The deadpool is full of failed online travel startups – it’s a crazy idea to even think about starting a new business, right?

The sector is dominated by the Priceline and Expedia groups which spend north of $3 billion dollars per year each in online advertising

Google’s flight and hotel search are increasingly taking up space on the search results page and phones, and there is intense competition from other innovating startups (Airbnb, Skyscanner, Hopper, etc).

NB: This is an analysis by Luc Guilhamon, co-founder and vice president of marketing at Viajala.

What’s more, the metasearch model comes at a significant monetization disadvantage relative to booking sites.

So when we started Viajala in 2013, raising the money needed to go head to head against this level of competition was not an option.

As a result, we decided to focus on meticulous execution in online marketing, which would enable us to be more efficient than the competition, and obtain a disproportionately larger volume of traffic.

We set out to build scalable, software and data-driven marketing campaigns, and we now have over two million users per month, and operate profitably.

A quick disclaimer: this column is based on our own experience and you may not find the same outcome should you apply the same practices to your business.

Different business models, markets, brands can lead to very different outcomes.

Our objective is to present you with online marketing practices ideas, not to give you a blueprint for your own traffic acquisition.

Firstly, we ensure that we work only on projects that add significant value, and that we can handle with our limited resources. We only work on marketing practices for which the answer to the following questions is “yes”:

• Can this impact our KPIs meaningfully? For example, we do paid marketing only on Adwords and Facebook, other networks and ad formats are too small (note we operate in Latam and Spain).
• Is it scalable / can we automate this? For example, we do little optimization for seasonal ads, as these are ephemeral and difficult to automate.

These questions are the filter to all our initiatives and tests.

In terms of resources, we have had only three people working on our marketing campaigns at any point in time:

• One advertiser to run the campaigns and write the tools requirements
• One data scientist to create the models
• One developer who builds the tools.

This small scale was initially due to our low budget, but in hindsight we wouldn’t want it any other way.

When it comes to software teams, we find small is beautiful, and it’s better to have a few focused people rather than invest in a large team, as the distraction from management is too expensive.

This concentrates all of the knowledge on just a few people, which makes us quick and flexible.

Adwords is the advertising platform of focus of our paid marketing strategy for the volume opportunity it offers, and its countless optimization options.

Our approach from the beginning was to build an in-house SEM platform that would allow us to automate our campaigns.

Below are some of the solutions we implemented.

1) Fully automatic keyword management

Travel SEM is all about the long tail – especially when you cannot afford the same CPCs as your competitors, so we use Google’s targeting ideas service to identify new keywords at scale.

We input the keywords we currently bid on, collect the keyword suggestions from Google’s API and identify which suggestions are relevant by comparing them to our inventory.

For example, if a keyword suggestion contains terms that are relevant to our inventory – such as “flights” and a destination name such as “new york”, we would automatically upload it.

However, we exclude suggestions that contain terms we identified as irrelevant – for example “crash” or “delay”.

We then automatically add the relevant keywords to our accounts.

This solution enables us to find the newest relevant keywords, but also limits the risk of unnecessarily bloating our accounts with useless keywords, since these suggestions have been searched for at some point on Google.

It is worth noting we also treat search terms we are broad matched on as we do the keyword suggestions, and they follow the same vetting and upload processes.

2. Precise adcopy

We try to take full advantage of ad customizers, as this is a great way to have ads that are easily applicable across adgroups, whilst being very targeted.

For example, many of our ads include both dynamic price and destination names:

• Flights to {destination_name}
• From only {destination_price}
• Search and compare flights on Viajala to find the best deals

We also use ad extensions as much as possible. For example, when users are searching for ‘flights to Barcelona’ in Spain, we show price ad extensions for the most popular routes:

• Flights from Madrid – €99
• Flights from Valencia – €49
• Flights from Sevilla – €139

Again, this is fully automated, so our ads are highly targeted across all of our 7 markets, and the only investment is building the tool.

We run marketing profitably, which ensures we don’t rely on fundraising to afford advertising.

In order to maximize profit, we designed a tool that uses a random forest machine learning algorithm that calculates the most profitable bid for every keyword, and subsequently automatically updates the keyword’s bid.

This maximizes the financial performance of our campaigns, and since it’s fully automated it saves a lot of time.

Furthermore, in order to maximize bidding accuracy, we duplicated our adgroups by device and search remarketing audience.

This goes against what Google usually recommends but enables more accurate bids for each audience.

Social media

Facebook is another important source of marketing traffic for us as it enables us to drive large volumes of qualified traffic in a scalable manner.

For lack of resources we haven’t invested as much effort optimizing our Facebook campaigns as we did on Adwords, but we have found the following targeting methods work well:

• Dynamic remarketing – this is the highest quality traffic we obtain from Facebook, and improve our new users’ lifetime value as we retarget them after their first visit.
• Lookalike targeting: Facebook excels at identifying new users who are interested in our service. We build and target ads at a lookalike audience based on our own users, and this enables us to advertise to a large audience of new users which monetizes well. In our experience, the quality of the traffic from that audience is much better with Facebook than with other competing advertising platforms.

SEO is not dead – far from it

Despite the fact Google has pushed the traditional organic results down the search results page over the years, SEO still drives a lot of great traffic.

Today, it is our largest traffic source in terms of profit – and we continue to invest a lot of efforts to grow it. Our approach has been to focus on three main areas:

• Technology: focus on site speed, usability and crawlability.
• Content: We hired a team of content managers to produce unique, high quality content for our site.
• PR: We told the press our story – a small team of local entrepreneurs who want to fill a gap in the Latam flight market – and obtained great coverage from the main local publications.

Flight and hotel booking often is a long decision process, so we put in a lot of effort to bring users back to our site. We do that by sending users price updates for their previous searches in a number of ways:

• The aforementioned remarketing campaigns on Facebook; and we also implemented something similar on Google’s display network.
• We send price updates to users’ devices based on their previous searches via Chrome push notifications. This is a simple, yet very effective channel for us, thanks to Chrome’s large user base.
• Email marketing: we maintain a database of subscribers who are highly engaged in order to ensure a high open rate and good deliverability with ISPs. We do this by removing users who haven’t opened our emails after just 4 weeks. The content of the emails is personalized based on the user’s previous behaviour on our site to maximize engagement.

Not all plain sailing

This column so far has been a list of successful marketing practices – but we have had our fair share of failures too along the way.

These are equally interesting to learn from, and we have shared a couple of these.

We tested a third party remarketing network instead of directly advertising with Adwords and Facebook – the objective was to improve scale and optimize our ad serving practices to every user across the various publishers.

This network turned out to require much higher CPCs for a lower volume of traffic – not a great recipe. We also found they offered much less control than Facebook and Adwords, as their user interface was basic.

Overall this was a terrible test – and was part of the reason we subsequently decided to strengthen our commitment to focus on only two advertising platforms.

Our strength lies in direct marketing, but we did A/B test branding campaigns.

We showed video ads on Youtube and Facebook in a specific region – and measured the impact in direct traffic and brand searches for that region.

Whilst we did see a positive impact to our KPIs, we also found that we made $0.20 revenue for every $1 we spent, even at $0.01 cost per view.

We believe that branding campaigns can return positive results over time, however it isn’t the right initiative for us at the moment.

Ultimately, all these learnings, both successful and not, are a result of our testing approach.

We avoid making assumptions whenever possible, and test ideas in the most effortless possible way to validate its value.

What’s more, the tests often also help us better understand how best to implement the idea, so it also improves the final product.

NB: This is an analysis by Luc Guilhamon, co-founder and vice president of marketing at Viajala.