Skip to content
GullySales

Notes for owners · Digital marketing

Consumer audience targeting when the segments have gone

Detailed interest audiences have thinned out and narrow targeting now costs more than it saves. Exclusions, a sharp advertisement and enough budget to learn replace them.

The GullySales team · Updated 21 Sept 2026 · 8 min read

Consumer targeting in India has changed shape. The detailed interest and behaviour segments campaigns used to be built on have thinned out, income was never available here, and three categories have had demographic targeting withdrawn. Narrow targeting now usually costs more than it saves, because a small audience sees the same advertisement too many times. The work has moved to three things instead: what you exclude, how sharp the advertisement is, and whether you have spent enough in one place to learn anything.

If you have not yet written down who the buyer is, start with how to identify your advertising audience and come back. This page is about what to do with that answer once you have it.

Narrow audience, or broad with a sharp advertisement

Narrow targetingBroad, with a filtering advertisement
Who sees itA small, defined setAnyone the platform thinks might respond
Frequency after two weeksHigh. The same people, repeatedlyLow, because the pool keeps refilling
Cost per clickUsually higher, and risingUsually lower
Wrong enquiriesFewer, but you paid for the filteringMore, unless the advertisement does the filtering
How fast you learnSlowly. Too few results to readQuickly, if the budget is in one place
What breaks itAudience fatigue in week threeVague creative that could apply to anyone

The right-hand column only works if the advertisement itself disqualifies people. A price, a model name, a delivery window, a locality, a starting figure. Put one of those in the first line and the wrong reader moves on before you pay for a click.

Exclusions do more than selections

This is the part most accounts have never been set up properly, and it costs money every day.

  • Existing customers, for anything they have already bought and will not buy again this year
  • People who enquired and were quoted, who belong in follow-up rather than in a fresh campaign
  • Your own staff, your suppliers, and the agencies who keep clicking to check on you
  • The pin codes you cannot serve, or will not travel to
  • Job seekers, who arrive in volume on any advertisement that looks like a company post
  • Anyone who bought once on a discount and never returned, if you are building a lookalike

For a retailer with a few thousand customers, excluding buyers alone often shifts the cost per order more than any new audience does. You stop paying to advertise to people who already walk in.

Spend enough in one place to learn something

A test that produces nine results tells you nothing, however neat the report looks. Before you split a budget, work out how many enquiries you need before the difference between two sets is real. For most small businesses that means forty or fifty results per set, which usually means two sets and not eight.

Give it a fortnight without changes. Editing the budget, the creative and the audience in the same week means you will never know which one moved the number. This is the most common way an Indian small business wastes a monthly budget, and it costs nothing to stop.

Your own data, and the rules around it

First-party data now outperforms bought audiences, and it carries obligations. Under the Digital Personal Data Protection Act you need consent for the purpose you are using the data for, given when you collect it, with a way to withdraw it. A phone number given at the billing counter to send an invoice was not given for advertising.

So fix the notice at the counter, on the enquiry form and in the WhatsApp opt-in. Then the customer list, the enquiry list and your exclusions become assets you can use without worrying about who asks.

Test the language, then answer in it

Run the same offer in English and in the state language, with the budget split evenly, and let the cost per enquiry decide. The winner is not predictable, which is exactly why it is worth testing rather than arguing about. Local audience targeting explains why the language setting in the platform is not the same thing as the language of the advertisement.

The half nobody completes is the reply. An enquiry that arrives in Kannada and is answered from an English template loses the advantage you just paid for.

Judge it on the right number

Click-through rate tells you the creative got attention, not that it found a buyer. Read cost per enquiry, the share of enquiries answered within an hour, and cost per order.

For example, a furniture retailer in Bengaluru might drop interest targeting altogether. One broad campaign across a twelve kilometre radius, existing customers excluded, and "solid wood, delivered in four weeks, from a stated price" in the first line. Two creatives, one language test, a fortnight untouched. Any figures in a plan like that are illustrative until the first month is measured, but the structure is what transfers.

In every audit we record enquiries by source, reply time, conversion and cost per order before anything is changed. Without that baseline, the argument about targeting cannot be settled, and it usually turns out not to be an argument about targeting at all.

What to do next

Open your ad account and look at the exclusions on your best performing campaign. Empty? You have found this month's easiest saving. Then count how many results each audience produced last month, and merge everything that produced fewer than forty. Book a free audit if you want that account read alongside your enquiry log.

Questions

Questions owners ask.

Should we still use interest targeting at all?
Use it to start a campaign, not to run one. Interests in India are inferred from pages followed and things clicked years ago, so they age badly and overlap heavily. Begin with two interest-based sets if you like, but plan to move the budget to a broad audience with real exclusions once you can see which creative is working.
Can we upload our customer list to build an audience?
Only if the people on it were told their details would be used this way. Under the Digital Personal Data Protection Act, collecting a number to deliver an order and then using it for advertising is a purpose question, not a technical one. Fix the consent notice at the point of collection, then upload.
How many audiences should one campaign have?
Fewer than you have now. Splitting a small budget across eight audiences means none gets enough data to learn from and you spend the month comparing noise. Two or three, with enough spend each to produce forty or fifty results in a fortnight, tells you something you can act on.
Our cost per click is fine but nothing converts. Where do we look first?
At the hour the enquiries arrive and how long they wait for a reply. In most of the accounts we audit, the targeting is not the problem and the reply time is. Fix that before you rebuild the audience, because a rebuilt audience with the same reply time produces the same result.

Get a free audit of how you sell, and a scored report of where the work is.

Book a free audit