Winning the Benefit of the Doubt

IPA Effectiveness Paper

Client: The Conservative Party

Agency: M&C Saatchi

Principal Author: Richard Storey

Contributing Author: Steve Parker

2

3

Summary

This paper explores the role communications played in helping the Conservatives secure a General Election victory most commentators, pollsters and voters didn’t expect.

Specifically, we show the marginal effect paid-for advertising had over and above the omni-channel communications onslaught that characterises elections.

The 2015 Election was shaped by six parties, not the usual two or three. So, the concept of swinging voters from Labour to Conservative was superseded by a strategy of swaying supporters of other parties.

Forensic up-weighting of Target constituencies swayed the borderline decisions of marginal voters considering voting for ‘others’.

Seat by seat analysis pinpoints the decisive contribution this activity made. Without it, an outright majority would not have been won.

With the binary organisational objective met, we assess payback under the new criterion of Reasonable Utilisation of Marketing Assets, against which the campaign made a more than reasonable return.

4

5 An Abnormal Market

General Elections are quite unlike most markets evaluated for IPA Effectiveness submissions.

The product is sold over one 15 hour period every five years1.

Between these periods, there’s much influence, brand repositioning, packaging change, product reformulation and a surfeit of brand tracking.

But sales on the day is the only measure that counts.

Purchase is ‘free’, but customers know there’s some cost to themselves, their families, communities and country. Judging the nature, extent and personal impact of that cost is key to the purchase decision.

Brands are fairly consistent nationally, but only offer ‘local’ products, with variation in delivery and presentation.

Competitive choice varies locally. The constituency system makes this critical. Under ‘First past the post’, the brand leader isn’t the one purchased most nationally, but the one purchased most in more places.

Distribution of sales is therefore critical, especially relative to competitors.

Seat Distribution

General Election Result 2010

Source: BBC

This highlights the peculiarity that a customer’s choice doesn’t necessarily deliver them that choice. Who they choose instead of can be as significant.

We’ll see the implication of this for the Conservatives’ 2015 strategy and its evaluation, beginning with the prognosis.

1 The 2015 General Election was the first held since the 2011 Fixed Term Parliaments Act and so took place 5 years after the 2010 Election. Previously, scheduling of the election was at the incumbent’s choosing, (unless forced by a vote of no confidence), with the mean interval of a little over 4 years. 6 Prognosis

The Conservatives weren’t expected to win.

Incumbent governments, irrespective of party, typically lose votes and seats. Since 1832, the governing party seeking re-election lost on average 3.9% of the vote and 59 seats2.

Change in Votes and Seats of Incumbents Seeking Re-election

Change in number of seats held Change in share of vote (%) 100% 5.0%

0% 0.0%

!100% !5.0%

!200% !10.0%

!300% !15.0%

Share Vote

!400% !20.0% 1835% 1837% 1841% 1847% 1857% 1859% 1865% 1868% 1885% 1895% 1918% 1929% 1931% 1935% 1945% 1951% 1959% 1979% 1987% 1997% 2001% 2005% 1852% 1874% 1880% 1886% 1892% 1900% 1906% 1910% 1910% 1922% 1923% 1924% 1950% 1944% 1964% 1966% 1970% 1974% 1974% 1983% 1992% 2010% Source: Politics Resources

The last Prime Minister in office for over 18 months who increased vote share was Lord Salisbury in 1900, at the height of the Boer war. Every established incumbent since, including Thatcher and Blair, lost share3.

Polling suggested that would happen again, predicting a 33.7% Conservative share, compared to 36.1% in 2010, suggesting a loss of 34 seats4.

Furthermore, although relatively well funded, the Conservatives had less to spend than in previous elections, with considerable less available for advertising5.

Conservative Campaign Spend & Proportion Spent on Advertising

17.9% 16.7% 15.6%

12.8%

Spend (£m)

8.2% 7.5% Total%

4.4% 3.6% 45.8% Adver7sing% 44.9% 34.4% 23.1%

2001% 2005% 2010% 2015%

Source: Electoral Commission

Effective asset deployment would be key.

2 Source: Politics Resources. With population and boundary changes, the number of seats in Parliament has varied between elections, invalidating changes in seat numbers as a meaningful measure for some years. 3 John Major in 1992 was the last Premier to increase share of vote. But he was relatively newly installed as party leader. 4 Source: Poll of polls, 8 January 2015. The 2010 figure is actual, not a poll. As we’ll see, the distinction between the two is critical. 5 Source: Electoral Commission. Under Election rules, all campaign spend must be declared, together with invoices. This includes spend on manifesto, PR, broadcasts, admin and overheads, transport, rallies and events, unsolicited materials and market research. Conservative advertising spend was reduced in favour of, amongst other things, greater investment in market research. Advertising spend includes paid-for digital activity. 7 Base strategy

The bedrock of the strategy was a national push to shore up Conservative votes by a). Increasing Conservative voters’ likelihood to actually vote and b). Ensuring they didn’t defect.

The factor most likely to influence voters was confidence in handling the economy; This was twice as influential for intending Conservative voters.

Labour voters were more influenced by cost of living and healthcare, which became central planks of Labour’s strategy. However, wooing these voters wasn’t a key priority.

Importance of The Economy to Voting Intention

Single most important issue that will influence how you vote (%)

40" All"Voters" Conserva3ve"

30"

20"

45.8% 10" 44.9% 34.4% 23.1%

0" Economy Immigration Healthcare Cost of living Welfare Education Other Housing Pensions Crime Defence Environment Source: BMG, May 2015

Importance of The Economy to Voting Intention

Single most important issue that will influence how you vote (%)

40" All"Voters" Labour"

30"

20"

10"

0" Economy" Immigra2on" Healthcare" Cost"of"living" Welfare" Educa2on" Other" Housing" Pensions" Crime" Defence" Environment"

Source: BMG, May 2015

Labour were not well regarded on handling the economy amongst Conservative voters and a key theme of the 2010-15 Coalition Government had been reducing the deficit and returning the economy to growth.

The Recovering Economy

Annual GDP £ bn

1800$

1700$

1600$

1500$ 2007$ 2008$ 2009$ 2010$ 2011$ 2012$ 2013$ 2014$ 2015$ Source: ONS, 8

So, the Conservatives adopted the narrative; ‘The economy is recovering, don’t let Labour wreck it (again)’. This was tightly controlled across all communications channels, from party-election broadcasts to hustings6.

Example Communications

The second most influential factor was suitability for Prime Minister, the leader of either party with the higher rating having won the last nine elections.

Importance of Perceived Suitability for Prime Minister

Party leader satisfaction rating on eve of election, by ultimate outcome

Source: YouGov Dec 2014

David Cameron was a key asset compared with Ed Miliband or . This was particularly true amongst supporters of each party.

6 Most digital activity was created by Craig Elder and Tom Edmonds with the Conservative’s in house communications unit. Posters and some election broadcasts were produced by M&C Saatchi. The remaining activity was created in house. 9 Suitability for Prime Minister

38#

19#

4#

Cameron# Milliband# Clegg#

Who would make the best Prime Minister (%) Source: YouGov Dec 2014 Amongst all intending to vote

Suitability for Prime Minister Amongst Supporters

96#

53#

37#

Cameron Milliband Clegg

Who would make the best Prime Minister (%) Source: YouGov Dec 2014 Amongst all intending to vote for that party

So, the Conservative’ campaign made extensive use of , urging voters to ‘Vote for the person you most want as Prime Minister’.

Upweight strategy

However, the base strategy wouldn’t be enough to win outright. The Conservatives needed a plan to gain incremental votes and seats.

So, a list of Target Seats was drawn up, with up-weighted resources7 allocated to them.

Approximately half the seats were ‘must hold’, where the result was expected to be particularly tight. Half were ‘aim to win’ seats, where a Conservative uplift could snatch victory8.

This approach was commonplace for most parties. Indeed, Labour named its 106 seat ‘hit list’ in 2013.

However, 2015’s unique context made the nature of the Conservative list unlike any previous.

7 Principally, posters, digital, Ministerial visits, and up-weighted door to door and telephone campaigning 8 This was originally launched as the 40:40 initiative, with further seats added later. Despite plenty of speculation amongst commentators in the press, the specific number and names of seats remain a critical secret. 10 Changing context

Traditionally regarded as a two horse race, elections since the 1930s were fought on the concept of ‘’; turning Labour voters Conservative or vice-versa. (Although the average swing since 1945 was actually only 3%).

The “Swingometer”

Robert'McKenzie'with'a'swingometer'on'elec5on'night,'1964.'(BBC'Archive)'

However, since the 1950s, there had been a steady rise in ‘other’ parties, to over a third of the vote.

Rise of the ‘Others’ Vote

Total share of vote for parties other than Labour and Conservative (%) 40"

30"

20"

10"

0" 1945" 1950" 1951" 1944" 1959" 1964" 1966" 1970" 1974" 1974" 1979" 1983" 1987" 1992" 1997" 2001" 2005" 2010" Source: BBC

In 2015, parties focused on immigration, independence and the environment accelerated this trend. Six parties were expected to poll over a million votes, with the prospect of more support for ‘others’ than either ‘main’ party.

More Undecided Voters in 2015

2015 2010

Don’t Know 15# 15#

Uncertain 36# incl. Don’t know 45#

Uncertain 32# 36#

64# Certain 53# 55# Certain 49#

Definitely#Decided# May#Change#Mind# Definitely#Decided# May#Change#Mind#

Sources: ComRes Jan 2015, Ipsos MORI Feb 2010

11 In addition, there were also more undecided voters than usual9.

Examined seat by seat, the implications were critical. Only 42% of seats were ‘straight’ Conservative/Labour battles10. Even within England, that figure was 46%.

This was highly pertinent for the Conservative’s list. Only 8% of Targets were ‘straight’ Conservative/Labour battles. The remaining 92% were effectively contested with ‘others’. 10% were three-way marginals. 37% were versus Liberal Democrats. In 60%, UKIP posed a threat of either winning the seat outright, or ‘stealing’ sufficient votes to tip the result from the Conservatives.

‘Aim to win’ seats were even more skewed. ‘Other’ parties were key in all. None was ‘straight’ Conservative / Labour.

So, the critical audience was those voters favouring ‘others’ parties, principally Liberal Democrat or UKIP, or un-decided.

From ‘Swing’ to ‘Sway’

The concept of ‘sway’ superseded ‘swing’. Rather than swinging Labour voters to Conservative, the aim was to sway those inclined to ‘others’ or undecided.

‘Swing’

Lab$ Con$

‘Sway’

Lab$ Con$ Green$ UKIP$ Lib$Dem$ Un2decided$

Earning the ‘benefit of the doubt’ amongst swayable voters was to be the key battleground.

As with the base strategy, economic competence was a strong influence here, notably amongst Liberal Democrats11.

Importance of The Economy to Voting Intention

Single most important issue that will influence how you vote (%)

30" All"Voters" Lib"Dem"

20"

10"

0" Economy" Immigra1on" Healthcare" Cost"of"living" Welfare" Educa1on" Other" Housing" Pensions" Crime" Defence" Environment" Source: BMG, May 2015

9 This uses two different sources at two different slightly different points in time relative to polling day. 47% of 2015 voters hadn’t definitely decided how they would vote four months before polling (including 15% who claimed not to even know whether or not they had decided!) 10 There were 274 seats where Labour and Conservative finished first and second in 2010. 11 The relatively recent emergence of UKIP meant it wasn’t possible to break out data on it in much of the polling. 12 But a second message emerged from analysing the ‘suitability for Prime Minister’ theme.

‘Others’ voters were not keen on Ed Miliband.

‘Others’ Voters View of Suitability for Prime Minister

38#

Cameron# Milliband#

26#

7# 6#

Lib#Dem# UKIP# Who would make the best Prime Minister (%) Amongst all intending to vote for ‘others’ parties Source: YouGov Dec 2014

But there were others they wanted to see in Downing Street even less. The SNP and Alex Salmond in particular, were critical liabilities.

Trust in Other Leaders

Net trust in leader (%)

9

5 3

Sturgeon Farage

Cameron Clegg Miliband

-7

-13 Source: BMG

Other Figures’ Suitability for PM

Would make best Prime Minister (%) 34

20

7

Boris Johnson Nigel Farage Alex Salmond Source: BMG

13 The Detriment of the Doubt

Again, context was critical.

With all polls and commentators predicting a close result, 70% of voters believed there would be a ‘hung’ parliament12.

Labour could then only govern with another party’s co-operation, most likely the SNP, who were tipped to become third largest party13.

The SNP made clear their eagerness to co-operate with Labour and ‘lock the Tories out’.

“I call on Labour today to match that pledge, to make clear that if Labour and the SNP combined have more seats than the Tories, they will join forces with us in a vote of confidence to lock David Cameron out of Downing Street.”

Nicola Sturgeon, Keynote Election Speech, March 28th 2015

‘Alex Salmond: I would bring down any Tory minority government’

New Statesman March 2015

“The Tories would have to go for a vote of confidence and we’d be voting against. So if Labour joins us … then that’s Cameron locked out.”

Alex Salmond, quoted in New Statesman 24 March 2015

SNP Election poster

However the public, particularly our audiences, were firmly against the SNP holding any balance of power.

Party Preferred to Hold Balance of Power

If there was a , which of the smaller parties would you most like to see having the balance of power? (%)

All voters All non-Labour voters 44" 37" 42"

26"

20" 17"

11"

3"

Liberal UKIP Not sure SNP Liberal UKIP Not sure SNP Democrat Democrat Source: YouGov

12 Source: YouGov, April 2015 13 At the time, the SNP were forecast to win 42 seats and the Liberal Democrats 18. The Liberal Democrats were unlikely to hold sufficient seats to confer the balance of power, and had, in any case, already intimated that they would not form a coalition with Labour. 14 In particular, ‘others’ voters, especially those in Target seats, objected strongly to the prospect of a Labour government propped up by the Nationalists.

Favourability of a Labour government Relying on SNP Support

Would it be a good thing or a bad thing if there were a minority Labour government that relied upon SNP support? 91# (%) Split by voting intent 81#

Good#thing# Bad#thing# 71#

48#

34#

20#

9# 3#

Labour# Lib#Dem# UKIP# Conserva

This issue was about more than the Union. Voters in Target seats saw the SNP as something of a bully, intent on diverting money, initiative and influence away from their constituency to a Scottish agenda.

In short, they worried about Ed Miliband ending up in Alex Salmond’s pocket.

The Up-weight Campaign

The Conservatives’ adopted the theme of ‘A vote for anyone else risks a Labour government propped up by SNP’.

Research indicated a ‘swayable’ minority open to this argument. This was some 2.4m voters nationally, with around 400,000 in Target seats, heavily skewed to ‘others’ supporters14.

Potential Sway Voters

Millions of likely voters

Already intending to 10.6 vote Conservative

17.6

Not intending to vote Conservative, but think hung Parliament is likely, 2.4 think Labour would do a deal with SNP, this would be a bad thing and under these circumstances would prefer a Cons government Source: YouGov April 2014

Posters, digital advertising and increased campaigning were deployed in Target seats to win the benefit of the doubt amongst this audience,

They made the key messages simple, direct and engaging.

14 Source: YouGov April 2015. 7.9% of the sample agreed ‘Not intending to vote Conservative’ AND ‘Think a hung parliament is likely’ AND ‘Think Labour would do a deal with SNP’ AND ‘Think this would be a bad thing’ AND under these circumstances ‘Would prefer a Conservative government’. 15

Poster Advertising

16 Digital Advertising

17 Amplification

Although running only in Target seats15, advertising generated national exposure via intense media interest.

Examples of Broader Media & Social Coverage

Media Reporting of the Labour / SNP Agenda

Number of press headlines about SNP and Labour, Miliband and Sturgeon or Miliband and Salmond 1200"

1000"

800"

600" Poster first 400" appears

200"

0" 08-Jan 15-Jan 22-Jan 29-Jan 05-Feb 12-Feb 19-Feb 26-Feb 01-Apr 08-Apr 15-Apr 22-Apr 29-Apr 04-Mar 11-Mar 18-Mar 25-Mar 06-May Source: Lexis Nexis

15 Paid for advertising only ran in Target seats, except on two occasions, when ‘Ad Vans’ were deployed in Westminster for PR purposes. 18 The issue rapidly became a key political agenda, prompting Labour to rule out a formal coalition, in an unsuccessful attempt to draw a line under the argument.

Source: TheGuardian.com 16 Mar 2015

This was the first time a poster had achieved a political direct response.

Source: TheGuardian.com 17 Mar 2015

Further campaigning fuelled speculation of a non-coalition ‘arrangement’, adding further detriment of the doubt.

19 Examples of Further Campaigning

Telegraph, 10 April 2015

Continued Media Speculation

Telegraph, 11 April 2015 ITV, 19 April 2015

Social media, 4 April 2015

Mirror, 17 April 2015

20

Eve of the Election

By the eve of polling, there had been a clear effect on perceptions.

Increased numbers trusted the Conservatives with the economy, with support having risen on that issue more than non-campaign themes.

Satisfaction with Managing the Economy

Do you think the Coalition is managing the economy well or badly? (%)

49 % 48 % 45 % 44 % 42 % 40 %

Campaign

Source: YouGov

Shifts in Trust by Issue

Would handle issue better Conservative lead over Labour (%) 20

May$14' May$15' 15

10

5

0 Economy Law & Order Taxation Asylum & Unemployment Europe Education & NHS immigration schools

-5

-10

-15

Source: You Gov

Trust in Handling the Economy

Which party would best handle the economy in general? (%) 40%

35 %

22 % 22 %

11 % 8 %

4 % 4 % Campaign

Source: YouGov

21 Increased numbers trusted Cameron as leader.

Trust in Leadership

Which if any of the following would you say you trust more (%) 40 %

33 %

31 %

23 % 21 % 21 %

5 % 6 % Campaign

Source: Opinium

Satisfaction with David Cameron

Satisfaction / dissatisfaction with David Cameron as leader (%) 60#

50#

40#

30# Dissatisfied

Satisfied Campaign

20# 01-Oct 01-Nov 01-Dec 01-Jan 01-Feb 01-Mar 01-Apr 01-May Source: Ipsos Mori

Concern at a possible Labour / SNP arrangement had increased.

Shift in Disapproval of Labour / SNP Arrangement

Would it be a good thing or a bad thing if there were a 42" minority Labour government that relied upon SNP support? 37" Early Late

21" 21"

16" 14"

8" 6"

Very Good Fairly Good Fairly Bad Very Bad Source: YouGov 22 Significantly, these shifts were strong amongst intending Liberal Democrat and UKIP voters.

Conservatives Trusted to Handle the Economy by ‘Others’ Voters

Trust Conservatives more to handle the economy amongst those intending to vote for UKIP and Lib Dem (%)

UKIP Voters 37 %

34 % 31 %

Lib Dem voters 27 %

Campaign

Source: YouGov

Shift in Disapproval of a Labour SNP Government

Would be dismayed at a Labour SNP Government (%) amongst target groups 80

Early Late 70

60

50

40

30

20

10

0 All Likely to vote Lib Dem Likely to vote UKIP Expect hung Parliament Source: Com Res

Polling specifically in marginal seats suggested the key messages had cut through more in those targeted with advertising.

Key Measures in Up-Weighted Seats

All voters agreeing (%)

70

Non-upweight Up-weight 60

50

40

30

20 Satisfied with Cameron Trust Conservatives with Disapprove of Labour SNP economy government Selected marginal seats only Split into Target and non-Target Closest poll in seat before Election Source: Ashcroft polling, Survaction

The Conservative’s Research16 indicated this would win the benefit of the doubt.

But this belief was not shared by most.

16 The Conservatives invested significantly in private polling and campaign strategy, via Crosby Textor. This research remains confidential and this paper only references information already in the public domain. 23 A Poor Forecast?

Polls and commentators unanimously forecast the parties tied and Parliament hung17.

This was largely based on absolute levels of claimed voting intent, versus Labour nationally. This wasn’t reflective of the Conservative strategy.

Nevertheless, there were indicative patterns supporting Conservative confidence.

The majority of final polls suggested more of those who voted Conservative in 2010 intended to do so again18.

Voting Intention by 2010 Vote

Proportion of those claiming to have voted for party in 2010 intending to vote for that party again in 2015 (%)

90

Conservative Labour

80

70

60

50 Survation Survey YouGov Ipsos Ashcroft Com Res ICM Opinium BMG Populus Monkey MORI Sources: As shown

In addition, a significant minority of those who chose ‘others’ indicated they would now vote Conservative19.

Voting Intention by 2010 Vote

Proportion of those claiming to have voted Liberal Democrat in 2010 intending to vote Conservative / Liberal Democrat in 2015 (%)

50

Conserva)ve* Liberal*Democrat* 40

30

20

10

0 Survation Survey YouGov Ipsos Ashcroft Com Res ICM Opinium BMG Populus Monkey MORI Sources: As shown

With these indications stronger in up-weighted than non-upweighted seats, there was directional evidence within the polls of the benefit of the doubt at work.

It was now down to the only poll that mattered.

17 Poll of polls May 6th 2015. Conservative 34%, Labour 33%, UKIP 13%, Liberal Democrat 8%, Green 5% 18 Sources: All published final polls. Proportion of those claiming to have voted Conservative in 2010 intending to vote Conservative again in 2015, compared with the equivalent for Labour. Note that some polls factor-in 2010 claimed voting to simulate don’t knows’ intent. 19 Sources: All published final polls. UKIP vote was far smaller in 2010, so it is only possible to show this analysis for Liberal Democrat 24 The Result That Mattered

On May 7th the country voted.

Counter to prediction and precedent, the Conservatives attracted a 5.7% increased vote, gaining 24 seats to secure a majority20.

The Final Result

Source: BBC

Shocked at the outcome, a number of commentators were made to eat their hats!

Commentators Eating Their Hats

Source: BBC

20 Source: BBC. Data from Returning Officers. The Conservatives vote of 11,334,576 was 5.7% up on the 10,726,614 it polled in 2010. David Cameron announced a majority Conservative government on May 8th. 25 The campaign had secured base votes, with Conservative share up 0.8%.

Further insight into how that was achieved comes from re-contact surveys, comparing actual with intended voting21.

This reveals that around 15% who stated an intention to vote either didn’t, or changed their choice.

A number of polls show marginally more people claiming to vote Conservative than said they would.

Actual Versus Intended Voting

Claimed Conservative vote minus claimed Labour vote (%) compared immediately before and after Election

Before After election election

Sample of respondents known to have voted Matched before and after election Intended versus actual voting Data from six independent polls Source: NCRM

The Conservatives succeeded in ensuring a greater proportion of their supporters actually voted, with more of them voting Conservative.

Intended, but Did Not Vote

Proportion of intending voters subsequently reporting they did not vote (%) Split by party originally intending to vote for 6.9%% 6.8%%

6.4%%

5.8%%

Cons% Labour% Lib%Dem% UKIP% Source: Survation

Voted as Intended

Proportion of intending voters subsequently reporting they voted for the party they intended (%) Split by party originally intending to vote for

87.7%

83.7% 82.6%

73.6%

Conservative Labour Liberal Democrat UKIP Source: Survation

21 A number of polling companies re-contacted their samples on or after May 7th. This allows us to correlate intended with actual voting. Both are claimed data, but nevertheless provide directional insight. 26 This meant an increased Labour vote was insufficient to close the gap, with Labour still 6.5% adrift.

The net effect on seats was near neutral.

Conservative Gains / Losses vs. Labour

Seats won / lost by the Conservatives

8"

Lost to Labour

Won from Labour

#10"

Source: Returning Officers

In line with the strategy, there was a net decrease in votes for ‘others’ parties. This reflected a marked decrease in Liberal Democrat vote and smaller than feared increases in UKIP and Green votes, offsetting the predicted SNP landslide.

Fall in ‘Others’ Vote

Total share of vote for parties other than Labour and Conservative (%)

40

35

30

25

20

15 1992$ 1997$ 2001$ 2005$ 2010$ 2015$ Source: BBC

Re-contact polls indicated a greater proportion of people intending to vote for ‘others’ hadn’t in the end, with more of them swaying to Conservative.

Proportion of Others Intenders Switching

Proportion of Liberal Democrat and UKIP intending voters subsequently reporting they actually voted for Labour or Conservative (%) Split by party originally intending to vote for

Conservative Labour 7.7%

6.1%

4.7%

2.6%

Liberal Democrat UKIP Source: Survation 27 We noted the larger than normal level on ‘un-decideds’. Around a quarter of voters claim to have made their decision on poll day or in the last few days22.

Unsurprisingly, a larger proportion of ‘un-decideds’ didn’t vote, but more of those that did appear to have voted Conservative.

How Late Deciders Voted

Party claimed to have actually vote for (%) Amongst those making up their minds ‘on polling day’ or ‘A few days before’ 32.9

25.7

21.7 19.7

Did not vote Conservative Labour Other

Source: Greenberg Quinlan Rosner / Ashcroft

This pattern, when translated into seats, meant the Conservatives minimised possible losses to UKIP and Green, whilst gaining heavily from Liberal Democrats.

Conservative Gains / Losses vs. Others

Seats won / lost by the Conservatives 27#

Lost to UKIP 0#

!1# Lost to Green Won from Lib Dem

Source: Returning Officers

Having, shown how the campaign as a whole worked, we now make a case for advertising specifically, by examining up-weighted Target seats.

Up-weighted Results

With votes cast and counted in 650 individual seats, we can make comparisons between different sets of constituencies, pinpointing the influence of advertising and up-weighted campaigning.

As we’re dealing with gross actual votes, not polls or samples, there’s no statistical or sampling error to consider here.

22 Sources: Ashcroft, Survation, ComRes. Between 11% and 17% claim to have decided on polling day. Between 12% and 15% claiming ‘in the last few days’. 28 Comparing up-weighted seats with non-upweighted reveals a pronounced incremental uplift in Conservative vote share23.

Change in Conservative Vote Share

Change in Conservative share of vote 2015-2010 (%pts) 3.6 in up-weighted versus non-upweighted seats

0.3

Non-upweighted seats Up-weighted seats Source: Returning Officers

Given the strategy, the effect on ‘others’ voters is particularly revealing.

The analysis shows greater decline in Liberal Democrat support, and lower rises in UKIP and Green support where advertising ran.

This is consistent with greater numbers choosing to vote Conservative rather than for ‘others’.

Changes in Voting Share in Advertised Seats

Change in share of vote 2015 - 2010 (%pts) in up-weighted and non-upweighted seats by party 9.8 9.3 Non-upweighted seats Up-weighted seats

3.6 2.9 2.3 0.3 Lib Dem

Conservative UKIP Green

-15.1 -16.7

Source: Returning Officers

These patterns tally with analysis of individual constituency polling.24

We look at the uplift in Conservative share between final poll predictions and actual votes cast. Splitting this between up-weighted and non-upweighted seats reveals the same pattern of incremental Conservative uplift driven by reduced ‘others’ voting.

23 Votes are aggregated into two groups, those cast in all up-weighted seats and those in all remaining, non-upweighted seats. This allows a side by side comparison to be made. 24 Ashcroft Polling conducted polls throughout the campaign in individual marginal. Some, but not all of the seats polled by Ashcroft fell in the Conservatives’ Target list and therefore received up-weighted support and advertising. For this analysis, we have aggregated results into those for seats in the Target, and those not. This analysis only covers some of the target seats, ie. those polled by Ashcroft. 29 Actual Voting versus Polling Results by Seat

Difference in the Election result vs. final constituency poll (%pts) By Up-weighted / Non-upweighted seat

10.3

BaseNon- upweight Upweight

5.9

Labour UKIP Green Liberal Democrat

Conservative -0.4 -0.9 -1.0 -1.4 -1.5 -2.3 -3.2 -3.7

Difference in the Election result vs. final constituency poll (%pts) Source: Ashcroft Polls

Again, this indicates a pattern of ‘others’ voters giving the Conservatives the benefit of the doubt.

These results also correlate with the shifts in the key benefit of the doubt drivers shown earlier.

In particular, dissatisfaction with a possible SNP-backed Labour government ran especially high in up-weighted seats.

The net effect on Target seats was pronounced. The Conservatives won 76% of their ‘aim to win’ seats, accounting for all but one of their gains. They lost only 18% of their ‘must holds’. To put this in perspective, Labour won just 9% of their hit list.

Source of Seat Gains

Seats won / lost by the Conservatives 34

1

Up-weighted non-upweighted Source: Returning Officers

Notably, we aren’t alone in concluding advertising and campaigning swayed voters.

In accounting for their failure, the opposition parties all publicly cited the same effect. Some specifically reference Conservative advertising as a factor.

“It was heard consistently on the doorstep that this scaremongering raised concerns. It may have reinforced the views of those who had already decided not to vote Labour, and may have had a decisive impact in a small number of constituencies.”

Margaret Beckett, Labour Taskforce Report

“One close Miliband adviser admitted he was initially perplexed why posters were appearing all over English towns which depicted Miliband in the pocket of Alex Salmond. “They stumbled on this SNP thing. We did not realise how much impact it would have.”, the close advisor said. “It was going to persuade 2.5% of the [electorate previously allied to UKIP] to go back into the Tory fold. It made us the risk.”

The Guardian, 4June 2015

30 ‘Despite our failings, we might not have done so badly except for the strength of the late Tory message about the weakness of Ed Miliband and English voters’ fear of the SNP in control in Westminster. No party had a response to this … Liberal Democrat chances were simply swept away.’

Liberal Democrat Election Review

‘All of us really felt a Labour government dancing to the tune of the SNP really chilled the English heart and you could see lots of people that would traditionally have voted for the Liberal Democrats playing it safe and saying the best guarantee (against) was to vote Conservative.’

Nick Clegg, radio interview on LBC

‘The results tend to indicate that many Conservative supporters who may have considered UKIP cast their votes for David Cameron for fear of the ‘coalition of chaos’, with ‘Labour dancing to SNP’s tune’. The ‘Vote UKIP, Get Labour’ message persuaded some UKIP supporters to ‘go home’ to the Tories rather than risk handing power to Alex Salmond.’

UKIP’s Performance in the 2015 General Election

“It was heard consistently on the doorstep that this scaremongering raised concerns. It may have reinforced the views of those who had already decided not to vote Labour, and may have had a decisive impact in a small number of constituencies.”

Margaret Beckett, Labour Taskforce Report

“One close Miliband adviser admitted he was initially perplexed why posters were appearing all over English towns which depicted Miliband in the pocket of Alex Salmond. “They stumbled on this SNP thing. We did not realise how much impact it would have.”, the close advisor said. “It was going to persuade 2.5% of the [electorate previously allied to UKIP] to go back into the Tory fold. It made us the risk.”

The Guardian, 4June 2015

‘Despite our failings, we might not have done so badly except for the strength of the late Tory message about the weakness of Ed Miliband and English voters’ fear of the SNP in control in Westminster. No party had a response to this … Liberal Democrat chances were simply swept away.’

Liberal Democrat Election Review

‘All of us really felt a Labour government dancing to the tune of the SNP really chilled the English heart and you could see lots of people that would traditionally have voted for the Liberal Democrats playing it safe and saying the best guarantee (against) was to vote Conservative.’

Nick Clegg, radio interview on LBC

‘The results tend to indicate that many Conservative supporters who may have considered UKIP cast their votes for David Cameron for fear of the ‘coalition of chaos’, with ‘Labour dancing to SNP’s tune’. The ‘Vote UKIP, Get Labour’ message persuaded some UKIP supporters to ‘go home’ to the Tories rather than risk handing power to Alex Salmond.’

UKIP’s Performance in the 2015 General Election

A Significant Effect

Scale of effect shouldn’t be confused with significance of effect.

With this marginal ‘sway’, the Conservatives won seats they wouldn’t have done without advertising.

We show this by reanalysing individual constituency counts, calculating a projected result by applying a shift to the 2010 vote in that seat. We simulate the effect of withdrawing advertising from Target seats, by applying the lower vote shift observed in non-upweighted seats.

This indicates the Conservatives would not have won 17 of the seats they did, loosing their majority.

Accounting for Other Factors

None of the myriad other factors at play, fully explains this significant effect.

Labour Opposition

Sceptics might argue a weak Labour offering gifted the Conservatives victory.

This wasn’t the case. Labour actually improved their appeal versus 2010.

Amongst Labour supporters, ‘suitability for PM’ and ‘trusted with the economy’ scores were strong.

Trust in Labour Amongst Supporters

All agreeing Labour would be best at handling the economy / Miliband would make best PM

53 All voters Labour voters

47

22 19

Handling the economy Make best PM

Source: Com Res

31 Labour also made substantial gains on the policy areas they focused on.

Labour Trust by Issue

Agreeing Labour would handle issue better (%)

40" MayC14" MayC15"

30"

20"

10"

0" Economy"" Law"&"Order" Taxa7on" Unemployment" Educa7on"&" NHS" schools" Source: You Gov

As a result, Labour increased their vote on 2010. However, the source of that increase was significant.

Broadly, the themes Labour campaigned most on appealed most to core Labour supporters, meaning Labour increased support largely amongst those already inclined their way. We see Labour’s largest vote share increases occurring where they were already strong.

The Conservatives prioritised their own supporters and ‘others’ voters. Labour’s performance was ultimately weaker amongst these audiences, making their share increase in more closely fought seats modest.

Skew in Increased Labour Support

Labour change in share of vote 2015 – 2010, by seat type

All safe All closely All seats Labour seats contested seats

Change in vote share +1.5 % +7.0 % +0.6 %

Source: Returning Officers

So, the picture was less of Labour being ineffective and more about the Conservatives being most effective where it mattered most.

Liberal Democrat Collapse

An alternative explanation is wholesale defection from the Liberal Democrats, following their record in Coalition, Conservatives simply picking up their votes and seats.

Whilst the scale of the Liberal Democrat decline suggests disapproval, it’s somewhat perverse to protest against a party’s record in government by voting for their partner in that government.

It’s certainly disingenuous to suggest the Conservatives were simply gifted deserting Liberal Democrats’ votes. We illustrate this with further seat by seat analysis.

The Conservatives fought 208 seats ‘versus’ Liberal Democrats nationally 25. A small proportion of these were Target seats.

25 Seats where the Conservatives and Liberal Democrats were positioned first and second in 2010, in either order 32

Directly comparing up-weighted versus non-upweighted Conservative / Liberal Democrat seats, the effect is visible.

All Conservative / Liberal Democrat seats show a greater decline of Liberal Democrat support as there was a larger vote to lose26.

Lib Dem Share of Vote Decline in Conservative / Lib Dem Seats

Change in Liberal Democrat share of vote 2015 – 2010 in Conservative vs Liberal Democrat fought seats (%pts)

Comparison of Up-weighted and non-upweighted seats

Non-upweight Up-weight

-17.8 -18.6 Source: Returning Officers

But the distribution of that lost support is revealing.

In non-upweighted seats, Labour gained more than the Conservatives. Whereas in up-weighted seats, the Liberal Democrat decline was greater and the Conservatives increased their share of vote twice as much, whereas Labour gained less.

Share of Vote Uplift in Conservative / Lib Dem Seats

Change in Labour and Conservative share of vote 2015 – 2010 in Conservative vs Liberal Democrat fought seats (%pts)

Comparison of Up-weighted and non-upweighted seats 4.1

Labour Conservative

3.0

2.2 2.1

Non-upweight Up-weight Source: Returning Officers

UKIP

In the end, UKIP grew support, attracted 12.6% of the vote, but gained only one seat. Their failure to win more helped secure the Conservative majority.

However, we showed earlier that UKIP support grew less in seats the Conservatives up-weighted.

Seat by seat remodelling indicates that if vote shares had increased at the non-upweighted level instead, UKIP would have effectively tipped 2 extra seats away from the Conservatives, pressurising their majority.

26 In these seats, the Liberal Democrats had, by definition, received enough votes in 2010 to make them either first or second party. 33 SNP

Undoubtedly the SNP had a successful election. But that success didn’t put the Conservatives into power.

The prospect of SNP having balance of power swayed non-Scottish voters. But in the end the Conservatives won an outright majority and SNP seats didn’t materially influence the outcome.

Even if the SNP landslide had not happened and Labour hadn’t lost any seats in Scotland, they still wouldn’t have held enough to prevent a Conservative majority.

Projected Seat Allocation without SNP Rise

Distribution of seats in the house if SNP had performed at 2010 levels

Source: Returning Officers

Policy

We have purposefully not discussed policy thus far, instead presenting evidence on brand preference.

The indications are that no party’s policies set them significantly ahead. Each party had ‘star’ policies that were liked, in aggregate no party appeared to hold a net advantage27.

Support for Party Policies

Which party has the best policies and ideas for Britain Amongst whole sample and supporters of each party

62 60

Conservative Labour

26 24

All Supporters

Source: YouGov

27 Source: YouGov / The Times. Each party had popular policies, but none individually or collectively more than their opposition. Aggregating each party’s most preferred policies shows no sizeable advantage. Data for the top 10 policies was very slightly more favourable towards Labour. 34 Opinion of Party Policies

Average agreement / disagreement with each party’s top 5 policies

66 66

Conserva)ve* Labour*

22 19

Good idea Wrong Priority Source: YouGov

Moreover, policies are largely set nationally, outlined in each manifesto. The popularity of those policies was fairly evenly distributed geographically.

Opinion of Party Policies by Geography

Average agreement with top 5 policies for each party by regionality of voter

90

Conservative Labour 80

70

60

50

40

30 National London South Midlands / North Scotland Wales Source: YouGov

Whatever its influence on national voting patterns, policy couldn’t explain local seat by seat differences, specifically, those between up-weighted and non-upweighted.

There were policies issues pertinent to individual constituencies, but our analysis aggregates multiple seats, minimising the influence of individual local factors

Other campaigning

General Elections are giant exercises in integrated (sometimes disintegrated) communications.

In this case every admissible communications channel was aligned behind the strategy, to good effect.

The question is whether there were substantial differences between up-weighted and non- upweighted seats, beyond paid-for advertising.

Certain factors, such as leafleting or ministerial visits, were capped under election rules or simple resource availability. These are discussed later.

Others, such as TV debates or media influence, were broadly national in impact and wouldn’t explain variations in voting patterns by seat. 35 Target seats did receive up-weighted campaigning, principally on the doorstep and telephone. However, opposing parties increased their effort in these seats too, nullifying any advantage.

In the absence of a metric for auditing weight of campaigning, we examine voters’ own reporting of their experience of approaches from parties.

From this voter impact perspective, Conservative ‘ground campaigning’ was visible, but less so than Labour’s.

Voter Experienced Weight of Campaigning

Whether have been contacted over the last few weeks by party Includes by leaflet, letter, e-mail, telephone or door to door (% contacted and party contacting most) 90

80 Been$contacted$ Most$contact$

70

60

50

40

30

20

10

0 Conservative Labour Liberal UKIP Deomocrat

Source: Ashcroft Polls

Likewise, the pattern is similar between up-weighted and non-upweighted seats.

Voter Experienced Weight of Campaigning

Whether have been contacted over the last few weeks by party Includes by leaflet, letter, e-mail, telephone or door to door (%) Target vs. non-target seats 90

80 Target'seats' Non,Target''

70

60

50

40

30

20

10

0 Conservative Labour Liberal UKIP Deomocrat Source: Ashcroft Polls

Other factors

Briefly, we can dismiss a number of other possible factors.

Turnout was up a modest 1%. Higher turnout disproportionately favours Labour.

Weather was fair, also traditionally favouring Labour.

Postal votes were included fully in all votes analysed above.

Target seats were distributed all over England & Wales, with no pronounced geographical bias.

36 Media influence was largely national.

Any Scottish Referendum effect is covered in comments about SNP.

So, we can reasonably claim advertising made the difference that made all the difference.

We now examine the loaded question of its return.

Return on Investment

We’re acutely aware how much election outcomes matter to the country and electorate. Or indeed that not all would agree the outcome was beneficial.

So it’s not only speculative, but divisive to estimate any longer-term economic or social value.

Rather, on a binary ‘succeed or fail’ basis, the campaign was clearly worth doing from the perspective of those paying for it. But did it represent reasonable utilisation of marketing assets?

RUMA

The Conservatives spent less overall and less on advertising than in 3 previous elections, against 45% greater competitive spend28.

Spend versus Competition

Total campaign spend and share of voice for Conservative (£m)

Total Conservative 42.9

37.5

31.8

27.1

17.9% 16.7% 15.6% 12.8%

41.7% 47.2% 52.5% 41.6%

2001 2005 2010 2015 Source: Electoral Commission

This lower share of voice nevertheless secured 58% more ‘seats per pound’ than 201029.

A crude justification of efficiency, but could the result have been achieved through other, potentially cheaper, means?

Campaigning: Reliant on volunteers, door-stepping, leafletting and telephoning is indeed cheaper per contact than advertising. But reliance on volunteers limits scalability. Simply put, the Conservatives did as much campaigning as they could find willing supporters to do it. They could only have done more by paying campaigners, making the cost less favourable than advertising.

28 Source: Electoral Commission. Party campaign spend is capped by Election rules at £30,000 per seat contested and the Conservatives spent less than this. 6 major parties reporting spend over £250,000, together spending £21.9m (versus £15.5m in 2010). A further 42 smaller parties and 3 non-parties submitted smaller additional returns totalling £1.8m, not included in this analysis. 29 Sources: Returning Officers, Electoral Commission. Votes per pound was up 55%. Versus 2005, seats per pound were up 154% and votes per pound up 96%. 37 PR: Media coverage of the election was already at saturation. It’s far from certain that any additional cut-through could have been achieved. We also note that much of the PR was consumed by those interested in the Election and actively ignored by those not. Many of our swayable audience were in the latter camp.

TV programming and party political broadcasts: Broadcasters’ self-imposed impartiality and election rules make it impossible to secure more airtime.

Policy ‘giveaways’: Arguably, it might have been possible to secure more votes via ‘generous’ policies, such income tax or VAT cuts. Notwithstanding political desirability, the state of the economy and budget deficit simply precluded such actions. Indeed, they would have been of a different order of magnitude in cost compared with advertising.

Social and ‘Earned’ media; Social media was already a significant component of the campaign, primarily in the form of paid-for activity. Non-paid-for, or ‘earned’ social media falls under the same argument as PR. Although it should be noted elements of the campaign did achieve significant levels of social sharing and comment, as shown earlier.

Events and ministerial appearances; Maximum usage was made of ministers and PR opportunities. It would have been unrealistic to achieve any more.

Beginning sooner; By campaigning sooner, greater use could have been made of time-limited assets mentioned above. Knowing the election timing, Labour began their campaigning as early as 2013. However, the borderline voters we were seeking to influence didn’t engage until late in the campaign, some not until polling day. So it’s unlikely earlier campaigning would have had any influence.

Altogether therefore, we conclude the campaign’s utilisation of assets was entirely reasonable.

How Atypical?

Having outlined the category’s oddities, consider this.

Here was a traditional market impacted by the rise of new competition30.

Careful analysis revealed the old enemy no longer represented the best source of growth.

Instead, effort went ruthlessly into winning the benefit of the doubt amongst middle-ground customers torn between newer, fresher alternatives and the familiar devil they knew.

Resources were orchestrated31 into a base trust message across all channels, supplemented by a sharp fighting theme in advertising, it’s poignancy, obliqueness and clarity surprising everyone.

Ultimately a billboard using just two words said it all, helping wrong-foot competitors and shape the outcome.

The art of sacrifice requires both rigour and courage. If nothing else, if this story imparts a little of both, it will have succeeded.

Words 3736

30 With apologies to political historians for including Liberals amongst the new. 31 Particular acknowledgement is due to Crosby Textor for rigorously orchestrating the orchestration. 38