The judgment of Mr Justice Warby in Monroe -v- Hopkins [2017] EWHC 433 (QB)  has already attracted a lot of attention.   Here I want to look at the issues relating to the evidence.  The case is one of the first that deals with defamation by twitter. There are important issues in relation to the retention of evidence and Twitter analytics in particular. Another important factor was the very early offer from the claimant to settle the action, that offer was ignored by the defendant.


The action related to what were found to be two defamatory tweets by the Defendant.    The claimant requested the first tweet be deleted (and it was). Afterwards there was an offer from the claimant to the defendant.

“At 8.14pm Ms Monroe tweeted again, this time using Ms Hopkins’ Twitter handle: “Dear @KTHopkins, public apology +£5k to migrant rescue & I won’t sue. It’ll be cheaper for you and v. satisfying for me.”

No damages were forthcoming and the claimant issued damages for libel.


“The trial has been short. The only oral evidence has been that of Ms Monroe, who has been cross-examined for the best part of a day. Mrs Hopkins has not given evidence. She relies on two witness statements made by her solicitor, Ms Harris, analysing a body of twitter records. Those statements have been admitted into evidence by agreement, without cross-examination, though without accepting the accuracy of the analysis. Ms Hopkins also relies on facts and documents that have been put before the Court by agreement.”


The judge had to consider the numbers who had read the offending tweets.  Problems occurred because the first defamatory tweet was deleted and precise analytics were not available [see the section below]. The claimant had to make estimates of the likely readers.
“57. Mr Price submits that the claimant’s estimates are no better than guesses, with no solid basis. He argues that any substantial traffic to his client’s home page in the 2 ½ hours at issue is likely to have been from followers of Ms Monroe and followers of Ms Penny, drawn by the offending tweet. He submits, and I accept, that there were few who followed both Ms Penny and Ms Hopkins. Further, says Mr Price, it would be wrong to assume that every impression amounts to a viewing of a tweet. The claimant’s case, he submits, is at the very top of the range of conjecture. He argues that it would not be safe to put any figure on the readership, and that Ms Monroe has not satisfied the burden of proof. Alternatively, he invites me to find there was some 1 See How Twitter Works  relatively small publication in the high hundreds or just into four figures, made up mostly of followers of Ms Monroe and Ms Penny.
58. Precision is of course impossible, but nor is it necessary. It is enough if I can make a sound assessment of the overall scale of publication. The submissions for Ms Hopkins that this cannot be done are not only unattractive but also unrealistic in my view. I am satisfied that the readership of the First Tweet was comfortably into five figures. An estimate of around 20,000 seems to me to be entirely reasonable, and much more likely to be an under-estimate than an over-estimate. The Twitter Analytics that are available show that individual tweets over the 25 hour period I have mentioned generated between 21,000 and 252,000 impressions, and between 198 and 14,196 engagements. The higher of these figures are those for the Second Tweet. Although it was not retweeted as often as some others, it was clearly “popular” and by a fair margin the tweet with the most impact of all those dealt with in these records.
59. An “impression” does not mean that the person on whose screen the tweet appeared actually read it. The figure must be discounted to arrive at an estimate of readers. Mr Bennett allows a discount of 60% to arrive at a figure of 100,000 readers for the Second Tweet. I accept that approach. The figures for the Second Tweet are significant so far as the First Tweet is concerned. They indicate the probability that the figures for the First Tweet were high. The analytics show that other tweets in the evenings generated six-figure numbers of impressions.
60. Mr Bennett’s other method of estimation leads him to a figure of around 25,000 readers, by this route: (a) he divides the 5.74m profile views in May 2015 by the 31 days of the month to arrive at an average of 185,161 views per day; (b) he divides that by 18, on the footing that the audience was a domestic one, and would spent some hours asleep, arriving at an average of 10,286 views per hour; (c) he multiplies that by the number of hours the First Tweet was available to view. This is an approach that starts with a known and undisputed figure and applies a rational analysis to it. It may perhaps overstate the position a little. For instance, the scale of views on particular days may depart from the average, for all we know. But taking account of what the Twitter Analytics show, I consider that the 20,000 figure I have mentioned is a safe conclusion.
61. I accept that the First Tweet was probably re-tweeted extensively. It is impossible to estimate the extent, except to say that the evidence suggests it was probably in the hundreds. It is not necessary to go further than this in analysing the scale of the likely readership of the First Tweet. These conclusions are enough to enable me to decide the other issues that I have to address. But I add that, though it would be idle to engage too much in analysis of the Twitter Analytics, the document does show the text of individual tweets, so it is possible to compare the higher figures for impressions and engagements with the content. That comparison seems to me to support the conclusions I have reached on the other bases mentioned. 62. As for the Second Tweet, the Twitter Analytics tell the story clearly enough. I have set out the figures already. They are clearly substantial. It was re-tweeted 40 times.”


By way of observations at the end of the judgment the judge stated:-
“83. This case has been about the particular tweets complained of by this claimant against this defendant. It may have little wider significance. But I cannot leave it without making two observations. The first is that the case could easily have been resolved at an early stage. There was an open offer to settle for £5,000. It was a reasonable offer. There could have been an offer of amends under the Defamation Act 1996. Such an offer attracts a substantial discount: up to half if the offer is prompt and unqualified. Such an offer would have meant the compensation would have been modest. The costs would have been a fraction of those which I am sure these parties have incurred in the event. Those costs have largely been incurred in contesting the issue of whether a statement which on its face had a defamatory tendency had actually caused serious harm.
84. The second point is that there have been difficulties over disclosure especially on the claimant’s side, of which others should take note. The deletion of the First Tweet, at Ms Monroe’s request, meant the Twitter Analytics were unavailable. And Ms Monroe’s Twitter records were extensively deleted. I am not able to attribute responsibility for that on the basis of the evidence, and I do not. What I can say is that this highlights in the Twitter context the responsibility of a litigant to retain and preserve material that may become disclosable, and the responsibility of a solicitor to take reasonable steps to ensure that the client appreciates this responsibility and performs it.”

There has been much discussion in the media in relation to the costs of the action itself.  There are newspaper reports that the defendant was ordered to pay £107,000 on account of costs.  However we are not told whether the case was costs budgeted and what percentage of the claim for costs that represents.