Consumer Market Analysis: Sizing, Segmentation and Gaps, With a Worked Example From Public Data

Public data is free and large but rarely matches your category exactly. A survey matches your category but rests on what people say. Use each to check the other.

MoSPI's Household Consumption Expenditure Survey 2023-24 covered 2,61,953 households between August 2023 and July 2024. It reports average monthly per capita consumption expenditure (MPCE) and the share spent on each item group (MoSPI press note, December 2024).

Take "beverages, refreshments and processed food", which MoSPI says has the largest share of food spending in rural and urban households. Its share of MPCE was 9.84% in rural areas and 11.09% in urban areas, against average MPCE of ₹4,122 and ₹6,996 (same source, figures without the imputed value of free items).

Multiply the last row by the population of your target area, in lakh, from a source you name in the report. That gives the ceiling for the whole item group. Your category, say packaged snacks, is a slice of it, so the next step is a share you estimate, ideally from retail data or your own survey.

Three cautions. The item group is broad. The figures are national averages, and MoSPI shows wide state gaps. And they are 2023-24 rupees, so adjust if you are sizing a later year.

Bottom-up sizing builds the market from buyer behaviour:

Market value = households in the target area × share who bought in the period × purchases per buyer × average spend per purchase

An illustration with made-up survey results for one city, so you can see the arithmetic:

Now compare with the top-down ceiling, which is per person, so convert households to people first. If the city's households average, say, 4.5 people (use the household size from your population source), 2,00,000 households hold 9 lakh people. The urban ceiling for the whole item group is then 9 × ₹93.1 crore = ₹837.9 crore a year, and the bottom-up ₹13.44 crore is about 1.6% of it. If the bottom-up figure were bigger than the ceiling, an input would be wrong. Check the recall period, the wording of the purchase question and whether you screened the right household member.

Averages hide the spread. MoSPI's fractile figures for 2023-24 show how far apart households are (MoSPI):

In urban India the top 5% spend about 8.5 times as much per person as the bottom 5% (₹20,310 ÷ ₹2,376). Among major states, urban MPCE ranges from ₹8,978 in Telangana to ₹4,927 in Chhattisgarh (same source).

Spending data gives you the economic frame. Your survey then segments buyers within it, by need, usage or attitude. Decide the segmentation approach before fieldwork:

• A priori segments use variables you choose, such as city tier, NCCS and age. They are easy to target and repeat.
• Data-driven segments cluster respondents on their answers. They can reveal groups you had not considered, but need enough completes for each cluster and a test of whether the clusters hold up.

Our market segmentation survey guide covers the questionnaire.

These questions cover the competitive and gap parts of the analysis. Adapt the wording to your category.

For the gap analysis, plot each attribute with importance on one axis and satisfaction on the other. Attributes high on importance and low on satisfaction are your candidate gaps. Treat them as hypotheses and test them with a concept test before you build anything (concept testing).

In our view, AI tools help most with the slow parts: drafting a first questionnaire, grouping thousands of open-ended answers into themes, and summarising tables. They do not supply the sizing inputs. Penetration, frequency and spend must come from real respondents or published data.

Check every AI summary against the table it came from, and code a sample of open-ended answers yourself to see whether the themes match. Our best AI survey tools guide sets out a test for this.

You need actual sales and market share. A survey measures claimed buying. For shop-level sales, use retail measurement data; see our NielsenIQ alternatives guide.

Your category is sold mainly to offline or rural households. Our panel reaches smartphone users only, so the bottom-up inputs would describe connected buyers.

We can field the bottom-up survey and the question bank above. Say you size the category in two cities with 400 completes each: the bill is 800 × ₹120 = ₹96,000 on the Pro plan (see Pro). Recall wording drives the whole estimate, so trial the purchase questions on a free run of 30 answers before you buy the bundle.

Respondents can be targeted by city, tier, region, NCCS and age, which lets you set the quotas your segments need. Each segment's answers are tabulated with significance flags, and the open "what is missing" answers are grouped into coded themes (panel methodology).