In short
- Lift is the gap between people who saw the content and a comparable group who did not, in percentage points (absolute) or as a share of the control figure (relative).
- A survey needs about 1,100 people per group to detect a rise in awareness from 20% to 25%. Smaller samples cannot confirm small lifts.
- Branded search is the cheapest lift signal. Read it against a baseline and a control term, and remember that Google Trends is a relative index, not a count of searches.
- Organic creator posts give you no clean control group. Partnership ads, regional creators or time-staggered waves do.
To measure brand lift from influencer marketing, compare people who saw the creator content with a similar group who did not, on a question such as "Have you heard of this brand?", and report the gap in percentage points. In practice you have four routes: a survey, branded search movement, a platform lift study on creator ads and a geo or time holdout. Use at least two, because organic creator posts rarely give a clean control group.
Our guide to measuring influencer marketing ROI lists these methods in one table. This post shows how to run each, with sample sizes and worked examples. Lift matters because Indian consumers do use creators to discover products: 63% do for discovery, according to a 2025 Kantar-backed study reported by Campaign India. For the wider plan, see the complete guide to influencer marketing in India.
What lift means: four formulas
Take a survey where 20% of the control group and 26% of the exposed group know your brand, and a campaign that reached 5,00,000 unique people at a cost of ₹3,00,000.
| Measure | Formula | Example |
|---|---|---|
| Absolute lift | Exposed % minus control % | 6 percentage points |
| Relative lift | (Exposed minus control) ÷ control | 30% |
| Incremental people | Absolute lift × unique people reached | 30,000 |
| Cost per incremental person | Campaign cost ÷ incremental people | ₹10 |
The last row is the useful one, because it lets you compare creators with other channels on the price of each extra person who knows you. It only works with unique reach. If you add each creator's reach together, overlapping audiences are counted twice and the lift is overstated.
Choose one question as the primary measure before you start: unaided recall (which brands in the category can you name), aided awareness (have you heard of this one), consideration or purchase intent. The rest are diagnostics.
Four methods and when each fits
| Method | What you need | Cleanest when | Main weakness |
|---|---|---|---|
| Survey, exposed against control | Survey tool or panel, enough respondents | Exposure is known, as with creator ads | Self-reported; the exposed group is not a random sample |
| Search lift | Search Console, Google Trends, a baseline | Any campaign, at almost no cost | Other marketing moves branded search too |
| Platform lift study | An ad platform account and a minimum budget | Creator posts run as partnership ads | Covers only the ads, not organic reach |
| Geo or time holdout | Sales data, comparable regions or periods | Regional creators or geo-targeted ads | Needs order volume; audiences spill across regions |
Survey lift: design and sample size
Ask both groups the same question in the same order, randomise the brand list so yours is not always first, and include a few decoy brands that do not exist or are not in the category to estimate false claims. Ask about the category first, then the brand.
The hard part is deciding who counts as exposed. For creator posts run as ads, the platform knows. For organic posts, you cannot know who saw them, and the options are weak. You can ask respondents which creators they follow, but self-reported exposure is unreliable, which is why decoy creators help. You can survey the creator's followers and compare them with a matched panel, but followers are already fans, so the gap overstates the campaign's effect.
How many respondents? This table uses the standard two-group comparison at 95% confidence and 80% power. It is arithmetic, not an India benchmark.
| Awareness in control | Awareness in exposed | Lift | People needed per group |
|---|---|---|---|
| 20% | 30% | 10 points | about 290 |
| 20% | 25% | 5 points | about 1,100 |
| 10% | 12% | 2 points | about 3,800 |
A small survey cannot confirm a small lift. With 500 people per group and awareness near 25%, each group's estimate carries a margin of error of about 3.8 points, and the gap between the two groups carries about 5.4 points. A measured 5-point lift on that sample is indistinguishable from chance. If you cannot afford the sample, do not run the survey. Use search lift and a holdout, and say that is why.
Search lift: the cheapest signal
People who are curious after seeing a creator often search the brand by name. That makes branded search a behavioural signal rather than an opinion.
| Source | What it shows | Limit |
|---|---|---|
| Search Console | Your site's impressions and clicks for queries containing your brand name | Google Search only; your site only |
| Google Trends | Relative interest on a 0 to 100 scale, by country, state and time | Not search volume; 100 is the peak in the period you selected |
| App store search data | Installs that began with a search, in the store's own console | Stores differ in what they report |
| Marketplace search and rank | Search position or traffic where the seller dashboard shows it | Only some marketplaces provide it |
Read search lift against a baseline and a control. Take four weeks of branded clicks before the first post. Then compare the campaign window with the baseline, and compare a control term, such as the generic category word, over the same weeks to remove seasonality.
Worked example, made-up numbers: branded clicks average 120 a day in the baseline and 168 a day during the campaign, a rise of 40%. The control term rose from 300 to 330 a day, up 10%. Adjusted lift is 1.40 ÷ 1.10 minus 1, about 27%. Reporting 40% would overstate what the campaign did.
Also look at timing. Check the 72 hours after each post. Branded search that rises on posting days and falls back is evidence for that creator. A steady climb across the month points to something else. Note, too, what an index can and cannot say. BBH India's chief creative officer reported (agency-reported, not independently verified) that Garnier Men's 2025 comedian-led campaign saw Google search interest rise from 0 to 100, alongside a 44% sales rise, as covered by Branding in Asia. On Google Trends, 100 marks the peak of the chosen period. "0 to 100" shows the shape of the spike, not how many people searched.
Platform lift studies on creator ads
When a creator post runs as a partnership ad from your account, the ad platform can in principle hold out a randomly chosen control group, show the ad to the rest and survey both. That is the closest thing to a controlled experiment in influencer marketing, because exposure is known and assignment is random. Meta says thousands of Indian advertisers run Partnership Ads, as exchange4media reported in March 2025.
Before you plan around it, ask your platform representative which study types support partnership ads, what the minimum budget and audience size are today and which questions the survey asks. We do not quote thresholds, because we have none sourced and platforms change them. Two limits remain: the study covers only the paid exposure, not organic reach, and the survey belongs to the platform. Our creator ads and whitelisting service covers the set-up.
Holdouts: audience, geo and time
A holdout withholds the creators from a comparable group and measures the difference.
- Audience holdout inside creator ads. The platform randomly withholds the ad from part of the target audience. It is the cleanest design and needs enough volume.
- Geo holdout. Run creators in a set of cities and withhold them from comparable cities. Compare the change before and after in both.
- Time holdout. Release creators in waves and watch the step in orders or searches each time. It is the weakest design and the cheapest.
Worked example, made-up numbers, for a geo holdout. The test cities went from 10,000 orders in the four weeks before to 12,600 during the campaign, a rise of 26%. The control cities went from 8,000 to 8,800, up 10%. Lift is 1.26 ÷ 1.10 minus 1, about 14.5%. Had the test cities grown only at the control rate, they would have reached 11,000 orders, so the campaign added about 1,600.
There is an India-specific catch. Organic creator posts reach audiences everywhere, so you cannot hold out a city from a Reel. A geo holdout works when creators' audiences are concentrated, as with regional-language creators (see our regional influencer marketing service), when you run geo-targeted creator ads, or when you sell through local delivery or store markets. Pick clusters whose orders moved together before the test, keep offers equal in both, and use a baseline of at least four weeks.
Reading and reporting the result
- Within the margin of error means "no detectable lift", not "no effect". Say what the sample could and could not detect.
- Report both forms of lift with the sample size, the window and the method.
- Convert it to cost per incremental person and compare it with your other channels.
- Do not turn lift into a sales promise. Lift is a leading indicator. It does not carry a fixed conversion rate, and we found no India figure that links one to the other.
Our post on influencer marketing attribution shows how to place a lift result beside codes, links and survey answers, and the report template shows where it goes.
Which method for which campaign
| Campaign | Best first method | Add |
|---|---|---|
| One creator or a small micro-programme | Search lift and post-level spikes | A post-purchase question |
| Regional-language programme with many creators | Geo holdout | Google Trends by state |
| Creator ads at volume | Platform lift study or audience holdout | A survey of your own |
| Launch combining PR and creators | Search lift plus survey | Staggered creator waves |
Mistakes to avoid
- Surveying followers and calling it lift. They were fans before the campaign.
- Comparing campaign weeks with a baseline and no control. Seasonality fakes lift.
- Treating Google Trends as volume. It is an index.
- Running an underpowered survey. A small sample can only confirm a large lift.
- Expecting lift in last-click sales. Last-click reporting credits search or direct visits for what a creator started.
If you want a lift design for your next campaign, send us a brief. Our performance influencer marketing service includes holdout or survey design, and the KPI guide shows where lift sits in a scorecard.
Sources
- Indian consumers and influencers: coverage of a 2025 Kantar-backed study of consumers and brands, Campaign India (12 Jun 2025)
- Creative dive: Parikshit Bhattacharya on Cannes Lions-winning Bassi vs Garnier Men (sales and search interest figures reported by the agency), Branding in Asia (4 Aug 2025)
- Meta launches AI-powered creator marketing solutions for brands (Partnership Ads in India), exchange4media (26 Mar 2025)
Frequently asked questions
What is brand lift in influencer marketing?
Brand lift is the increase in awareness, consideration or purchase intent among people exposed to creator content, compared with a similar group who were not exposed. It is usually expressed in percentage points. Because it measures attitude and behaviour before sales, it suits campaigns where the goal is to be known, not only to sell.
How do you measure brand lift from influencers?
Four ways: a survey of exposed and unexposed people, branded search movement against a baseline, a platform lift study on creator posts run as ads, and a geo or time holdout that compares sales or searches where creators ran and where they did not. Use at least two, because each has a weakness the others cover.
How big a sample do I need for a brand lift survey?
To detect a rise in aided awareness from 20% to 25% with 95% confidence and 80% power you need about 1,100 people per group. A 10-point lift needs about 290 per group; a 2-point lift needs about 3,800. If you cannot reach the sample, use search lift and a holdout instead.
Can I measure brand lift for organic influencer posts?
Only roughly. You cannot know who saw an organic post, so you cannot form a clean control group. Self-reported exposure is unreliable, and followers already like the brand more. Partnership ads, regional creators with geo-concentrated audiences or staggered waves give better evidence. See our creator ads service.
What is search lift in influencer marketing?
Search lift is the rise in people searching for your brand by name during or after a campaign, compared with a baseline. Read it in Search Console for your own site's branded queries, and adjust for seasonality with a control search term. It is the cheapest lift measure, but other marketing can move it too.
Is Google Trends enough to prove brand lift?
No. Google Trends shows relative interest on a 0 to 100 scale where 100 is the peak in the period you picked, so it shows the shape of a rise, not its size. Use it to spot spikes after posts and pair it with Search Console clicks, a control term and ideally a survey.
How much does a brand lift study cost?
We do not quote a figure, because we found no published India price and platform minimum budgets change. Ask your ad platform for the current minimum spend and audience size for a lift study on partnership ads, and price a survey from the sample sizes in this guide.