Results from 13 paid traffic tests [Case Study]

One of the things I’m trying to do with Informly is work out a scalable way to grow and ideally build some early revenue. I see content marketing as the best way I can grow my business long term but it takes a long time to build up momentum.

So as part of marketing for the agency version of Informly I have tested a number of paid traffic strategies and made a bunch of estimates to work out which ones might be suitable ways to grow. First let’s start with everything you need in place before you start messing around with paid traffic.

Note 1 – I should preface this by saying I’m not a paid traffic expert so a lot of aspects of this can obviously be improved and take what you want away from my results. 

Note 2 – If you want to be spared all of the gory details skip to the end and check out the takeaway section.

Basic analytics tracking

300x250_adroll

Because most of these strategies enable me to use trackable links I used the Google URL Builder to create trackable links for each campaign. An example might be as follows.

The right for example would be placed on a site like clientsfromhell.net and I would create a URL with the Google URL Builder that enables me to track traffic just from that ad.

In some cases I’d use more than one ad from each traffic source and have different trackable links for each campaign.

You can learn the specific steps to this in my guide on The practical guide to actionable Google Analytics.

Estimates of Lifetime Value (LTV) and Cost Per Acquisition (CPA)

Unless you have a bucket of money to throw at advertising then you probably want some idea of whether the advertising is bringing in customers for less than they are worth long term. If you want something that scales ideally you want to acquire customers for significantly less than they are worth to you long term. Some say up to 3x is a good result. In my case I was aiming at least to have CPA lower than LTV and I felt if I could achieve that then I would be able to further optimize down the track to improve it over time.

There are a few things you are going to have to estimate to work this out.

Lifetime Value of a Customer (LTV)

To know whether the advertising is worth investing in you need an estimate of customer lifetime value. In my case my plans range from $19 – $75 / month but I can assume that people will stick around for more than one month. Because I haven’t been around long enough to know how long they will stick around I have to make some estimates. Here are the calculations I used:

  • Average plan value – $49 / month
  • Fixed costs per customer – $9 / month
  • ROI period – 24 months
  • 2 yearly churn – 20%

I constantly review these estimates as I get more and more real data to estimate my lifetime value which based on the numbers above is:

(Average plan value – Fixed cost per customer)* ROI Period*(1-Churn)

or with real values

($49-$9)*24*(.8) = $768

Again I can’t stress enough that this is an estimate but it’s important to have some idea of this value before you start spending money to acquire customers.

Cost Per Acquisition (CPA or CAC)

Analytics and the paid traffic provider will tell you how much each click costs you (even if you pay CPM you can still look at how much traffic you got and divide it by the total cost to work out CPC). However there are a few steps between someone visiting your site and becoming a customer.

And in addition to that issue you may not have enough signups from a particular source to use real data for the calculations so you might have to make some estimates.

I use our free CPA calculator to work out how much a customer costs based on how much I’m paying per click.

I’ve embedded our calculator below but you can visit the CPA calculator here and even embed it on your own site if you want.

Let’s use Facebook traffic as an example. In the calculator below enter $1.81 for the CPC,  1.7% for the conversion of visit to free trial (Step 2) and 3% for the 3rd step which is my estimate of trial to paid. It will come out at $3,549!

Here are the results you should expect (bar a few dollars for decimal place differences).

facebook_traffic

CPA Calculation

The calculation going on above is as follows:

Cost per trial = (Cost per visit * visits) / Trial signups)

CPA – Cost per trial * trial to paid conversion estimate (3%)

Notes

I had a rough idea that 3% of people who sign up for a trial would convert to paid (a combination of assumptions and real data). So if I didn’t have enough signups from a particular source (i.e. above 5 signups is not enough people to get real data) then I could just estimate that 3% of them would become paid.

You can do the same thing if you don’t have enough traffic from a source. For example if I only had say 50 visits from Facebook that may not be enough to test the visitor to trial signup conversion rate so I could estimate that to get the value for ‘Sign ups’. I would be very careful with this estimate though since the conversion rate will vary considerably from source to source. The best case would be to make sure you have just enough traffic that you can draw some conclusions.

So the example above shows you how quickly spending $1.80 can turn into a traffic strategy that will bankrupt your business. I can’t acquire customers for over $3,500 when they are only worth $768 to me.

If you have any questions about these calculations please feel free to post them in the comments below.

Sources tested and results

So with the tracking set up I tested a bunch of different paid traffic strategies (over the last few months). Here are the high level results with the explanation of key points addressed in the notes below.

paid_traffic_results

 Notes:

  • The twitter auto follow is by far the best cost wise. For this one I’ve estimated the costs to be $150 which is basically my time to set it up. This strategy isn’t scalable however so it just runs at a small scale in the background. I have written a detailed guide on this strategy if you want to learn more see our Twitter auto follow guide
  • Our cold email strategy involves my team building lists of web design companies to email and ask them if they want to check out our app. So far this seems like the only scalable paid way of generating traffic I have been able to uncover so we are continuing to experiment with this.
  • Adroll retargeting is very effective but it can’t be scaled without getting significantly more traffic to my site to start with. We only deploy the Adroll code on our agency pages which is only a small amount of the traffic on the site. You can learn ore about this strategy in our Retargeting guide.
  • Most of the paid traffic was set up using BuySellAds.com, some of it was banner ads and others was paying for ads in email newsletters, paying for tweets etc.
  • Outbrain.com shows related stories on large blogs like Mashable. It’s cheap traffic but converts very low since they are more interested in reading articles than buying product. However it could work well in conjunction with Adroll which I will test further.
  • In some cases you can see there are no conversions to trials like with the Designer Depot item (which is a tweet by the way). In cases like that I took my average visitor to sign up conversion rate of 8% and applied it. This obviously gives you a much higher number of conversions than actually occurred but in any case the CPA is too high to act on. If you didn’t want to take this approach you could simply change the 0 conversions to 1 just to see how bad the CPA would be had you had one person sign up!

Takeaways

Since this stuff can ge a bit complicated I thought I’d list out the things I think you can take away from this post. Again I stress I’m not a paid traffic expert but if you are a founder like me and you wanted to look at paid traffic then hopefully the points below will give you some guidance.

  • Don’t ever do paid traffic if you don’t have estimates for your key metrics. $1.80 can very easily turn into $3,500 to acquire a customer without you realizing it. 
  • Banner ads in general didn’t perform well enough in my case. There could be a lot of reasons for this but I think I’ve tested enough to not want to pursue the strategy in general until I’m more established and I am certain about my metrics and the funnel is well oiled. 1 or 2% shifts here and there could change negative ROI into positive ROI very quickly. I as hoping to find a few easy sources of cheap leads that I could easily scale but that was not as easy as I hoped.
  • ‘Paid for tweets’ generally performed far worse than other paid strategies. I will be avoiding those in future.
  • Retargeting is very effective. Check out our retargeting guide for more on this.
  • Cold emails are surprisingly effective. I will continue to test and refine this strategy.

If you have any questions please feel free to leave them below.

 

About

Dan Norris is a co-founder at WP Curve and a passionate entrepreneur with an obsession for content marketing.

27 responses to “Results from 13 paid traffic tests [Case Study]”

  1. Jake Hower says:

    Good wrap up. Incredible how some of the strategies can work out so expensive per acquisition. Cold emails look like a great place to start.. You should zoom in a little on your strategy with them 🙂

  2. Were your banner ads Direct Response ads or usual Brand Building stuff? Good to see good results from retargeting. Collecting my list of people there too with AdRoll.

  3. I’m wondering Dan what the results would be if instead of pursuing the conversion to trial member immediately, you instead were pursuing a conversion to subscriber that was then nurtured by email for a time and then with a relationship in place offered the trial. What’re you’re thoughts?

  4. Dan Norris says:

    Thanks Jake. I think if a lot of people took the time to estimate the acquisition in this way they would find that advertising is generally very expensive. It’s just a lot of people don’t realise.

  5. Dan Norris says:

    Hey Adrijus the banner ad example above is a real example, they were all along those lines.

  6. Dan Norris says:

    Hey Scott, yes I tried this as well. The problem with this is it ads another step into the funnel and if you look at the calculator above adding one more step and yet another % conversion significantly diminishes the outcome.

    Getting the email opt in increases the conversion rates slightly but then only a small amount of those opt ins end up signing up for an account.

    I think as a long term organic strategy it probably works well but for paid traffic the dollars don’t add up.

  7. Well, that’s why they are low performers. It’s showing features, not benefits. And no direct response. When you have time, look into http://www.dankennedy.com (I think Tim Conley is a fan of him too and promotes direct response ad usage on his FA site, even some podcasts there).

    I’m sure you’d get improvement man, esp with your free reports ready!

  8. Dan Norris says:

    I’m aware of Dan Kennedy, Recurring revenue is a benefit not a feature. The example above actually includes both. Recurring revenue (benefit), automated reporting (feature). I experimented with just benefit and just feature, the results impacted a little bit on click through rate but that doesn’t really affect the ROI too much. The cost per click is reasonable either way. The onpage conversions was an issue with most of the display ads. They were well below normal on page conversions.

  9. Will Swayne says:

    Hey Dan – that’s a nifty click to lead to sale conversion calculator. I like how you’ve got the +/- buttons in place to elegantly deal with different sales processes. Nice 1.

  10. Dan Norris says:

    Thanks Will! I’ll probably build others (LTV calculator etc) when I get a spare second.

  11. I’d try smaller and simplier ”Retain ( or Gain) more clients and get more Recurring Revenue”.. or even ”Increase Your Revenues while doing little additional work”

    Writing headlines sucks tho.. hard to come up with them, I know that. I’d say yours need more to satisfy people’s selfishness..

    Best of luck!

  12. @scottdennison:disqus how exactly will you set this up because I see sense in what you are saying?

  13. @PropertyLion:disqus – what Dan is doing is going for the sale directly off of the click. The number of buyers ‘directly off the page’ is typically lower than you’d find if you used your click to build an email list and then build a relationship with the list.

    Not pound them with offers but actually take time and connect with them then make recommendations for products that will help them solve problems.

    I recently started testing an opt-in page that shares a high value tip and offers the reader 4 more tips if they subscribe. Here’s the interesting thing. There’s over 1500 words on that opt-in page but so far the conversion rate of visitor to opt-in is almost 3x what it was with a standard ‘squeeze page’.

    As I send more traffic to the page those numbers will likely drop some but I think it will produce a higher quality of lead overall.

    As to how to set it up – instead of sending paid traffic to an offer, send them to a landing page and get them to subscribe. Then nurture them for a bit before putting offers in front of them. Good luck!

  14. Interesting statistics. Thanks for sharing.

  15. Can you share the link to your landing page so that I can check it out and see how you have set it up? Thanks.

    I recently came across something like your idea in the “preemimence squeeze” page model by Andre Chaperon. http://affiliatebully.com/docs/CovertSqueezeModel.pdf

    His advise is to show tremendous value upfront to the leads before sending them to an optin form. He says this model does not lead to more leads but highly pre-qualified leads who are ready to buy from you.

  16. Dan Norris says:

    The problem is in adding an extra step to the funnel you drastically reduce the conversion rate because you are multiplying it by yet another low percentage rate. Let’s take a few examples (and I have done a little bit of testing with this but I will continue to do more).

    Funnel without email course

    * CPC $2
    * Onpage conversions 8% (best case)
    * Trial to paid 3%

    CPA – $833

    Funnel with email opt in

    * CPC $2
    * Onpage conversions 16% (let’s say you could double the conversion rate). Note when I tested this my overall conversions were actually a lot lower – the click through rate was almost half – on page conversions were better but for fewer people to the cost to get an opt in wasn’t much better.
    * Email to trial conversion 25% (this is where the problem is, it’s an additional multiplier. 25% is about a standard open rate let alone getting someone to click and then sign up. You are adding so many steps 25% is generous).
    * Trial to paid conversion 3%

    CPA – $1,667

    I’ll continue to test this more but it’s hard to see this funnel working with paid traffic. My preference is always to get them to sign up for the product if I can – that adds them to my list as well but the primary goal is to get people using the product. I’ll update this post if I get any drastically different results using this approach. I’m about to test it with a guest post.

  17. Dan Norris says:

    Thanks Darren I’m glad you found it interesting, thanks for the comment.

  18. Dan Norris says:

    Also when I looked at the people who eventually signed up for the app – going straight for the signup resulted in 3X the number. I only tested the email option on 100 people though so the sample isn’t big enough but I think going for the direct signup is probably going to work out better.

  19. Bob McCarthy says:

    Dan

    I think you should give Scott’s suggestion more consideration.

    Instead of shooting for the trial right away, offer something with a lower commitment to get the email address.

    You can offer the trial too, but prospects who visit your landing page and don’t convert aren’t prospects at all – because you don’t know who they are. By using a soft offer (like tips, or a self-help guide or anything short of the trial), you can capture more response and start emailing them.

    Yes, you’ve added another step to the sales process, but who cares? It doesn’t really cost anything – and you come away with more prospects for continued lead nurturing. If you test a number of different conversion offers, you may see a significant (not a slight) increase in conversions.

    On the trial, do you require a credit card? If not, say it and highlight in promoting the
    trial.

    This was a very interesting post Dan. Thanks for sharing your numbers. But I was wondering why you haven’t included paid search (Adwords) in this experiment.

  20. Hey Dan: Very nice study with some very interesting stats and takeaways. Thanks.

  21. Dan Norris says:

    Thanks Steve I’m glad you found it useful, thanks for stopping by to comment.

  22. Darren Thompson says:

    Hi,

    Great article, but there was a couple of phrases I did not understand…

    – What do you mean by “Twitter Auto Follow” ?
    – What is “Clients from Hell” ?

    Kind regards,

    Darren

  23. Dan Norris says:

    Hey Darren no worries mate I should have been clearer. You can check out this post to see what I’m talking about with Twitter Auto Follow http://inform.ly/twitter-auto-follow/ Clients From Hell is a website where I was running some paid ads. I bought the paid ads through buysellads.com

  24. David says:

    Hi Dan

    Thanks for the sharing of the info. I will be interested to know if you do dip your toe in the water again with banner ads and how you will look to measure success

    Cheers

    David

  25. Dan Norris says:

    No worries thanks David!

  26. Hmmm, very interesting … I’ve done about 75% of those you speak of and , Twitter is cheap & effective “If You Automate the process & keep up with your personal profile” can also attract people who you might not want on your list…” Tire kickers” …etc.

    Emails are being tested now and i’ve seen the more consistent “slightly predictable” results so far.. Good Article 🙂

  27. Dan Norris says:

    Thanks mate, keep us posted with the results.

Leave a Reply

Your email address will not be published.

WordPress problems?

Our WordPress experts have you covered.

Hyper-responsive 24/7/365 WordPress support, maintenance and small fixes.