Archive for ‘Analytics’

4 Ways to Optimize for Lifetime Value Using Segmentation

By July 3rd, 2014

In the previous post we discussed how to segment your customer base for lifetime value (CLV). In our fourth and final post of this lifetime value series, we’ll provide some recommendations for optimizing for CLV.

Once you have identified audience segments, you’ll want to investigate things like:

  • Where are these people? What marketing channels do they cross?
  • What messaging or value propositions are relevant to this segment?
  • What are different ways that you can reach a customer within a particular segment?

The idea of optimizing for CLV is to achieve increasing profits from your existing customer base, or find new high-value customers.

Be channel agnostic when looking for marketing opportunities

When asking these questions, it’s useful to take a channel agnostic approach to your thinking. Instead of relying on data from one specific channel, you’ll want to incorporate data signals across different marketing channels like search intent, behavioral data, and audience characteristics, in addition to your own data, in order to paint a detailed picture of your high value segments and develop strategies to increase your customer CLV.

Once you have a good understanding of your high-value customers, you can start developing strategies to drive higher CLV. A few potential scenarios follow below:

Scenario 1 – Find new high CLV customers

A financial company could improve its acquisition efforts by refocusing its spend towards an acquisition channel that has proven to attract high CLV customers. Alternatively, it could also expand its reach by leveraging 2nd and 3rd party data to create look-alike audiences for additional targeting opportunities.

Scenario 2 – Reduce customer churn

An insurance company creates an audience segment for current customers with contracts expiring within the next 30 days in order to run a search retargeting campaign that ensures the brand has top that the brand is well represented when the user conduct relevant searches.

Scenario 3 – Increase repurchase rates

An online clothing retailer rolling out a new seasonal line could remarket its product ads on Facebook to customers who haven’t bought from the company within the past 6 months.

Scenario 4 – Cross-promote or upsell to potential customers

A home improvement retailer could look for opportunities to cross-sell to potential customers who’ve recently shown interest in home loans. Alternatively, the retailer could promote upsell opportunities to sell kitchen upgrades to customers who’ve recently indicated interest in or purchased a new appliance.

The possibilities are numerous. By taking basic customer lifetime value information, developing useful segments, and connecting that information with the customer data that you’ve collected or have access to, you can open up a number of new, interesting ways to make smarter marketing decisions and increase profits.

Looking for additional posts in this series?
3 Reasons Why Segmentation is Necessary to Understand Customer Lifetime Value
Gathering Customer Lifetime Value Data – Start Small and Build
Using Lifetime Value to Create Audience Segments

Using Lifetime Value to Create Audience Segments

By June 27th, 2014

Now that you have your data in order, you’ll want to start parceling out relevant customer segments.

The goal is to identify segments that are practical and helps you make better marketing decisions. You’ll want to strike a balance between segments granular enough to execute targeting strategies, but broad enough to scale your targeted marketing efficiently.

There’s no shortage of advanced statistical models to determine customer segments, but in this post we’ll provide a few straightforward, practical recommendations.

Start segmenting based on customer value.

The 80/20 rule, which states that 80% of your profitability comes from 20% of your customers, is a good place to start your segmentation effort. You might actually find your customer profitability distribution is even more extreme with over 100% of your profits coming from your top customers, and your worst customers actually taking away from your bottom line. However, at the most basic level, the distribution might look like the above graph.

In this case, we’ve identified high CLV, medium CLV and low CLV segments. However, these segments are still too broad for much use. From here, you’ll need to rely on your data to help unearth demographic or behavioral commonalities within each segment.

Ideally, you’ll want to try to identify segments with commonalities that you can act upon and measure changes in performance.

For example, you might find that:

  • Customers of a particular demographic or geography might have different CLV characteristics
  • Certain acquisition channels may be more effective at driving higher CLV customers
  • A customer segment may show high loyalty, but lower CLV, suggesting opportunities to drive increased business

Once you’ve identified different customer segments, you can identify opportunities to optimize for your higher CLV segments. We’ll explore some ideas to optimize for lifetime value in our next post.

Gathering Customer Lifetime Value Data – Start Small and Build

By June 23rd, 2014

It’s important to segment your customer data to gain useful insights, but the relevant segments may not become apparent until you start wading through the financial data. In this post, we’ll discuss three important financial factors you’ll need for your customer lifetime value (CLV) calculations.

There are three important things you’ll need to find:

1. Retention rate/Customer lifetime
2. Revenue per customer segment
3. Margins per customer segment/sale

The latter two are pretty straightforward, and don’t leave much room for interpretation, so in this post we’ll focus on divining the retention rate. (One caveat on revenue – businesses with a brick and-mortar presence should also make sure to account for the “research online, buy offline” effect when calculating CLV. Otherwise,customers acquired via online channels will be undervalued.)

Customer “lifetime” doesn’t mean forever

“How long is a customer’s lifetime?” is a pretty common question. The “lifetime” in lifetime value isn’t quite so literal. You’re not expected to be projecting profits for the 75 years or so of a customer’s life. For one, that’s impractical. But more pressingly, the further out the revenue projections, the less reliable they are. Generally, we would recommend calculating your CLV within a five-year window (although some exceptions apply).

There are a number of different ways to figure out your customer’s lifetime. Two of the more common methods involve using your retention rate or your repeat purchase rate.

Scenario 1: Use retention rate

If you charge your customers a monthly or yearly retainer or subscription fee, you should know your monthly or yearly churn rates (retention rate = 1 – churn rate).

For example, an enterprise SaaS company with a yearly retention rate of 90% would have an expected customer lifetime of 10 years based on the equation below:

CLV formula retention rate




Scenario 2: Use repeat purchase rate

Repeat purchase rates are very similar to retention rates, except they are more relevant for companies whose customers have irregular purchase cycles. For example, a retailer with repeat purchase rates of 50% will have different customer lifetimes depending on whether their customer buys from them every 3 months, or every 3 years.

Scenario 3: You know neither

Some businesses may not know what their repeat purchase or retention rates are. In those cases, you’ll have to back into the number. You can do this by specifying a gap of inactivity where you decide a customer is no longer a customer, and then measuring the time period between the customer’s first and last purchase.

The period of inactivity in which you determine a customer has lapsed depends on the nature of your business. For example, a grocery store whose best customers buy weekly might determine that a 3-month dry spell is sufficient enough to indicate that they are no longer customers. Other businesses – auto dealerships for example – will have much longer lag between purchases.

Don’t forget about your segments

Regardless of how you calculate your customer’s lifetime, remember from our previous post that you’ll want to find the customer lifetime for each relevant segment you’re measuring, not just the total average customer lifetime.

This is part of a 4-part blog series. Missed the first post? Read 3 Reasons Why Segmentation is Necessary to Understand Customer Lifetime Value here. Our next post will dive deeper into how you can carve out relevant customer segments, so stay tuned!

3 Reasons Why Segmentation is Necessary to Understand Customer Lifetime Value

By June 19th, 2014

Marketing metrics like CPC or CPA only capture the cost part of the profit equation. This seems a bit odd when you think about how often successful marketing campaigns are judged on the revenue portion of the equation. It’s also detrimental over the long-term because it treats your marketing efforts as a cost-center vs. a revenue center.

Incorporating customer lifetime value (CLV) is important because it takes both revenue and costs into account. Our recent white paper provided an introduction on how to introduce customer lifetime value into your online marketing campaigns. This post is the first of a 4-part series that will provide pragmatic recommendations for building lifetime value models.

Prelude – Get your data ready

A lifetime value model is only as good as the quality of the data you keep. To develop a good CLV model, you’ll need to ensure you’re accurately measuring things like revenue per customer, margin per sale, and retention/churn or repeat purchase rates.

Once you have your data, there are three good reasons why you need to segment your customers to get an accurate understanding of your customer lifetime value.

Reason 1: Average Revenue Per User (ARPU) is incomplete

A simple example can illustrate. Let’s say a business has two customers:

Customer Lifetime Value example

Customer #1 is an unmarried, 25+ y/o librarian, $35K HHI, and buys a hatchback for $15K. Customer #2 is a married, 55 y/o executive with 3 kids, $250K HHI, and bought a sports car for $160K.

Based on this data, the average customer for this business buys $80K cars. Except in reality that $80K car is far outside the price range of Customer #1, and it might not be upscale enough for Customer #2. Furthermore, the average customer’s demographic info based on the details above is basically worthless.

This is why segmentation matters. ARPU tells you that you have paying customers, segmentation can tell you who’s actually driving your business.

Reason 2: Focus your effort on the segments matter most

Go back to the two customers above. Which segment appears to be more valuable to you? Which is the one you want to focus your marketing on?

That’s a trick question. The answer is it depends. If you can convert the hatchback buyer and she buys several cars from you over her lifetime, then she could be worth more than the sportscar buyer. But if you were focusing on a window of 3-5 years, then you’d probably want to focus your marketing efforts on other sportscar buyers.

Either way, you can’t make this decision until you know how different your customers are.

Reason 3: Make better decisions

You might choose to target both customers. And that’s where segmenting for CLVcan be most helpful. Once you can divulge a segment’s CLV, opportunities for acquisition, upsell, and cross-promotion may become more visible. For example, you might leverage retargeting campaigns to upsell to segments with high CLV. Alternatively, you may find that there are cross-promotion opportunities with a segment with lower CLV but potential for growth. Finally, you might decide to cut marketing in channels that consistently acquire customers with low CLV.

Ultimately, understanding your customers’ CLV enable you to be a better marketer. This is a 4-part blog series, so stay tuned for more best practices to follow! [Part 2: Gathering Customer Lifetime Value Data – Start Small and Build]

Why Marketers Need to Think More Like CIOs

By September 25th, 2013

Marin Software PPC SEM analytics data optimizeIt’s no secret that, due to the measurability of online advertising, marketers now have to face an avalanche of marketing data. To turn all this rich data into valuable insights which can then drive revenue, you have two options:

  1. Hire a small army of analysts to manually analyse this high volume of data, and turn it into actions which optimise revenue outcomes.
  2. Use technology to analyse the data, and turn it into actions which optimise revenue outcomes.

Hiring an army of analysts to manually digest data is an expensive and resource intensive approach, so unsurprisingly most marketers are choosing to go down the technology route. This way they can use technology to digest their data and analysts to think more strategically, adding an important layer of human insight.

This shift towards technology is evidenced by a recent Gartner statistic which claims that by 2017 the CMO will spend more on technology than the CIO. This highlights that marketers are now beginning to build out technology architectures to cope with their avalanche of marketing data. However, most CMOs and marketers in general have very little experience with technology implementations. We thought we’d share a few best practices to help marketers who are building out their technology architecture, based on our experience of hundreds of marketing technology implementations for enterprise class advertisers around the world:

1) Don’t hoard data, do something with it. I spoke a lot about this in my previous post “How to Avoid Being a Data Squirrel.” The key is to get the correct analytics, attribution and optimisation layers in your technology architecture and distribute the results across the enterprise. This way you can ensure you’re turning your analytics and attribution data into actions that optimise revenue outcomes.

2) Your technology layers don’t all need the same logo. Just because you wear Nike trainers, doesn’t mean you drink Nike beer, use Nike cleaning products or use a Nike tablet. Focus on finding best of breed technologies in each area of analytics, attribution and optimisation, then integrate them. The enterprise technology space is increasingly open, and pioneered this by bossing the CRM space, whilst integrating data from best-of-breed technologies in other areas through the App Exchange.

3) Ensure technology saves your team time. Ultimately, if technology doesn’t save you time then you may as well just hire an army of analysts. Look for technology which integrates data that would otherwise be done manually, creates more efficient workflows, and makes the management and optimisation of campaigns more efficient.

4) Look for third-party technology. When it comes to the layer of technology responsible for attributing media spend, it’s important to look for third-party technologies independent from media owners. If you allow a media owner’s technology to attribute spend then there are potential issues of bias involved towards their own media channels. In essence, it’s a bit like letting an estate agent decide how much you’re going to pay to buy a house.

5) Optimise to revenue, not just leads. Revenue is what drives your business and increasingly marketers are becoming accountable for revenue. When building your architecture ensure that your technologies are optimising towards bottom-line revenue for your business, not just form fills, quotes or other softer lead targets. This can be achieved by integrating marketing programs with CRM or ERP data, offline conversions over the phone and in-store or integration with proprietary data in other areas of your business. This bottom-line marketing optimisation is what will satisfy your CEO and CFO that the marketing technology investment is worthwhile.

We will be talking in more depth about these tips and best practices to help marketers build effective technology architectures at a number of events across Europe before the end of this year. So if you’re going to be at Ecommerce Paris, iStrategy London, Search Congress Amsterdam, Ecounsultancy JUMP, SMX Stockholm or our own Marin Masters conference in London at the beginning of November then be sure to join the conversation in person. Otherwise, feel free to join the conversation below.

How to Avoid Being a Data Squirrel

By August 21st, 2013

analytics data optimize

With the amount of data now available to digital marketers, they’re in danger of becoming data squirrels that hoard data but do nothing with it. There aren’t enough analysts in the world or hours in the day to manually analyse all the available data, and crucially, turn it into actions which optimise revenue outcomes.

The modern day digital marketer needs to consider how to turn all the data they have into actions which optimise revenue outcomes through advertising technology. To do this they need to understand the three layers of advertising technology:

  • Analytics
  • Attribution
  • Optimisation

What you don’t measure, you can’t manage. Analytics platforms gave marketers the tools they needed to finally end the age-old problem of not knowing which half of their advertising is working and which half isn’t. Measurement is crucial, and it’s exactly why digital marketing has had the sensational growth rate it has.

However, whilst analytics is vital, when in a silo it can’t help marketers understand how and when customers interact with multiple channels. Furthermore, it doesn’t interpret the data into campaign adjustments which optimise revenue outcomes.

Whilst analytics platforms have gathered data for marketers, the attribution technology layer is what helps them make sense of that data. Attribution has been a hot topic for marketers over the last few years as they’ve looked to understand their consumers’ path-to-conversion, and how to better allocate marketing budget and tailor messaging as a result.

However, whilst attribution platforms make recommendations on how to better allocate spend, they don’t implement these changes. For example, attribution platforms do not translate their output into bid recommendations for particular keywords on search engines or banners on display networks.

The final layer of digital marketing technology is the optimisation layer. These technologies take the data from the previous two layers and turn them into actions that optimise revenue outcomes for the marketer. They drive incremental performance gains and time savings.

However, these platforms would struggle to do anything without the data provided by the previous two layers of digital marketing technology. Data is the lifeblood of the marketing technology architecture, and without it the key final layer of optimisation couldn’t happen.

Final thoughts
It’s worth considering that not every layer in digital marketing technology needs the same label on it. It’s a bit like choosing shampoo and conditioner; just because you prefer shampoo from Brand A doesn’t mean you can’t buy conditioner from Brand B if you prefer. The same applies to marketing technology, focus on finding the best technology in each layer. Today’s enterprise technology world is open and connected, best-of-breed technologies can facilitate data flow between themselves seamlessly.

Introducing the BoostCTR Account Performance Grader – Optimize Your Campaigns Today

By April 2nd, 2013

BoostCTR Account Performance GraderFor many search marketers, identifying opportunities for optimization within paid search campaigns is challenging. Monitoring and maintaining top performing ad groups, keywords, and ads is a standard best practice; but as campaigns grow, keyword lists expand, and creative tests multiply, this approach fails to scale and provide incremental improvements in paid search performance. With so many optimization opportunities hidden in an ocean of data, how can search marketers give the required attention each campaign deserves? Where do you even start?

To help search marketers answer these questions, Marin Software is thrilled to announce our partnership with BoostCTR to offer a free paid search diagnostics tool that not only provides insight into account performance, but also opportunities for optimization. The Account Performance Grader is designed to analyze historical performance across keywords, ads, quality scores, and ad groups for AdWords and Bing Ads campaigns. Simply sign up and enter the required information to receive your customized report.

Among other best practice recommendations, this report will provide actionable insights for pausing poor performing keywords and ads, as well as reveal quality score trends that identify areas where keyword relevance can be improved. With the Account Performance Grader, search marketers can remove the guesswork out of campaign optimization and focus their time on more strategic, high impact tasks.

Sign up here and start optimizing your campaigns today!

GRP = Get Ready (for another online) Performance (metric)

By September 27th, 2012

I began my love of data-driven marketing nearly a decade ago when I started at The Nielsen Company. While my time there was limited to the Consumer Packaged Goods and Telecommunications industries, I was hard pressed to get away from the heart of the Nielsen business—or at least what they were best known for—their television ratings. That’s where I started to become familiar with terms like Gross Rating Point (GRP), which is:

“A unit of measurement of audience size. It is used to measure the exposure to one or more programs or commercials, without regard to multiple exposures of the same advertising to individuals. One GRP = 1% of TV households.” (Source: Nielsen Media Research)

Gross Rating Point (GRP)

GRP is the foundation by which media buyers compare the advertising strength of various media vehicles. So why should digital marketers care? Nielsen, in addition to other companies like Comscore, wants to give marketers new GRP-like metrics by which to measure the effectiveness of their advertising efforts across channels (TV and online).

Aside from providing a single lens for viewing performance across platforms, a GRP-type metric would also lend itself to informing advertisers on how much they would be willing to pay for certain digital media impressions. This could change the way advertisers currently manage their online bidding—only paying for those impressions that they feel will be most valuable to their business. The end goal would be to obtain the highest possible GRPs at the lowest possible cost, while remaining focused on the target market—all of this now being done across both TV and online channels.

As with any foray into new metrics and crossing the chasm of advertising channels, there are pros and cons to the idea of using GRPs. Critics have argued that GRPs are not a guarantee, but rather an estimation of the audience that could be reached and, therefore, aren’t the best gauge for what media channels are the most effective. On the other hand, this is one of the first efforts to bring TV and online channels together and I applaud the effort. I believe this is an inevitable step in the evolution of advertising and will continue to be a focus for marketers as they continue to maximize budgets, refine their advertising and hone in on high-value customers.

While these digital GRP metrics are relegated to mostly display advertising channels at this time, who’s to say search isn’t far behind? With search retargeting now becoming a reality, a search GRP system could be on the horizon as well.

Compelling Trends from Marin’s 2012 Q1 Report

By April 13th, 2012

Marin is proud to announce the release of our 2012 Q1 online advertising report. This report, which identifies significant year-over-year paid search trends, was compiled using data from over 1,500 advertisers and agencies who invest over $3.5 billion annually in online advertising through Marin.

At a glance, our study revealed an increase in click-through-rate (CTR), with cost-per-click (CPC) remained relatively steady. More specifically, we found a significant increase in CTR and a drop in CPC on Google. Some of our key findings include:

  • 46% increase in Google click volume
  • 14% increase in CTR on Google
  • 4% increase in the share of clicks coming from Exact match

Q1 2012 Industry Click Through Rates










So what does all this mean? The increase in CTR coupled with a 12% lower CPC points to Marin users increasing their efficiency on Google. This finding is further validated by the increased usage of exact and phrase match type keywords, as users continue to identify and fill gaps using Marin’s keyword expansion tools.

Q1 2012 Click Share by Device











Device targeting, specifically smart phones and tablets, continues to soar in popularity. Increases in click volume give evidence of the growth in consumer adoption. With smart phones and tablets showing higher CTRs and lower CPCs compared to desktops, mobile search should continue to be top of mind for advertisers.

Want to see other Q1 industry trends from 2012 with our recommendations? Download the full report here.

People Love to Search on their Smartphones and Tablets

By March 30th, 2012

We love our mobile devices, and according to our recent study of mobile paid search, we love searching on them. In looking across our client base the trend was unanimous, mobile search is up, way up.

In the U.S., we saw ad clicks from mobile devices increase 132% during 2011, and by the end of this year mobile will comprise 25% of all paid search clicks. Similarly, in the UK mobile ended the year with 15% of all clicks in the UK. And, even though it’s not as significant a percentage, mobile clicks in the Eurozone more than doubled in 2011.

Things get even more interesting for marketers when looking at the differences between smartphones, tablets, and desktops. Generally (UK was the sole exception), smartphones carry higher CTRs and lower CPCs, but the lowest conversion rates. Tablets beat desktops in CTR and CPC, come close to trumping desktops in conversion rate, and edge all devices out in cost per conversion.

So, what’s this all mean?

Mobile devices are not only changing the way consumers search and shop, but how marketers advertise. The immediate response by advertisers is to devote more budget to mobile search (we project ad budgets will fall just a bit short of click volume in 2012). However, down the road as savvy marketers adapt to mobile search scenarios, click to call, location-based promos, and integration with social will all become common place. Furthermore, attribution becomes a much larger issue, particularly in a scenario where a mobile search directly leads to an in-store sale. Who gets the credit?

How do you foresee search marketing changing with the increased adoption and use of smartphones and tablets?


Find us on Facebook