticity he uses is an average across categories, and category is exactly what decides the answer. The DIY case is the one place in the piece where a prediction met a real till.

The same barrier, two different machines

Sandeep Budhiraja is founder and CEO of Spark Sensory AI, and his piece ran on Greenbook on 27 October. In June 2026 his team surveyed 1,000 adults, 500 in the United States and 500 in India, mixing people already dealing with diabetes, prediabetes or obesity with people who have no diagnosis. 2
Awareness splits hard: 70 per cent of the American sample had heard of GLP-1 therapies against 38 per cent in India. Openness to the treatment if a doctor recommended it came back at 70 and 72 per cent. Two markets 32 points apart on awareness sit 2 points apart on willingness, and actual use runs the other way again, at 20 per cent current or past users in the United States against 6 per cent in India. His reading is that India is under-informed rather than resistant, which moves the strategic question from how to persuade people to what they need to know first. 2
Cost is the leading barrier in both countries, at 42 per cent in the United States and 46 per cent in India, and the identical label covers two different machines. In the United States affordability runs through insurance coverage, prior authorisation, employer benefits and patient assistance programmes, so a consumer can want the treatment, have physician support, and still fail to convert. In India it runs through out-of-pocket economics with much less category familiarity underneath. His line is the one to keep: a barrier percentage is not an insight until you understand the mechanism underneath it. 2
The physician gets the same treatment. Physicians matter in both markets, which on its own is useless; what differs is when they enter. Sixty per cent of the American sample knew someone who had used a GLP-1 therapy, so the doctor often joins a journey already under way and validates an existing intent. In India the physician is more often where understanding of the category begins. He closes on a research point rather than a marketing one: healthcare research is good at describing populations and weak at watching movement, and movement is what carries someone from awareness to relevance, or from a recommendation to a prescription that keeps being filled. 2
The boundary to keep. This is the author's own study published under his company's name, and the piece does not report how the thousand people were recruited, what quotas or weighting were used, or whether the two samples were balanced the same way. The openness question is hypothetical in any case: a stated intention towards a medicine that costs money and requires a prescription.

Children finishing the task

Jaime Solórzano is head of insights at Q2Q Global and Martha Llobet runs the company. Their piece went up on Greenbook on 30 October as the first instalment of a series built on proprietary research in ten countries with children aged eight to fifteen. 3
The standard adaptation of an adult method is to shorten the questionnaire, simplify the wording and add a few visual tasks. Their argument is that this buys less than people believe, because the environment itself is built on adult habits: a linear questionnaire assumes a child can hold a thread, a retrospective question assumes they can explain why they did something, and most formats assume the platform world these children live in stops at the door. 3
In the qualitative phase, children answered at their own pace, in whatever form they chose — a voice note, a photo, a drawing, an emoji — and came back later with more. The finding that matters is that these were not decoration around a text answer: for this cohort the voice note and the drawing carry the evidence. 3
Then the figure. In the quantitative study, 66 per cent of children said they adjusted what they posted according to the reactions they expected. A child entering a research session is already working out which answer will be rewarded, what will sound acceptable, how the moderator will take it, and a method can either relieve that pressure or reproduce it. The authors also name who each method favours: interviews give room to confident speakers, long surveys reward sustained attention, and in a tightly moderated group the quickest answer tends to become the group's answer. 3
Their design rule is to ask what is the most natural way for this participant to express this particular experience, rather than how to make the usual method suitable for children. Engagement is separated from entertainment: a colourful game is not participant-centred unless it lets a child say something a conventional question could not reach. AI appears as support — moderating exchanges, flagging answers that deserve a follow-up — with interpretation, cultural and developmental expertise, and safeguarding built into the process rather than added as a final review. 3
The boundary to keep. The authors run the firm whose proprietary study this is, and this is the first instalment of a series, so the sample sizes, the recruitment approach and the questionnaire are not disclosed yet. The 66 per cent figure is self-reported and comes from their own study. The broader claim, that a generation is different in kind, is easy to make and hard to test; what can be tested is narrower and more useful.

What to take into the week

Three numbers: minus $25 million, 42 per cent, two thirds. The first was profit measured against a benchmark nobody had questioned. The second was a barrier counted in two countries and built from two different machines. The third was children telling researchers that they were being watched.
Before a number goes into a deck, write one line underneath it: what was this a reading of, and what was it measured against? If the answer is a model's average, say so. If it is a question people answered hypothetically, say that. If it is a task people finished, say that too, and then go and find out what they did not say while they were finishing it.

References

  1. 1
    When Do Pricing Strategies Work for Retailers? — Research World

    researchworld.com

  2. 2
    The GLP-1 Adoption Paradox — Greenbook

    greenbook.org

  3. 3
    Researching Gen Alpha Series — Greenbook

    greenbook.org

Deep Reads Weekly
Deep Reads Weekly@Ana Askari

This story was produced automatically by a channel. One sentence is all it takes for Neodrop to keep producing for you.

Related content