This column is re-published from one of Brian Wieser's Pivotal Research Group research notes to Wall Street investors.
Accounting choices and financial disclosures can be an arcane topic even among the investor community. This is especially true in a sector where revenue momentum frequently drives valuations, as is the case with ad tech. These issues re-emerged for us last week following earnings reports from both Facebook and Yahoo, where investors were effectively reminded that not all economic activity associated with each company is necessarily booked as gross revenue. Specifically, Facebook clarified on its earnings call that its ad network revenues are booked on a net basis (excluding media costs). Similarly, Yahoo – which provided guidance for the first quarter that was essentially flat year over year, despite the new Mozilla search agreement and acquisition of BrightRoll – conveyed that much of this new activity is booked on a net basis, as well. By contrast, activity at Google (such as its exchange and display ad sales) are, we believe, booked on a gross basis. Meanwhile, Twitter stated on its earnings call yesterday that its exchange revenues are booked at net, while any network revenues it generates will be booked at gross. This topic has also emerged within the ad agency sector over the past year, where WPP and Omnicom are reporting proprietary trading activity associated with programmatic buying units as part of gross revenue, but only WPP is consistently disclosing net revenue separately from gross.
Economic activity in one form may be booked as revenue at one company while a similar endeavor may be differently classified at another. In principle, there are clear rules around what revenue is and how it is recognized. However, the specific deal structure that a company arranges with its partners may lead auditors to deem an activity is not reportable as revenue. For example, Rubicon Project notes among its revenue recognition policies that it bills buyers for gross advertising purchased plus fees, if any, and then remits to sellers amounts spent less fees that sellers pay them. The underlying activity that Rubicon facilitates is not defined as revenue, but “managed revenue” (which is to say the reported revenue figure is an appropriately conservative definition). “Revenue” reflects the fees that Rubicon retains. While we believe this is an appropriate approach, use of this metric limits comparability between Rubicon and, say, Rocket Fuel unless we reference a more common metric which excludes media costs for both companies, such as gross profit. (To this point, we think the most appropriate single metric against which to compare ad tech companies is the gross profit line rather than anything else.) While we can clearly characterize both companies as operating with very different business models today, the fluid nature of the industry may mean that both companies might look more alike each other in the future, while accounting choices may remain very different.
However, even if we get comfortable with equating gross revenues and net revenues (or revenues less media costs or revenue ex-traffic acquisition costs) among different companies, another issue arises around rebates and commissions to third parties, especially to agencies, which can be significant for some companies in the ad tech space, and modest for others. For example, in ad network MaxPoint’s newly filed S1, the company notes that it “paid…channel partners a commission, or fee, in exchange for providing the company with new business. The company recorded reductions to revenue for consideration paid to its channel partners.” This also seems appropriate. However, we have heard of other instances where rebates are possibly deducted out of sales and marketing expenses rather than out of gross revenues or before the activity is even booked as revenue. This would help to increase the perceived size of a company in terms of its revenues. Presumably the specific contractual language of payments dictates accounting choices, even if a behavior is fundamentally the same.
At the agency holding companies, there are differences in how agency trading desk or other similar activity is reported, partially reflecting different business model choices and partially reflecting different managerial choices. Inclusion or exclusion of media costs associated with trading desk or programmatic platform buying activities as absence of any distinction between net revenues and gross revenues in some instances are beginning to distort top-line organic revenue growth comparisons across an otherwise mostly homogenous group of competitors.
WPP’s CEO Sir Martin Sorrell has been vocal about the need for other agency holding companies to be more transparent in disclosing pass-through costs, especially following a transition in their own accounting last year to recognize proprietary trading (primarily associated with programmatic platform Xaxis) as gross revenue. This has led the company to provide separate organic growth figures on both a net and gross revenue basis, which we agree is the right approach for the industry.
Sorrell’s comments have mostly been pointed at Omnicom, which also began booking proprietary trading activity associated with its Accuen trading desk as revenue last year, but without the similar breakdown of net vs. gross, distorting growth figures reported on its earnings releases and its securities filings in the process. Omnicom’s commentary on earnings calls allow one to make a reasonable estimate of apples-to-apples organic revenue growth, but this pre-supposes that everyone hearing the disclosure understands the nature of the underlying business activity. The gross / net issue is slightly less of a matter for Interpublic, which does not book activity associated with its programmatic buying unit Cadreon as net revenue, as it does not take principal risk on inventory. However, IPG does book other pass-through costs as revenue. When those activities are significant, the company highlights them on their earnings calls. Proprietary trading associated with its Orion barter unit are, however, booked as revenue and not typically broken out, although the rate of growth at that unit is unlikely to be comparable to activities associated with trading desks, and so is not likely distorting organic revenue growth trends. The same would likely be true with Omnicom’s barter division, Icon International.
Overall, we think that agency holding companies would benefit from the provision of more explicit break-outs of net and gross revenues, especially as proprietary trading activities become increasingly common going forward. As a general rule, we believe when investors better able to understand a company’s revenues and revenue growth in comparison to its peers, capital will be more freely supplied. The direct comparability of the holding companies is, we think, a major advantage for investors, as it is relatively easier to benchmark each holding company’s growth trends. We believe this facilitates a more efficient flow of capital (which means capital essentially gets cheaper as investors can make more intelligent choices with more higher quality information) to companies in the sector. As proprietary trading and programmatic platforms become increasingly common for the agency industry, greater transparency around net and gross organic revenue growth trends will therefore become increasingly important as a result.
Another issue related to revenue recognition that will grow in importance is how regional activities are quantified and whether or not (and when) they are disclosed. Regional revenue recognition is particularly important for agencies and digital media companies given the global nature and distributed activities associated with each company’s revenues, as reporting – or lack of reporting on – trends in a secondary region or in international markets may distort perspectives on a company’s core business on a like-for-like regional basis. There are many instances in agencies and ad tech companies where activities which seemingly occur in one region are recognized in others. The problem occurs for agencies when client teams operating predominantly in one country service a client based in another. Where the revenue should be recognized may be inconsistent across companies, although it is not likely to be material in most periods for holding companies, as most revenue generating entities (and allocated P&Ls) within agencies are located close to clients.
By contrast, for ad tech companies, a greater proportion of activity is better able to be geographically removed from their clients, and presumably the geographic location of a specific activity is what dictates the geography in which revenue is recognized. For example, Rubicon is forthright in describing its operations as global, but we know the company has significant marketplace traction in many markets around the world. However, for reporting purposes “substantially all of the company’s revenue is U.S. revenue, determined based on the location of the Company’s legal entity that is a party to the relevant transaction. Revenue originated in foreign countries was not material during the years ended December 31, 2011, 2012, and 2013” according to its prospectus from last year. While the reporting here may accurately reflect the nature of Rubicon’s business, the absence of reporting of international vs. domestic revenues in securities filings or on earnings releases limits assessments of industry-wide growth trends on a regional basis.
There are other international revenue recognition issues beyond these items associated with the timing of accurate data provision. For example, some companies, such as YuMe, do not provide international growth figures on earnings releases, limiting the meaning of the headline figures until weeks later, when international earnings are included in the company’s 10-Qs and 10-Ks. Separately, we can point to companies including Facebook, which reports geographic skews of revenues on press releases and on earnings calls apportioned by users, but more appropriately reports the geographic skew of revenues apportioned by paying entities (i.e. advertisers) on 10-Qs and 10-Ks which similarly appear later.
Other issues we are mindful of are likely a function of what is required of companies in terms of segment disclosures. For example, when businesses have separate internal P&Ls or divisional general managers who report to the company’s CEO directly, we always find it odd that companies report only one segment for a company, or a handful of segments which mash up different kinds of revenues across multiple business units. Rightly or wrongly, this is standard practice for most of the companies we cover, and is especially problematic for companies which are increasingly diversified away from a core advertising business such as is the case with Google. Investors are left guessing about the real underlying trends driving company performance, and we believe that capital flows less freely when this is the case.
Although the underlying economics of a business should not be impacted by reporting and disclosure choices, inconsistencies across an industry can impact external perceptions of different businesses and potentially distort outsiders’ efforts to size markets and assess prospects for any individual business. Accounting and disclosure choices do not matter much for companies with large ad tech businesses such Google, Facebook, Yahoo and Twitter as investors have a wide range of metrics to focus on and areas which seem to be inconsistent across those companies are still small. However, they matter much more in rapidly emerging sectors such as ad tech and for agencies, where ad tech-related activities are a potentially meaningful source of incremental growth, at least on a relative basis. As such, we think that differences in accounting choices and disclosures warrant more attention than they have received in the past.