Methodology & Glossary
Sally Dickerson managing director Benchmarketing Analysis conducted by Sally Dickerson managing director Benchmarketing A graduate in Mathematics from Oxford, Sally joined Mars UK as a market research analyst, later joining their Management Science division. She co-founded The Decision Shop, part of Bates/Cordiant then in 1999 joined the then OMD group and set up ROI (Return on Investment) focused on market mix modelling, which became OMD Metrics, then BrandScience, and is now Annalect Marketing Sciences. In 2016 Sally created a new consultancy business, Benchmarketing, running strategic quantitative consultancy projects using meta analysis. She has also contributed to over 30 IPA advertising effectiveness awards, been an IPA effectiveness award and Cannes Creative Effectiveness judge multiple times, and run marketing effectiveness masterclasses for the Marketing Society and Chartered Institute of Marketing.
1 2 3 Methodology Advertising spend data A meta-analysis of econometric models Budget optimisation
SMI data provides a representative picture of actual media mix Actual media spend from booking data provided by 65% of UK agencies Data across all above the line media Detailed breakdown of digital spend across display, video and paid search Breakdown of newsbrand spend across print and digital So for this analysis we have used a source called Standard Media Index, which provides a much more accurate and representative picture of the actual media mix. It is based on actual media spend from booking data covering 65% of UK media agency billings. The data covers all above the line media. Importantly it provides a detailed breakdown of digital spend across display, video and paid search. So for newsbrands it covers spend both in print and in digital.
Total adspend - how it's calculated For this analysis we have used a combination of Nielsen and Standard Media Index. SMI is based on actual media spend from booking data covering 65% of UK media agency billings. The data covers all above the line media. Importantly, it provides a detailed breakdown of digital spend across display, video and paid search. So, for newsbrands, it covers spend both in print and in digital. In order to get figures as close to real world adspend levels as possible, we’ve applied a simple formula. We’ve taken offline adspend figures from Nielsen and used SMI digital adspend percentages. By combining these two pieces of data, we can fill in the gaps in each; so we calculate the percentage of total spend taken up by the Nielsen offline spend figures and also calculate total digital adspend fi gures, giving us an accurate proportion of the total media landscape.
Econometrics • Advertising x medium and message Identifying and assigning a weight to the ingredients driving sales and profit • Advertising x medium and message • PR, Media mentions, Buzz • Pricing vs competitors • Store universe changes • Product/Range changes • Brand awareness/perceptions • Competitor marketing • Competitor routes to market • Technological change • Seasonality • Economic change If sales are a cake, econometrics determines the recipe To understand return on investment of media, we need to be able to identify and quantify ALL key drivers of sales and profit. The ingredients will usually include: The Advertising x Medium and Message PR, Media mentions, Buzz Pricing vs competitors Store universe changes Product/Range changes Brand awareness/ perceptions Competitor marketing Competitor routes to market Technological change Seasonality Economic change The technique assigns a weight to each ingredient Once the ‘recipe’ is known, we can reproduce the cake by combining the weighted ingredients
A meta-analysis of 684 econometric models from 2011 to 2017 Benchmarketing conducted a meta-analysis of 684 econometric models from 2011-2017, across a number of different sectors. Meta-analysis is the interrogation of multiple sets of results data – “metadata”. It is common in pharmaceuticals, particularly in clinical drugs trials. One trial isn’t enough, you need hundreds to be sure of your results. Then the key skill lies in diagnosing why results between groups are different. In this work, the metadata comes from a single line for each model, with the inputs (spend, media mix) and outcomes (return on investment) all quantified consistently between models, together with the category of the brand being analysed. In this work, the metadata comes from a single line for each model, with the inputs (spend, media mix) and outcomes (return on investment) all quantified consistently between models, together with the category of the brand being analysed. Using our cake analogy – we can work out whether better tasting cakes always use butter rather than margarine in the recipe, and whether using three eggs works better than two
= PROI sales revenue x profit margin % ÷ media investment Campaign profit return on investment (PROI) PROI sales revenue x profit margin % ÷ media investment = Profit return on investment (PROI) is the revenue generated by advertising campaigns divided by the profit margin for each client over the short to medium-term. It takes into account the media investment and the cost of goods or services, so provides a much clearer guide to advertising payback than simply looking at the revenue generated. A key feature of this research lies in the concentration on total campaign PROI, as opposed to individual channel PROI. The most important thing is to prove the effectiveness of media on a total campaign level because this is what drives results and core business objectives. It’s important to note that profit calculations have already accounted for all media and cost of goods, so any PROI figure of over 1.00 is paying back on top of the initial investment in the short term. In this research, we’re concentrating on short to medium-term profit generation. This is increasingly the way that advertising is measured so it’s important to be able to prove the profitability of media channels in the short-term. This in turn will mean considerably higher long-term profit: Benchmarketing's estimate of the long-term effect is twice the short to medium-term PROI.
Profit return on investment (PROI) Scatter graphs allow us to see relationships in data. • In this example chart, used for illustration, we can see the relationship between media spend and PROI, looking at print newsbrands percentage of total comms spend and the PROI • Each dot on the graph represents an econometric model case in the results vault • Here the data suggests that as the % of print newsbrands in the mix increases, so does effectiveness
Creating tertile groups of cases shows the profit return for low, medium and high spend levels To focus on PROI, Benchmarketing have concentrated on splitting the cases within each category into three spend ‘tertiles’; low spend, medium spend and high spend (one third of the overall cases in each). Once this is complete, an analysis of the PROI of different levels of print and digital newsbrand spend is made. Using this data enables us to see the effects of different levels of print and digital newsbrand usage on the overall profit return of advertising campaigns. This allows us to make credible, evidenced, spend recommendations.
Creating tertile groups of cases shows the profit return for low, medium and high spend levels To focus on PROI, Benchmarketing have concentrated on splitting the cases within each category into three spend ‘tertiles’; low spend, medium spend and high spend (one third of the overall cases in each). Once this is complete, an analysis of the PROI of different levels of print and digital newsbrand spend is made. Using this data enables us to see the effects of different levels of print and digital newsbrand usage on the overall profit return of advertising campaigns. This allows us to make credible, evidenced, spend recommendations.
Here’s how it works for LEISURE AND PLEASURE super-category In the LEISURE AND PLEASURE category, the highest campaign PROI is found when print newsbrands are between 19.4% and 30.5% of the total media investment (the current average share of 8.8% for print newsbrands delivers the lowest campaign PROI). So, when print newsbrands are at this optimal share, the campaign will pack back £3.31 for every £1 invested. In the LEISURE AND PLEASURE category, the highest campaign PROI is found when digital newsbrands are between 1.7% and 3.0% of the total media investment (the current average share of 0.8% for digital newsbrands delivers a lower campaign PROI). So, when digital newsbrands are at this optimal share, the campaign will pack back £4.35 for every £1 invested.
Budget optimisation The data from the meta-analysis allows us to build response curves for each media channel. This works by analysing PROIs across data points in each category to generate average response curves. Curves that “go flat” suggest high diminishing returns and no benefits to additional investment. We can take an annual budget and optimise the overall PROI for the spend, by changing the mix. It’s a simple hill-climbing optimisation that picks the highest and the slopiest points by medium. From the curves we can calculate the optimum media split for any given budget.
Budget optimisation: Supermarkets example For the supermarkets category, the recommended average investment in print newspapers is 19% of the total media budget. This varies across different levels of spend, starting off at 8% for smaller campaigns up to £10 million, rising to 27% for campaigns around £40 million and then slowly reducing to 17% for larger campaigns at the £100 million mark. Until there are more cases with consistent spend where digital newsbrands are split out from all other digital display, it is only viable to recommend an overall spend share of 2.1% for all sizes of campaign. However, the opportunity for return from spend in digital newsbrands is very likely to be much higher.
Introducing the super-categories
EVERYDAY PICKUPS REFRESH AND REVIVE LEISURE AND PLEASURE SHINY NEW THINGS GROWN UP STUFF Frequent, habit driven purchases. Brands are generally favoured by default rather than thought. Low cost Choice is often welcomed and purchases are more considered, providing valued moments of ‘me time' or family time. Purchased relatively frequently at low-medium value Goods and services that are bought weekly (more or less). Medium brand consideration and medium value Goods and services mostly purchased or reviewed on an ad-hoc basis, the use of which often bring enjoyment. Quite high brand consideration and medium value Purchased roughly annually (sometimes more, sometimes much less). Varying degrees of interest but often significant spend and therefore often highly researched Examples – confectionery & sweets, household supplies, media e.g. newspapers or magazines Examples – alcoholic FMCG, gambling, National Lottery, entertainment & leisure, supermarkets Examples – computers & software, toys, games & consoles, clothing & accessories, consumer electronics, telecoms, retail (non-grocery) Examples – restaurants & coffee shops, drink – non-alcoholic FMCG like fizzy drinks, food FMCG, charities, beauty & personal care FMCG, health & medical FMCG, pharmaceuticals (not an advertised category in the UK) Examples – government and public sector, insurance, other offline services, other online services, travel & transport, energy & utilities, business services, motors, finance BSILC (banking, savings, investments, loans, cards)
Glossary of terms Adstock (advertising carry-over) Digital display Advertising carry-over rates measure the time period over which the media will drive a sales response. 50% carry-over rate a week means that if there were 100 impacts in the week of the advert, then there would be an effect of 50 in the second week and 25 in the third. Carry-over is identified by best fit and statistical confidence within the modelling process. It greatly influences the total return of a campaign. Digital display This refers to advertising online, in all of its formats (excluding search). Established media Established media refers to media channels that existed prior to the relatively recent rise of digital media, such a print newsbrands or TV. It’s important to note that established media forms have adapted to the digital landscape to become multi-platform. Meta-analysis A meta-analysis is a method for systematically combining data from several studies to develop ‘metadata’ and come to overall conclusions with greater statistical power than the sum of their parts (the individual econometric models that they’re made from). It’s common in clinical drugs trials in the pharmaceuticals industry. One trial isn’t enough, you need hundreds so as to be sure of your results. If all the trials come up with the same answer, that’s a very strong result. If the trial results are different, then being able to explain robustly why they are different – different dosage, different demographic sample – is again a result and new learning.
Glossary of terms Nielsen PAMCo Nielsen is an information and measurement company providing market research, insights and data about what people watch, listen to and buy. In this project, the data has been used to report offline media spend figures. PAMCo Audience Measurement for Publishers is the new JIC (Joint Industry Currency) for published media, using approved, world-leading methodology. It produces de-duplicated brand reach, allowing users to carry out reach and frequency planning to better understand audiences across all platforms. PROI – ‘Profit return on investment’ The revenue generated by advertising campaigns divided by the profit margin for each client over the short to medium-term. It takes into account the media investment and the cost of goods or services, so provides a much clearer guide to advertising payback than simply looking at the revenue generated. RROI – ‘Revenue return on investment’ The amount of gross income versus the costs of an ad campaign. This is not overall profit, it’s sales revenue. Sales response curves Sales response curves take a dynamic look at the changes in response to advertising according to different combinations of channels at different budget levels.
Glossary of terms Scaleable The ability to be resized to different proportions. In this project ‘scaleability’ refers to whether or not a media channel can be scaled up to take a higher proportion of the media budget and still deliver strong profits. For example, if a medium has low reach, additional investment might affect profitability as it will be hitting the same audience more frequently. SMI (Standard Media Index) SMI provides global adspend data straight from booking systems (covering 65% of UK media agency billings). In this project, it’s been used to calculate the proportion of total adspend attributed to digital media channels, as it separates different types of digital spend, including digital newsbrands. Tertile A ‘tertile’ refers to one third of any given dataset. In this analysis, we talk about ‘low’, ‘middle’ and ‘upper’ tertiles. Traditional mass media Also referred to as ‘established media’, traditional mass media refers to large audience media channels such as print newsbrands, TV or radio, which existed prior to the relatively recent rise of digital media. It’s important to note that ‘traditional’ have adapted to the digital landscape to become multi-platform.