Tuesday, 11 July 2017
Oneflare Pty Ltd v Chernih [2016] NSWSC 1271
(13 September 2016)

Background

OneFlare is an Australian website which connects tradespeople, cleaners, and other service providers with potential customers. In 2012, in a bid to boost their site's prominence on Google search results, the founders hired an SEO consultant (Search Engine Optimisation).

What is SEO?

SEO, or Search Engine Optimisation, is a set of tactics for setting up a website, so that it shows up earlier on Google's search results (known as a site's PageRank).

While many SEO tactics are ethical, some 'black-hat' strategies are considered unfair because they manipulate Google's results in an artificial way.

To figure out a website’s importance, Google's computers analyse the content and popularity of each of the millions of sites on the Internet. Popularity is based on the links between websites: a site with lots of links pointing to it is considered important, especially if those links are from large sites.

For example, your website’s PageRank would increase if you added a blog and others linked to it. Using black-hat SEO, you could create 1,000 new websites (all empty, and all owned by you), and make them all post a link to your blog, convincing Google's computers that your site must be popular.1

Google warns against using these black-hat methods, and punishes offending sites by demoting their standing in search results (if caught).

At first, the strategy worked, and as more customers found the website through Google searches, OneFlare's popularity grew. In 2013, however, Google deemed the SEO techniques unfair and in breach of their guidelines, and removed the site from the top page of Google search results. Web traffic dropped to near zero, and revenue fell dramatically.

OneFlare’s founders sued their SEO consultant for damages for loss of revenue in the Supreme Court of New South Wales. They alleged that the consultant had breached his implied contractual and common law duties to:

  • Use reasonable care and skill in providing services to OneFlare; and
  • Warn OneFlare of the risk of harm inherent in his strategy.

OneFlare called an accounting expert on the question of damages.

The damages issue:

To quantify damages, the expert compared OneFlare’s actual revenue with what it would have been, had they not used their ‘black-hat’ SEO strategy (and Google had not removed it from search results).

One of the expert’s key assumptions was that there was a direct, linear, and causal relationship between OneFlare’s website traffic and the revenue generated. In other words, if traffic had continued to grow, revenue would have grown along with it.

To justify this, he drew a chart of OneFlare’s actual traffic along with its revenue over time, which showed a correlation between the two lines:

Simplified version of the accounting expert's chart




The accounting expert pointed to the close correlation between these two lines to support his assumption: because the red line (revenue) follows the blue line (web traffic) so closely, a change to the dark blue line would have resulted in a similar shift in the red line2.
 
Counsel for the consultant argued that this approach was inadequate, since the expert ‘had not employed any appropriate form of statistical analysis, such as regression analysis, to establish a statistically significant relationship between traffic and sales’3

The criticism raised an important point – does the chart prove that changes in web traffic were causing changes in revenue? 

The phrase ‘correlation does not imply causation’ is often used to emphasise that apparently connected trends don’t prove that one is causing the other. For example, here’s a chart of the number of people who drowned by falling into a pool and the number of films Nicolas Cage appeared in each year4


Number of people who drowned by falling into a pool versus films Nicolas Case appeared in



FOR_17-05_Chart-1.png


A more serious example involved the debate over whether cigarettes cause lung cancer. In the 1950s, before carcinogens in cigarettes were widely understood, some experts dismissed the link between the two as correlation without causation. Ronald Fisher, a famous statistician, theorised that cancer caused smoking, because people might be taking up smoking to soothe their already irritated lungs. In 1965, Darrell Huff (author of a bestselling book named ‘How to Lie with Statistics’) was paid by the Tobacco Institute to present similar arguments in testimony before a US Senate Committee on the risks of smoking.5

In the present case, McDougall J’s judgment explains how this tricky issue was navigated with respect to OneFlare.


Judgment

McDougall J began by clarifying that a statistical or regression analysis is not necessarily required to establish a causal relationship:

“First of all, causation is a matter of common sense, not a matter to be established only by statistical analysis. Of course, statistical analysis may help to prove causation in a particular case. But it does not follow, in the absence of statistical analysis, that causation has not been proved.”6

McDougall J went on to explain his use of common sense when reviewing the visual trends on the chart:

“Secondly, Oneflare’s business model worked because people went to its website to look for a service provider and, in some cases at least, gave work to that service provider. The revenue that Oneflare derived was earned substantially through its website ‘introduction’ of the two parties. As a matter of common sense, and on the balance of probabilities, it is reasonable to infer a causal relationship between traffic and revenue (or sales).

Visual observation of Mr Cavanagh’s graphs is consistent with that inference. In broad but sufficiently accurate terms, they do show sales tracking traffic.

I conclude […] that there is a reasonably direct link between the volume of traffic to Oneflare’s website and the sales revenue generated. The latter, I think, is in a reasonably direct sense a function of the former… 
It follows, in my view, that [the expert’s] evidence and common sense both support the conclusion, on the balance of probabilities, that [the black-hat SEO] caused a loss of revenue” 7,8


Common sense also necessitated some limits on the accounting expert’s calculations. For example, even once causation was established, it would have been wrong to assume that OneFlare’s early growth would have continued at the same rate indefinitely. To reflect this, the expert applied a growth rate which reduced over time as the business matured.

In his judgment, McDougall J also considered the size of the industry, OneFlare’s ‘first mover advantage’ and the potential for disruption, along with the ‘inherent uncertainty that seems to attend so many internet-based businesses’.9 He applied a discount rate to take each of these factors into account.
 

Outcome of the case:

Despite his findings on the calculation of damages, McDougall J ultimately found that the consultant was not liable, ruling that:
  • OneFlare had retained the consultant specifically to devise an aggressive SEO strategy, and the consultant had done so; therefore there was no breach of duty to exercise reasonable care and skill; and
  • The consultant had no duty to warn of the risk of harm (and in any event, OneFlare were aware of the risks involved).
Key takeaway: The courts will not accept that correlation (by itself) implies causation – but it’s possible (even without a statistical analysis) to establish causation when the correlation is supported by common sense.

Sources and further reading
Endnotes
1. This is a simplistic example. In OneFlare’s case, hundreds of thousands of links were created using ‘linkbuilding’ techniques such as ‘linkwheeling’, and rather than simply creating new websites, OneFlare bought ‘aged domains’ to add weight to some of those links. This was combined with more ethical tweaks such as improving the website’s structure, including lots of relevant keywords on the site, and adding content such as blogs.
2.This chart has been simplified for illustrative purposes – the expert’s actual chart was more detailed with a more complicated shape reflecting different trends over time.
3. Oneflare Pty Ltd v Chernih [2016] NSWSC 1271 at 227.
4. From the website Spurious Correlations - http://tylervigen.com/spurious-correlations.
5. In an article for the Financial Times, Tim Harford noted that both Huff and Fisher ‘were paid as consultants by the tobacco industry and some will believe that the consulting fees caused their scepticism. It seems just as likely that their scepticism caused the consulting fees. We may never know’.
6. At 248.
7. At 249-252 and 255.
8. McDougall J also pointed to evidence of the consultant’s plan and budget, which showed that he expected his SEO services to increase OneFlare’s traffic and subsequently its revenue, including a conversion rate and budget. He noted that “[i]t seems a little inconsistent for [the consultant] to now argue that there is no link between traffic and sales”.
9. At 274.