Survey Evidence in Proceedings before the Australian Trade Mark Registrar

In 1990, French J (as he was then) made an observation about the utility of survey evidence in the decision  of State Government Insurance Corporation and State Government Insurance Commission v Government Insurance Office of New South Wales  [1991] FCA 121:

The trial of the action went for over 3 weeks, a substantial portion of which time was occupied with evidence relating to the reliability and admissibility of the results of market research surveys conducted on behalf of the applicant. In the event, there was a serious question about the extent to which, in competing logo cases, the Court can derive great assistance from such studies and about the relationship between their cost and the benefits to the litigation generally. For ultimately in logo and get-up cases, the characterisation of the conduct complained of is a matter for the judgment of the Court and the role of market research surveys is at best ancillary.

In court proceedings, survey evidence may be admissible in court proceedings as an exception to the hearsay rule: Arnotts Ltd v Trade Practices Commission (1990) 97 ALR 555 (Arnotts); Kellogg Company v PB Foods Ltd [1999] FCA 1610 (Kellogg case) ; Ritz Hotel Ltd v Charles of the Ritz Ltd (1988) 15 NSWLR 158. The Federal Court has now for many years had a series of practice directions dealing with survey evidence, so as to improve their efficacy and diminish the points of dispute between the parties relating to the survey.

But the delegates of the Registrar of Trade Marks, in a forum which has no rules of evidence, often are faced with surveys which are either confusing or very technical.

  1. Utility of Survey Evidence

Survey evidence plainly has utility in trade mark prosecutions. In Re NovAtel [2012] ATMO 42 (27 April 2012) Hearing Officer Thomson noted:

“But there will also be cases in which such materials will be of themselves inadequate to support a proper judgment. In some circumstances market or survey evidence might be necessary to form a concluded view.”

In Re Nordstrom, Inc, [2007] ATMO 24 (30 April 2007) Hearing Officer Beal noted the impact of the absence of survey evidence:

“The applicant has asked me to infer reputation in Australia from use and reputation overseas, but has not stated their intention to open retail operations or otherwise specifically target Australian customers. Nor has any indication of a growing reputation been demonstrated through evidence such as consumer surveys…”

In Re Majans [2016] ATMO 47,  Hearing Officer Brown noted, “I note that the Applicant has not provided any support for this assertion, such as a market  survey  or third party declarations attesting to the association.”

In Re Mount Everest Mineral Water [2012] ATMO 65,  Hearing Officer Thompson noted, “In some circumstances market or survey evidence might be necessary to form a concluded view”.

A failure to adduce survey evidence leads to risks in respect of how a hearing officer might apprehend consumer recognition of a trade mark.  In Re Discount Drug Stores Pty Ltd  [2014] ATMO 66  Hearing Officer Worth said, “I note also that market surveys may comprise direct evidence of consumer appreciation of a trade mark, but that none has been conducted to support the Applicant’s assertions.”

  1. Problems with Survey Evidence

The Kellogg case criticised survey evidence used for the following reasons:

  1. the respondents did not reveal exactly what question was asked of them, where it was asked, how their ‘interrogator’ identified him or herself and if there was any prior relationship between them, who asked the question, whether the person who carried out the survey was an independent survey firm or a person connected with the opponent.
  2. the results were presented as a bundle with no further information as to how the respondents were selected, whether these were the only respondents selected and what (if any) connection the respondents might have with the opponent.
  3. The fact that four of the seven respondents worked in the healthcare industry may suggest some form of bias in their selection or that they are not a representative sample.
  4. The number of people surveyed was also very small and quite possibly statistically insignificant even where the other concerns about the survey were properly addressed.

In Re Irene Notaras  [2013]  ATMO  98 (25 November 2013) Hearing Officer Wilson noted with apparent scepticism that:

“The circumstances under which this survey was made have not been supplied by the Applicant. I note that this evidence paints the unlikely scenario that not one of the persons surveyed failed to recognize the Trade Mark. The Applicant’s survey form contains a picture of the Trade Mark and two of the questions asked were:

  • Who do you recognise to be the source of the above shape?
  • Who do you recognise to be connected and supplier of the above?

It is likely that the Applicant surveyed a number of its own customers as many of them comment on the fact that they actually own an ‘ATOMIC’.”

This scepticism was mirrored in Re Cashcard Australia [2007] ATMO 78 , where Hearing Officer Lyons noted:

“The respondents who took part in the Daemon survey are described as both ‘cardholders’ and ‘Cashcard customers’. It is unclear to me whether the applicant’s own customers were being surveyed. If that were the case, then clearly they would comprise a skewed survey group. The evidence is simply unclear on this point and so I must treat the Daemon survey with caution.”

In Re Apple Inc [2015]  ATMO  25 (12 March 2015) it was noted that:

“The Survey results provided do not assist the trade marks [sic] claims for acceptance. An example of the survey conducted was provided. The 1st question provided a representation of different connector plugs. Contained within these examples was an image of the Applicant’s 30 pin device. The evidence supports that the level of distinctiveness of [the Trade Mark] increased when positioned on the 30 pin shape. As the applicant has disclaimed the shape the survey has the potential to lead the participant to a desired outcome.

The results of the survey are also unconvincing particularly due to the small sample size, exposure to the shape of the 30 pin connector which has been disclaimed and the results.”

In Samsung Electronics Australia Pty Ltd v LG Electronics Australia Pty Ltd [2015] FCA 227 the court considered that the following factors must be considered to give any weight to survey evidence:

  • quality and integrity of any questionnaire used for the purposes of the survey;
  • the methods of data collection and analysis employed; and
  • whether the respondents to the survey are sufficiently representative.
  1. Number of Survey Respondents

The survey should include a statistically significant number of respondents. There is little judicial guidance on this matter other than the statement of Branson J in Blount Inc v Registrar of Trade Marks:

In each case the evidence relied on by the applicant is to be evaluated in the light of any evidence tending to the contrary effect and having regard to the evidence which the applicant might reasonably be expected to be able to obtain in all the circumstances of the case.”

In Hansen Beverage Company v Bickfords (Australia) Pty Ltd [2008] FCA 406 at [93],  a survey of 264 respondents regarding their awareness of the MONSTER ENERGY mark was regarded by the Federal Court as too small. Similarly in Application by Multix Pty Ltd (2004) 64 IPR 128,  the survey evidence was given little weight as only 50 surveys were prepared and only 20 surveys completed.

However, in Citigroup Inc v Citiwide Holdings Pty Ltd [2013] ATMO 31, at [30],  a total of 503 respondents was considered to be a “reliable and statistically meaningful representation of the Australian Financial Services Market”.

The truth of the appropriate number of survey respondents lies in statistics, not law. The term is called “sampling error” – the margin for error in the results. A survey will often confined itself to 400 survey participants. This is because a population of 10000 people requires a sample of 370 people for a +/-5% sampling error – a 95% confidence interval (which, by any measure, is a very good degree of confidence). Surveying more than 400 respondents will affect the survey result, but the increased number of survey participants  will have a declining impact.  The chart below demonstrates a 5% sampling error associated with 400 respondents.  But if the sample size is increased fourfold to 1600 then the sampling error is reduced by only half (to 2.5%).

(Source: Catalyse – catalyse.com.au)

The other factor to consider is the survey result. As the result moves towards 0% or 100%, the sampling error declines.  The following chart is based on a sample size of 400 respondents.  More than 400 survey participants will not affect the sampling error, the close the result is to either extreme.

(Source: Catalyse – catalyse.com.au)

  1. Survey Sample Representative of the Relevant Cross-Section of the Public

Hearing Officer Thompson in Re Paramount International Export [2009] ATMO 94 noted that “difficulties exist with the informal survey conducted by David Yates. I am not sure how the declarants who have responded to the survey were selected, whether they are the only respondents to the survey and how they happened to be sitting by their phone and a computer at the time of the survey”.

In Australian Postal Corporation v Digital Post Australia [2013] FCAFC 153  the Full Federal Court held that the consumer surveys presented in that decision were inherently unreliable, noting that the respondents were recruited from a sample group that was already well-acquainted with the brand of one of the parties in an earlier survey, and so there was no telling whether the respondents already had the brand in their minds when answering the survey. At first instance, the judge said:

… the major problem with the Court’s use of the surveys to inform its consideration of the issue of deceptive similarity is that the class of survey respondents was not confined to potential users of the proposed digital mail service. The respondents were at large. I place no weight on the survey evidence. It does not assist me to determine what is, in any event, a matter for the Court.

On the other hand, it was held by the delegate in Citigroup that it was reasonable to direct the survey to the public at large when the market involved banking, financial and insurance services.

The delegate was critical of the sample population surveyed in Vantage Systems Pty Ltd v IVision Pty Ltd [2011] ATMO 90 which comprised tertiary educated people, in circumstances where the relevant consumers of electronic audio visual conferencing facilities were likely to be secretaries. The delegate noted that there was no evidence matching the survey respondents to the demographic of the relevant purchasers.

  1. Recruitment Methods of Survey Respondents

Two surveys filed by Australian Postal Corporation in Australian Postal Corporation v Digital Post Australia [2013] FCAFC 153  in relation to the deceptive similarity of the marks AUSTRALIA POST and DIGITAL POST AUSTRALIA were given no weight by the Full Federal Court because of undisclosed recruitment methods and unreliable recruitment methods which potentially tainted the responses provided.

Similarly, in Samsung Electronics Australia Pty Ltd  [2015] FCA 227 a survey relied upon by LG regarding consumers’ understanding of the term “Full HD” was given no weight by the Federal Court as the results were statistically unreliable because of questionable sampling methods of survey respondents.

  1. Relevant, Unambiguous and Non-Leading Questions:

In Vantage Systems (above) the survey question which allowed only a yes or no answer was defective as it would have been more appropriate to allow for a gradation of answers.

In Adidas AG v Pacific Brands Footwear Pty Ltd (No 3) 2013) 308 ALR 74 at 109, Robertson J gave the survey evidence little weight in part because it contained a leading question – “who do you think makes this shoe?”.

In Minnesota Mining & Manufacturing Company  Hearing Officer Purvis-Smith said, in respect of questions being asked of a survey group relating to canary yellow for adhesive post-it notes:

The application in question is for canary yellow, and it this precise colour which must be addressed by the evidence. At the very least, the respondents should have been asked whether, in their knowledge, there were shades of light yellow stick-on notes available. This would have assisted me in determining whether the average consumer is able to differentiate between the various shades or whether they were aware of any noticeable or distinctive difference between yellow notes generally. The failure to do this reduces the value of the survey in relation to a s.41(6) as it does not assist me in determining whether the trade mark distinguishes the goods from those of other manufacturers. I have no way of knowing what proportion of the respondents used the term post-it notes as a generic term or as a trade mark. For that reason, I cannot ascertain what proportion of respondents associated the colour yellow with the applicant’s products. If I cannot ascertain that point, I cannot determine whether a reasonable proportion of consumers would associate the colour canary yellow with the applicant’s product. This goes to the very heart of the s.41(6).

Concise questions in a survey, specifically crafted, will mitigate the risk of such criticism.

  1. Survey Results

In Re Apple Hearing Officer Thompson noted that:

“…while the recognition rate of 32% in the  survey might be regarded as low, there is no requirement that a trade mark be 100% recognised in order for the provisions of subsection 41(5) to be applied – such a high extent of recognition looks rather to reputation (or to the requirements of subsection 41(6)) rather than the capacity to become distinctive.”

But is 32% low? This is one in three people. Survey results can be misinterpreted unless explained very clearly. In Chocolaterie Guylian N.V. v [2007]  ATMO  30 (31 May 2007) the hearing officer reached the following conclusion:

“Apart from the above examples, the evidence upon which the holder has put significant weight is the ACNeilsen “Chocoholic” survey. The results of this survey suggest that just over 40% of the population who took part recognized the seashell shape as coming from the holder company. This is indeed a significant result but it leaves over half of the buying public unsure of, or mistaken as to, the producer of the particular shape of chocolate. I consider this to be an important result since other traders are also making chocolates in this shape. The majority of the buying public does not connect the shape with the holder. Therefore, while the holder has a strong presence in the market under its well-known Guylian name, it has not shown that the seahorse shape itself is capable of doing the work of distinguishing the holder’s goods from the similar goods of other chocolate makers.”

But what this omits is that, on the other hand, two in five people did recognise the seashell shape as originating from the well-known chocolate manufacturer Guylian, and those two in five people would be misled if it came from another source.

  1. Conclusion

The lessons from this summation are obvious: use surveys, and indeed the decisions encourage this. But they should be well-crafted, and not tossed together. Further, the advocate at the hearing should properly understand the statistics underpinning them so as to inform the examiner.