We use cookies to allow us and selected partners to improve your experience and our advertising. By continuing to browse you consent to our use of cookies. You can understand more and change your cookies preferences here.

Share this page

Broadband Speed Expectations


Advertising rules around the commonly used “up to --Mbps” wording are currently underpinned by the stipulation that 10% or more customers have to achieve the stated speed in order for broadband providers to be allowed to offer it.

The Committee of Advertising Practice (CAP) are looking to change this, and have proposed four new options for these qualifying statements informed by qualitative research[1]. These are:

- a median download speed (available to at least 50% of consumers) measured over 24 hours

- a median download speed (available to at least 50% of consumers) measured at peak-time

- a range of download speeds available to the 20th to 80th percentile of users measured over 24 hours

- a range of download speeds available to the 20th to 80th percentile of users measured at peak time

We conducted some online experimental research with a nationally representative sample of adults between June 13th and 16th to test the speed expectations people have when faced with these different speed claim descriptions, as well as with just the headline speed with no extra explanation.


This experimental research measured the broadband speeds participants expected to receive as a percentage of the stated headline speed for 5 types of speed claim statement.

1. xMb/s average download speed available to at least 50% of users, measured over 24 hours

2. xMb/s average download speed available to at least 50% of users, measured over peak-time (8pm to 10pm)

3. xMb package, typically between xMb/s and xMb/s measured over 24 hours

4. xMb package, typically between xMb/s and xMb/s over peak time (8pm-10pm)

5. Up to xMb

The actual speeds presented to participants were based on an analysis of Ofcom’s Home Broadband Performance Survey[2] (November 2016) for the three headline speeds we used.

Participants were split into three equal sized groups so results could be collected for Standard ADSL (20Mbps), Fibre to the Cabinet (FTTC) (76Mbps), and Cable (200Mbps) without introducing survey fatigue. We also tested Fibre to the Premises/Home (FTTP) (1,000Mbps), but as the nature of this technology meant that the measurement disclaimers were not relevant, the questions relating to it were asked of the whole sample in the same way as statement 5 only.

The expected speed data were collected as a percentage of the headline speed, with a scale offered of between 0% and 200%. The screenshots below gives an example of how the speed expectation data were collected from participants.


We designed this experiment in order to see whether speed expectations differ with different statement wordings (up to 50% get xMbps; or typical range of xMbps and Y Mbps), measurement times (24hour average, or peak time average), or delivery technology (i.e. ADSL, cable, FTTC, FTTP). We also looked at whether particular combinations of these wordings worked together to have a significant impact on speed expectation. The a-priori hypotheses were as follows:

1. Statement wording (i.e. available to 50%, or typical range of X and Y) will not affect speed expectations.

2. Measurement time range (i.e. over 24hrs, or at peak times (8pm to 10pm) will not affect speed expectations

3. Broadband technology (i.e. ADSL, FTTC, FTTP, Cable) will not affect speed expectations

4. There will be no significant interaction effects between the main factors in hypotheses 1) to 3) above.


The hypotheses set out above were statistically tested via two separate procedures. The first tested all the factors but included just ADSL, Cable and FTTC. The second compared FTTP to the other technologies independently because the speed disclaimers factors were not applicable to it, as there is no speed variation with this technology.

The main tests were conducted via a mixed repeated measures ANOVA that consisted of 2 ‘within groups’ factors (i.e. repeated for each participant), and one between groups factor (i.e. participants split into groups). The within groups factors were:

Statement type (3 Levels – 1. up to 50%, 2. typical range between xMbps and yMbps, 3. Straight xMbps)

Measurement time (2 Levels – 1. 24hr Average, 2. 8pm-10pm peak)

The between groups factor was:

Broadband type (4 Levels – 1.ADSL 20Mbps, 2. FTTC 76Mbps, 3.Cable 200Mbps, 4. FTTP/H 1,000Mbps).

Across all the conditions together, we found that people expected to receive around 91% of the headline speed, and that regardless of the different conditions we manipulated in this research, that figure never fell below 83%[3], even surpassed the headline for some combinations of factors, the maximum being 110% of the headline[4].

The statistical tests found that statement type and broadband type had significant effects upon speed expectations, whereas measurement time only changed expectation when as an interaction with broadband type.

More specifically, of the speed claims the unqualified statement (as in example 5 in the method section) attracted the most optimistic speed expectations at 94%[5]. The typical range statements (between xMbps and yMbps) did not differ significantly from the unqualified headline speed (92% vs. 94%), whereas the “available to 50%” statement attenuated expectation to around 89% of the headline speed given (significantly different from the unqualified statement)

Expected proportion of headline speed differed by broadband types with FTTP and Cable both expected to slightly exceed the headline speed on average (both 107% of the headline), Standard ADSL was expected to deliver 85% of the headline speed, and FTTC 83% of the headline speed. Cable and FTTP were both significantly different from ADSL and FTTC respectively.

The interaction between broadband type and the time measure (24hr or peak), showed that for ADSL and FTTC the peak 8pm to 10pm statement resulted in a very slightly closer expectation to the headline with the peak time condition, whereas the opposite was true for cable – the 24hr measure gave a higher expectation relative to the headline.


Consumer expectations of the broadband speed they will experience in reality are high. Overall they expect on average to receive around 91% of a headline speed.

The expectation level appears to vary according to the type of qualifying statement they receive, with the variant “available to at least 50% of users” generating the most conservative expectations when compared to the “typical range of X and Y”.

Whether the measurement time is stated as being over 24hours or peak times did not have a statistically significant impact on the expectation levels.

Consumers think that cable and FTTP will deliver speeds much closer to, or even in excess of, headline figures than they do of ADSL or FTTC.


[1] https://www.asa.org.uk/resource/qualitative-research-for-broadband-speed.html

[2] https://www.ofcom.org.uk/research-and-data/telecoms-research/broadband-research/uk-home-broadband-performance-2016

[3] this was for the combination of ADSL (20mbps) and the up to 50% conditions.

[4] this was for the cable (200mbps) group when given the straight headline speed with no qualifying statement

[5] estimated marginal means

Share this page