How Lalafo turned TV advertising into a performance instrument
Anna Polishchuk
AI powered peer to peer marketplace
Lalafo co-founder, Anna Polishchuk, shared tips on how to use TV advertising as a traffic generator, calculate CTR for each TV spot and know the cost of each user generated by TV.
How it was done before
Over the last 1.5 years at Lalafo, we launched 16 TV campaigns in 7 countries in Europe, Central Asia, Southeast Asia and the Middle East. We shot 150 TV commercials featuring 12 celebrities.
We were one of the first, who started using TV advertising for an online product in the CIS. This was in 2012, for a free classifieds site called Slando in Russia. When launching our first TV campaign in Russia we started working with an advertising agency. Using traditional media indicators such as reach and frequency, the agency’s own experience and campaign objectives, the agency helped us create our first media plan. The agency was able to answer all of our questions, except one:
How many people are going to use Slando as a result of our TV campaign?
Moreover, the agency warned us that TV is primarily an all branding instrument. They advised that you shouldn’t expect an immediate increase in the number of users when you launch your TV campaign because for people to start using a new service, the campaign needs to reach an effective frequency. We had nothing in response and launched our campaign, expecting results in 6 weeks.
We couldn’t be more pleasantly surprised when on the first day of the campaign, we saw huge spikes in traffic within the hourly statistics. Some hours had 10 times more traffic, some — 2, and some had none. Even when smartphone penetration was low we could witness “second screen effect”. TV viewers who had a notebook, or a computer at hand, visited the site immediately during and after seeing our TV commercial.
Over the next 6 weeks of our TV campaign, we were watching spikes in traffic and came to our first conclusion: for online products, if TV works, it works immediately.
After our first TV campaign we went on to launch 5 more TV campaigns for Slando in 3 countries. Before each campaign, we analysed the previous campaigns results. Practically every time we saw that our TV campaign achieved media KPIs — frequency and reach, but didn’t always achieve business KPIs: users, new listings and purchases. We tried to understand what we can optimise to achieve better business results.
Each and every time it was very difficult to figure out what exactly worked well and what didn’t.
How we do it now
We develop an AI powered marketplace called “Lalafo” in a number of countries in Asia and Europe. In Lalafo’s first year of operations we had to launch TV campaigns in 7 countries in Central Asia, Europe, southeast Asia and Middle East.
On the very first day, after the TV campaigns started in 3 countries, we realised that “the second screen effect” for Lalafo is massive.
Firstly, we discovered that 95% of TV viewers visit Lalafo from smartphones. Secondly, in Google Analytics by the minute stats we saw our entire media plan.
Every TV spot generated an immediate spike during and several minutes after a TV spot was aired. It was the moment when we realised that this was the puzzle piece we were missing at Slando, to finally figure out what worked well and what didn’t.
Direct response analysis
The chart below shows the effectiveness of each TV spot, relative to its gross rating points (GRP) — the traffic increase per 1 GRP (left) and cost (right) during a TV campaign. Colored dots TV spots on different TV channels which allows us to compare TV channels with each other. For instance, you can see that users that come from the TV channel Lider have the lowest acquisition cost.
On the 3rd day after the TV campaigns were launched, we received our first TV monitorings from the TV channels showing the exact timing of each TV spot, channels and programs. We compared the minute stats in Google Analytics with data from TV monitoring in an Excel table. In the same table, we calculated the traffic increase during and after each TV spot. We discovered that the effect of a TV spot lasts between 5 to 8 minutes.
As we knew the cost of each TV spot, we could calculate the cost of each user that came from TV. Moreover, now we were able to compare the real effectiveness of each TV channel, TV programme, time of day, day of the week, position in advertising block, creatives and their durations. We did this looking at cost per user and the volume of users.
We’ve discovered that for Lalafo the best performing TV ads are “hard selling” — the ones that explain what a viewer needs to do right now. If we’d like to get a direct response, the creatives should be aimed at that. We found what works is funny and entertaining TV shows, TV series, news and any programs on weekends. A lot lower response we get from political shows, morning shows, music channels and inspirational TV ads.
There are a lot of similarities between online and TV advertising optimization. Using direct response analysis, we were able to calculate the CTR of each TV spot. For example, if 100,000 people had an opportunity to see a TV ad, with conversions of 2%, you can expect that 2,000 people would use Lalafo during and straight after our ad campaign aired. In our experience, the response rate could vary from 0.1% to 2%. Similarly, the online advertising response rate depends on context, creative, and viewer engagement.
Our ideal TV programme is the last episode of a popular TV series which people watch attentively, while holding a phone in their hand to share updates on social media.
The results of a direct response analysis are used to do tactical media plan optimization: it helps you identify and cut everything out, that doesn’t work and increases what works well. For Lalafo, this resulted in a 30% lower TV budget with the same business results in the number of users, new listings and purchases.
You don’t necessarily need to launch a 6 week TV campaign to understand whether or not the TV is working for your product.
As a rule, 3 days is enough. When we enter new markets we launch a TV campaign for a few days to test all the TV content. In 3 days, we get enough data on which channels, programmes / types of programmes and creatives are the most effective. Using this data we create a media plan. For the sake of data clarity, we recommend launching a test campaign on a different TV channel on different days, so there’s no intersection between TV spots. If we choose to run a test on 5 TV channels, we would do a 5 days test campaign. In some large countries you can do a test on 3 TV channels, and extrapolate data to the channels with similar audiences and engagement.
Strategic planning of new TV campaigns
When the first Lalafo TV campaigns were over, many new questions arose. For example, we wanted to know when would be the right time to launch a follow up campaign, what duration and pressure it should be, how we can detect when a TV ad wears out, which duration split works best, etc.
Most importantly: how to forecast the results of a TV campaign in absolute numbers.
We realised we could move up from the direct response and analyse the increases in business KPIs relative to media KPIs per day, per week, and the entire TV campaign. Using data gathered over a period of time, we did econometric modeling of different configurations of TV campaigns and can forecast the results of future TV campaigns in absolute numbers with an 80% accuracy.
For instance, suppose we launched in a new market, and collected the data on effectiveness of the first TV campaigns. By this time, we know the number of users, new listings and contacts that we can get per 1 GRP and costs per user, listing, contact in a particular market. We also know:
- Baseline — number of users and their actions that Lalafo would have had without TV advertising
- Long tail — the number of people that visited Lalafo not during and immediately after the TV ad, but later
- Decay rate / increase rate — rate at which the number of users is decreasing or increasing organically
- Power of advertising — a coefficient similar to CTR in online advertising that also includes long tail
- Average advertising clutter in a country
We laid out the following task to our user acquisition team: which configuration of a TV campaign is the best, so that it generates 500,000 new listings posted on Lalafo? Historical data together with other indicators are loaded into the econometric model which generates different scenarios.
This way we can model it to achieve the targeted results. The most suitable would be a TV campaign consisting of 3, 2-week flights and 1, 3-week flight with a 1 week break between flights with the same weights of 250 TRP per active week. All other TV campaign configurations, for example, 6 weeks TV campaign without breaks, a campaign with decreasing or increasing weekly pressure would be less likely to achieve targeted results, or maybe would result in higher cost per action.
The model can also show that target results are impossible to achieve unless some of the input data is changed.
For example, it may turn out that in order to achieve our target results, we’d need to increase the power of advertising. We know that the power of advertising increases if we use celebrities. In this case, the user acquisition team would need to make a decision if they would like to increase video production costs to achieve target business results.
Recipe for a successful TV campaign
- Calculate the power of the second screen effect on your product
- Analyse direct responses and optimise the media plan. Cutting everything that doesn’t work
- Use data on business KPIs increase per GRP per TV spot, per day, per week and for the entire campaign to forecast the results of future TV campaigns.
Originally published at https://www.linkedin.com on July 10, 2017.
