A founder who has spent eight years managing consumer goods distribution across Maharashtra is considering a nutrition range for urban households. She knows which products move through general trade, which price points retailers accept, and which categories attract frequent investment from larger brands. Her uncertainty is narrower and more important: which product will a household buy again after the first trial, and which proposition is worth building a brand around?
That is the point at which traditional category judgement often becomes expensive. A founder may have a credible view of a trend, a familiar competitor set, and a product that looks sensible on a first purchase. None of those signals establishes repeat demand. In consumer goods, the first transaction can reflect curiosity, discounting, gifting, or a well-placed advertisement. The second and third transactions reveal whether the product earned a place in a routine.
At Palette Brands, we start with that distinction. We analyse purchase and reorder data from Indian e-commerce, then use the signals to find products with genuine repeat demand. From there, we build or back consumer brands around the strongest evidence. The work is not about replacing category judgement. It is about deciding where judgement should be applied, and where it should be tested before a founder commits to a product line.
Category belief is not category evidence
Founders usually enter a category with a belief system already in place. It may come from operating experience, conversations with distributors, trade publications, competitor observation, or a personal conviction that Indian consumers are ready for a better version of an existing product. Those inputs are useful. They help define the initial question.
They do not answer the question on their own. A category can be growing while most of its new products fail to generate repeat purchase. A competitor can have strong visibility while its customers remain highly promotion-sensitive. A trend report can describe rising interest in a need state without showing which format, pack size, price, or usage occasion becomes habitual.
We call this difference category belief versus category evidence. Belief says a problem is important or a market appears attractive. Evidence shows how people behave after they have already paid. For an early-stage consumer company, that behavioural sequence is often more useful than a large top-down market estimate.
The sequence also makes the question more precise. Instead of asking, "Should we enter personal care?" a founder can ask, "Which personal care products are reordered without an unusually deep discount, by which customer cohorts, at what interval, and in what basket context?" That is a decision that data can materially improve.
What purchase data can show before a launch
Indian e-commerce generates more useful pre-launch information than many founders initially assume. The most basic layer is SKU-level transaction history: when an item was purchased, at what price, in which pack configuration, and whether the order included a promotion. Even without personally identifiable information, aggregated records can show whether demand is distributed across many customers or concentrated in a small set of heavy buyers.
Order sequences add more detail. A founder can examine:
- the time between first purchase and reorder;
- the share of first-time buyers who return for the same SKU;
- whether customers repeat the exact product or move to another format, size, or price tier;
- the number of orders in which the product appears alongside complementary items;
- how reorder behaviour changes when a promotion is absent;
- differences between new, returning, and high-frequency customer cohorts.
Product ratings and written reviews can provide context, but they should not be treated as a substitute for transactions. A highly rated product may still have weak repeat purchase if it is bought occasionally, gifted, or selected mainly for a one-time need. Conversely, an ordinary-looking product page can sit behind strong reorder behaviour because the product performs reliably in a routine.
Timing matters as well. If purchases recur at a plausible usage interval, that is a different signal from a cluster of orders created by a sale or a seasonal event. The useful question is not simply how many units moved. It is whether the pattern resembles replenishment.
Reading the sequence before building the brand
Transaction data is most valuable when it changes what gets tested. Suppose a founder is evaluating household cleaning products. A broad category view may suggest that consumers want safer formulations, more pleasant scents, or convenient formats. Those are possible positioning territories, not validated propositions.
A sequence analysis might show that one concentrated format has a higher reorder rate among customers who initially buy a smaller trial pack. It might also show that repeat customers purchase at a predictable interval, while a larger pack attracts fewer second orders. That does not prove the concentrated format is the right product to launch. It does indicate where the first product and the first test should be more tightly defined.
The same approach applies to nutrition and personal care. Instead of beginning with a complete range, a founder can compare products by repeat rate, reorder interval, cohort retention, price changes, and basket relationships. A product that appears smaller in absolute sales may be more interesting if its buyers return consistently and are not dependent on a short promotion window.
At Palette, our early incubation tracking across three Indian consumer categories recorded a 58% average reorder rate across products, compared with a 31% category baseline. These figures are based on our internal pilot tracking. They are not a forecast for every category or a promise of future brand performance. Their practical value is that they give the team a way to distinguish an observed repeat signal from a general belief that a product should work.
Before spending on identity, packaging systems, large inventory commitments, or a broad launch plan, founders can use the sequence to define a narrower test. The test may compare two pack sizes, two functional claims, or two price points. It may focus on one use occasion rather than the whole category. It may also reveal that a product deserves no further investment, which is a useful outcome when the alternative is learning after a full launch.
Choosing the category and the first positioning
Category selection should not be reduced to finding the highest reorder rate. A high repeat rate in a small, tightly defined customer group may not support the intended distribution model. A product with broad demand may have weak economics after returns, discounts, fulfilment, and customer acquisition costs. The decision needs several signals considered together.
We typically want to understand four things. First, is repeat behaviour broad enough across cohorts to suggest more than a niche habit? Second, is the reorder interval compatible with the proposed product economics and working capital cycle? Third, does the product retain demand when price and promotion are separated from the analysis? Fourth, are there adjacent products or unmet usage occasions that give a new brand room to enter without simply copying an incumbent?
These questions help a founder choose between category entry options. They also inform the first positioning test. If customers reorder for a particular functional outcome, the initial proposition should make that outcome clear. If repeat buyers are concentrated around a specific usage occasion, the product and communication should be designed around that occasion. If customers regularly switch between formats, the opportunity may be in solving a trade-off rather than adding another undifferentiated variant.
Positioning should be tested against behaviour, not selected because it sounds distinctive in a workshop. A founder can test whether a benefit-led proposition attracts the right first customers, whether those customers reorder, and whether the second purchase occurs without unusually high incentives. This creates a more useful learning loop between product, proposition, and demand.
The boundary of purchase data
Purchase data tells you which products people buy again. It does not tell you why the proposition resonated, how to differentiate the brand story, whether the product experience is genuinely superior, or whether a customer would recommend it to someone else. Those questions require qualitative research, product testing, creative judgement, and direct customer conversations.
Data can also mislead when the underlying records are interpreted without context. A reorder may be made by a household member different from the original buyer. A discount may be hidden in a bundle. A subscription-like interval may reflect stock-up behaviour rather than routine consumption. Marketplace availability, search placement, delivery coverage, and stock-outs can all affect what appears to be demand.
This is why data-first brand building does not mean data-only brand building. Product quality, sourcing, formulation, pricing, packaging, distribution, compliance, and service still determine whether a repeat signal can become a durable business. Data can help decide what deserves a serious test. It cannot manufacture a product that customers want to keep using.
The founder in our opening example might therefore leave the analysis with one nutrition product, one primary usage occasion, and two positioning hypotheses rather than a twelve-SKU range. That is not a smaller ambition. It is a more disciplined commitment. If you want to map the purchase data in your target category before committing to a product line, that is exactly the kind of conversation we have with early-stage founders.
