Search has been one of the success stories of the past five years, accounting for more than half of all online advertising spend. In the early days of paid search, it cost so little to bid on even the most-successful keywords that measurement of the effectiveness of this spend was rarely a priority.
Indeed, Andreas Pouros, managing director, search engine optimisation (SEO) at Greenlight, says that at the time of hiring his agency, 50% of its current clients did not have adequate search measurement through web analytics in place. By 'adequate', he means tools that were capable of comprehensively measuring paid and natural search separately, down to the search term, and marrying that up with revenue delivery.
'The remaining 50% were not measuring search at all,' admits Pouros. In some instances this was because they had no real web analytics in place. When they did, the software often had not been set up to track search at all, thereby requiring more money from the client to create reports to get the information they wanted.
Even after the agency had advised them of the situation, Pouros says that 30% of existing clients still did not measure search six months later. It is clear, then, that it will take some time to help brands appreciate the importance of measuring their search performance on an ongoing basis.
According to a UK search engine marketing report by E-consultancy, 43% of company search marketers did not know what their return on investment from paid search was, while 61% did not know what the figure was from SEO.
'While advertisers are allocating close to a third of their online advertising budgets to search, including commercial and organic listings, very few have a clear and unified measurement methodology for the channel,' says Alain Portmann, founding partner at digital agency Web Liquid.
A third of budgets might be a conservative figure; according to the Internet Advertising Bureau, paid search accounted for 57% of online search during the first half of 2007.
As the market becomes more sophisticated, and keyword prices rise, brands are beginning to grasp the importance of such metrics. Accurate tracking of search engine activity is vital to a website's ongoing development - when a brand tracks its search engine paths, how long customers stay on its site, and the content they find most interesting, it can be used to stimulate more targeted communications.
More specifically, Lucy Allen, managing director of natural search and online brand-positioning agency LBi Netrank, advises marketers to consider metrics such as Google PageRank, site-traffic figures and conversions (see essentials). She says that if a page has heavy traffic but a high bounce rate (when consumers enter a site and fail to click on any additional pages within it in a specified time period), an SEO agency would look at the campaign's priorities and encourage the client to change the page to increase conversions, rather than trying to increase rankings in natural search results.
Measuring bounce rates on landing pages is one of the easiest ways of spotting unsuccessful targeting of campaigns, or design and content flaws within a website. 'One client's pay-per-click campaign was delivering traffic to a page with a bounce rate of more than 80%,' reveals Jenni Lloyd, online marketing consultant at social media agency Nixon McInnes. 'Each of those clicks represented a wasted £5 cost.'
Neil Morgan, vice-president, marketing, at web analytics company Omniture, identifies four types of online business, each with an interest in specific types of statistics. An ecommerce site, he says, will look at products sold, revenue, conversions per keyword, and revenue per keyword. A media site is interested in visitors' time on site, which ads they look at, and subscriptions sign-ups. For a B2B site, the focus is lead metrics. Self-service sites, meanwhile, such as online banking, need to examine the number of log-ins to customer accounts driven by search.
Morgan believes that many brands are simply measuring the wrong areas. 'If you look at something like car insurance, people will search on that term at the beginning of a sales cycle, then come back later in the sales process and search on more specific terms,' he explains. 'The keyword that produces the most click-throughs is not the one that drives the most conversions; and most companies are optimising their spend on click-throughs, rather than conversions.'
A surprisingly small number of companies, he believes, are tracking the business outcome to the search campaign, but the figure is rising. 'Companies like Centerparcs and Lastminute.com are doing this now,' says Morgan. 'We tell our clients: "Here is what is driving conversions so that's what you should be spending more money on".'
This 'last-click' issue is one of several obstacles that make the business of measuring the success of search a tricky one. According to Ryan Murdoch, analytics director at Mediaedge:cia, it means that non-brand terms, used early on in the sales process, do not receive the credit they deserve for their role in the purchase, since consumers are disproportionately likely to convert through searches on brand terms.
'Understanding this process is critical in order to plan search-marketing budgets in the most cost-efficient manner,' says Murdoch. 'By studying this at cookie level, we are able to examine how consumers behave when searching for products.'
There are other stumbling blocks, too. Today's consumer operates in a multichannel world - they may go online to get a quote for car insurance, for example, but complete the sales process by making a telephone call.
'You need the web analytics system to tie up with the CRM system. This will join the dots, but is often where it falls down,' says Richard Gregory, chief operating officer at Latitude. 'The web does not always integrate with the back office as smoothly as one would like.'
The problem is exacerbated by the fact that companies routinely have thousands of keywords under management at any one time.
With keyword inflation showing no sign of slowing, and companies having to justify their spend on search, as on all other forms of marketing, companies in such situations need to improve their lines of communication - and fast.
- 'There are four types of online business, each with an interest in specific statistics types: banking, ecommerce, media, B2B' - Neil Morgan, Vice-president, marketing Omniture
- 'Marketers should consider metrics such as Google PageRank, site-traffic figures and conversions' - Lucy Allen, Managing director LBi Netrank
- 'When hired, 50% of Greenlight's current clients did not have adequate search measurement in place' - Andreas Pouros, Managing director, SEO Greenlight
CASE STUDY - FLYTHOMASCOOK
Flythomascook's digital agency, BLM Quantum, was dissatisfied with the third-party bid management systems it used.
These all worked on the basis of 'click x conversion = target cost per acquisition (CPA)'. However, conversion alters from week to week. In the travel category, it is affected by factors such as the changing importance of specific destinations, seat availability, seasonality and price.
Last year, the agency developed its own system, which is designed to accurately calculate: 'click x expected conversion = target CPA.' While this difference may seem small, the agency says it has been significant.
The tool allows for all the factors influencing conversion rates, and enables BLM Quantum to find the expected conversion, based on the previous week's figures along with availability, price and seasonality (based on figures from the past few years) by destination.
Each week, together with Thomas Cook's yield management team, BLM Quantum inputs by destination: last week's conversion figures, availability, price and a seasonality factor. The amalgamation of these numbers forms an expected conversion figure for the following week.
Knowing the target CPA and the expected conversion, it can calculate the total average bid price across all keywords for that week. And with two years-worth of conversion data by keyword, it can factor the expected conversion and bid price down to a keyword level.
The tool has led to a dramatic improvement in results. Thomas Cook ran two reports to measure the difference in sales between the same two periods in 2006 and 2007 on Google.
While spend increased by 12%, the volume of sales increased by 71%, and the CPA decreased by 35%. As for ROI, in 2006, every £1 spent delivered £17 in revenue, while in 2007, the same delivered £27.
ESSENTIALS - WHAT TO MEASURE
- Google PageRank
- Site-traffic figures (including total visits, unique visits and page views)
- Referring URLs
- Cost per acquisition
- Bounce rate.