Reading time: 34 minutes

Tones, Questions & Responses: We Analyzed 400+ Political P2P Texts to See How They Begin

Published: Nov 25, 2021

“Hi, we must support Timothy Gregor in this election. Our future depends on his win. Do consider voting for him.”

How likely do you think this message will get responses? Let’s break down the obvious issues:

❌ There is no personalized or even impersonal address to the contact.

❌ The agent did not introduce themselves.

❌ The message is vague— No specific cause, campaign, issue.

❌ There is no CTA to act on the message.

This is a hypothetical example, of course. But in addition to the issues mentioned above, some other factors can also impact response rates; factors such as 

  • Whether the contact opted in for such texts,
  • If they had communicated with the campaign previously, or 
  • How inclined the contact is towards the candidate.

Political campaigns initiate conversations with electorates to nurture them to support the candidate throughout the race (and finally in the ballots). Texting has an essential role in these conversations. Specifically, peer-to-peer texting (or P2P texting) is instrumental in conducting personal conversations at scale and helping campaigns get closer to their supporters. 

The rules of a normal conversation apply here too. A great start can ensure a longer-lasting, more engaging, and potentially fruitful discussion. So, do political text bankers follow these rules?

We studied initial texts from nearly 450 campaigns to understand the language, tone, and semantics. Here’s what we discovered:


  • Message introduction
  • 45.2% of the messages began with a Hi/Hello/Hey {first_name}– indicating the popularity of personalization tools among political campaigns.
  • Next in line (but with a considerable margin) were cold opens. 17.4% of the messages began head-on, without any greetings, introduction, or descriptions. Of the cold opens, 1/4th were mobilization messages using the absence of greetings to convey urgency.
  • Message body
  • 45.66% of texts were between 161 and 306 characters (sent as two messages). The average response rate was the highest for this character length (12.25%), although we can’t be too sure that the length caused people to reply. 
  • 23.3% of the texts were campaigning for the candidate without an explicit CTA. 19.6% had a definite and prominent CTA.
  • Only 3.2% of the messages mobilized people towards ballot voting (the smallest share).
  • We observed four distinct tones in the messages– Neutral (38.13%), emotionally appealing (24.66%), formal (21.23%), and friendly (15.98%).
  • 48% of the campaigns used at least one non-rhetorical question. 47.94% used at least one link. Campaigns with one question got an average of 6.92% higher response rate. Those with two questions got a 0.86% higher response rate.
  • Only 0.45% of messages used either emojis or media.
  • Message conclusion
  • 38% of all messages ended with a question. Most of the questions were either a confirmation of support or a response prompt.

Among those that didn’t end with a question, 44.78% concluded with a link, and 19.03% with no distinct sign-off. 


  • We took out all test messages. These are messages that explicitly mention their testing status. Typically, at most, 10 messages per campaign are sent as a test while the initial text to contacts goes out in bulk. In some cases, the total sent messages were less than 10, but since they did not indicate that it was a test, we included them in the analysis.
  • While several factors serve as the reason a contact replies, one of the most important ones is the prompt– did the text ask or prompt for replies? This is why we have calculated the response rates against those texts that had such prompts. Additionally, we have calculated average response rates for factors such as the tone of the message. Pinpointing the exact reason a person replied is beyond the scope of this study. The average rates only give an overview and estimated likelihood for replies.
  • Nearly half the messages had at least one link in them. Even with a CTA to reply, these links could have prompted people to click on them rather than reply. So, a non-response in such cases does not necessarily mean non-engagement.

We’ve divided the message into three parts:

  1. Introduction
  2. Messaging (main body)
  3. Conclusion

Let’s dive deep.  

Message introduction


A text message is rarely as elaborate as a letter, and it can get difficult to distinguish between its introduction, message body, and conclusion. Often, two or more overlap or an element is missing. However, how a message starts can be an important factor in determining whether your contact reads it thoroughly. 

Unlike an email or social media message, texts are restricted in options to highlight the main parts (all caps, emojis, or excessive punctuations are some examples). So, a good start is essential.

What we analyzed:

  1. Nature of greetings 

Some greetings are very specific and close-ended. E.g., Hello {first_name}, with the only difference being in the position or absence of a comma/exclamation mark. Some are more subjective and diverse—especially the cold opens, disclaimers, and headline starters. 

We divided the greetings into 10 categories for the overview— a mix of objective and subjective openings.

Objective opens
Hello {first_name}. 
Includes a “hello” greeting followed by the merge tag “first_name” (of the contact). 
– Hi {first_name}
– Hello {first_name}
– Hey {first_name}
Hello {agent intro}Includes a “hello” greeting followed by the agent’s introduction. – Hi (agent intro)
– Hello (agent intro)
– Hey (agent intro)
Agent intro can include their name, designation, or both.
Greetings (Contact)

This type of greeting does not use personalization tools like a merge tag but addresses contacts as part of a community.
The “greetings” may include:
– Hi
– Hello
– Hey
– Good morning/afternoon
– Dear
– Respected
– Greetings

The “address” may include
– Voter
– Supporter
– Friend
– Neighbour
– Team
– Democrats/Republicans
Greetings {first_name}

This type of greeting uses merge tags and greetings other than Hi/Hello/Hey.
– Good morning/afternoon
– Dear
– Respected
– Greetings

Messages that start directly with the {first_name} merge tag without any greetings preceding it.
{first_name} may or may not be followed by punctuation.
These greetings are not followed either by a merge tag or by a community identification. 
– Hi
– Hello
– Hey
– Good morning/afternoon
– Dear
– Respected
– Greetings
Subjective opens
Cold open
Cold opens are when the message gets straight to the point without greetings, introductions, or disclaimers.
Subjective and varied in nature. An example would be, “We need smart growth in our cities.”
Message intro/descriptions
Messages that start with a 1-3 word introduction, warnings, or descriptions of what’s coming.
– Reminder
– Alert
– Warning
– Quick Reminder
Agent introduction
When a text begins directly with an introduction of the text banker without any greetings.
– I am/This is (agent name)
– I am/This is (agent designation)
A type of a cold open, texts sometimes start with news or headlines that draw attention to the following message (body)
Subjective and varied in nature. An example would be, “Today is election day!”

45.2% of the messages began with a Hi/Hello/Hey greeting followed by the merge tag {first_name}. Breaking down this introduction further, 36% messages began with a “Hi {first_name}, 2.5% with a “Hello {first_name}” and 6.6% with a “Hey {first_name}.

It was no surprise that the category with a short greeting and a merge tag took first place here. Merge tags are simple but highly effective ways of showing a person that the message was intended for them. Today most people akin to technology expect you to personalize and tailor messages for them. Getting a person to read their name in the first line is a classic way of maintaining engagement and showing them that this is not spam bulk text. 

First names are just the tip of the iceberg of default and custom merge tags you can use on CallHub. By default (in addition to the first name), we have the following merge tags in our system:

  • {last_name}
  • {mobile_number}
  • {phone_number}
  • {email}
  • {country}
  • {city}

Additionally, you can import other custom tags from your CRM and use them in your P2P, text broadcast, and calling campaigns. In the texts we studied, we came across the following merge tags used by political campaigns:

  • {Agent_name}
  • {Agent_firstname}
  • {link}
  • {surveylinkwithtag}

Learn more about these personalization tools here.

The next most popularly used introduction is with a cold open, at 17.4%. Following that was a “hello” greeting + agent introduction. An agent introduction is a great way of adding a personal touch at the beginning of your messages even without addressing the recipient by name or group. It shows that a person is texting them and eliminates the doubt or turn-off about bot messages.

Together, these three types of opens constitute 73.8% of the opens. 

Both, the “hello {first_name}” and “hello (agent intro) are considerably cold-ended introductions. But not cold opens. So let’s analyze this category further.

  1. Categories of the cold opens

Furthermore, we picked the cold opens to analyze their nature and categorize them. This was a deliberate choice considering all messages that started with a cold open fit into eight categories. While the actuals of news, agent introductions, and message descriptions are widely varied and distinct, the category name describes their overall nature. Not so for the cold opens. We identified the following sub-categories under cold opens:

  • Candidate introduction: Name or description of the candidate.
  • Campaign achievement: Sponsorship/endorsement/impact of the campaign or party.
  • Mobilization: Getting the contact to take action (e.g., vote, register, support, etc.)
  • Voter identification: Understanding how likely the contact is to support the candidate.
  • Voter persuasion: Convincing the candidate to support or vote for the candidate or not support the opposition.
  • Opposition attack: Negative campaigning against the opposition.
  • Campaign/Cause nature: Describing the nature, aim, or work of the campaign, party, or a specific cause.
  • Invitation: Inviting the contact to an event/rally.

Exactly a quarter of the cold opens were mobilization prompts such as “Vote on” or “Join us” or “Complete __ before….” When the motive is to get a contact to take action, such opens can work well with people who like short, quick reads. They don’t spend much time reading the entire message, and driving the message straight to the point can convey urgency.

The next most used category of the cold opens was that of candidate introductions. In most cases, these openings began with the candidate’s name but at times also with a recognizable introduction (e.g., their place of residence). Such opens connect your message directly to the campaign or candidate. The reader need not spend too much time understanding where it came from. It works best with confirmed supporters but least with definite and strong opposition supporters.

Top 5 P2P initial text openings (most used first)Bottom 5 P2P initial text openings (least used first)
Hello {first_name} 45.21%Message introduction/descriptions- 0.91%
Cold open- 17.35%Greetings {first_name}- 2.05%
Hello (agent intro)- 11.19%Greetings (Contact)- 2.51%
{first_name}- 8.45%Greetings- 2.74%
News/headline- 5.71%Agent introduction- 3.88%

Our tips for a P2P initial text introduction:

  • Use personalization tools wherever appropriate to form a connection with your reader. First or last names, city of residence, or custom tags that can form an instant relation (e.g., school name, income group, or ethnicity) can make your message relatable to the audience.
  • Add some form of greeting or introduction in the initial part of the text.
  • If going with a cold open, highlight the main part of the message with capital letters or emojis.

Message body

p2p initial text message insights message body characteristics

The main part of the message– the body. Here, political campaigns make a case for themselves and either prompt action or leave the reader with some information. The length of the messages we studied ranged from 30 characters to 930, and naturally, the amount of information, requests, or statements made in each differed considerably. Some messages had no definite introduction or conclusion, so the entire message had to be considered as the body (paraphrased example: Can we count on your support? Yes/No).

In any case, we dissected the message body and studied its tone, nature, and use of communication devices in each.

Here’s what we analyzed for the message body:

  1. Length (of the entire message)
  2. Nature
  3. Tone
  4. Use of questions, links, emojis, and/or media.

Here’s what we found:

Length of the messages

We divided the word count into six categories, starting at <160 and ending at 918+. These categories correspond with the ASCII standards of word count per text. We have also plotted the average response rate for each type, although there is no definite causation between the two. We can only intuitively judge why the trends are as they appear to be (too long a message may drag and put off a reader. A very short message may not compel them to reply).

Here are the standard ASCII character limits:

Text gets sent as/cost forCharacter limit
1 message160
2 messages306
3 messages459
4 messages612
5 messages765
6 messages918

P.S: If you use special characters (‘: @ # % !), the system automatically considers it UNICODE (instead of ASCII). The character limit for a Unicode message is 70 for the first text and 67 for all subsequent ones. 


For this study, we have used only the ASCII standards to mark the length of the messages. Although multiple campaigns used special characters, the difference between a 70-character word limit and a 160-character word limit determines the cost per text. Thus, it is irrelevant to the recipient.

45.66% of the messages were between 161 and 306 characters, while 31.74% were less than 160 characters. It shows that while many campaigns don’t worry about the message’s length until ~300 characters, they become wary beyond that. And for a good reason too. Too long a message can be off-putting.

However, we also noticed another possible issue. And that is the inclusion of long links in the text messages. About 35% of the texts included at least one long link. Although great as branded devices to compel a user to click, these long links consume many characters in a text. They can stretch your text beyond the character limit. At the same time, non-branded short links aren’t as compelling as branded ones.

Here’s the distribution:


Our tip: We at CallHub noticed this issue with many text campaigns and so have launched our in-house link shortener and tracker. With this feature, you can:

  • Shorten a link 
  • Brand it with your campaign/candidate, or use case name (or anything else, really).
  • Personalize it with a contact’s name (using merge tags).
  • Track total and unique clicks as well as who clicked on the links.

It’s an exciting turn for P2P message tracking, really. If your P2P text contains a link and you’re not getting expected response rates, this link tracker can tell you if people have jumped to your website instead. Read more about the link shortener+tracker here: Measure engagement on text messages.

Nature of message 

We categorized the nature of the messages as CTA, request, negative campaign, informative, transactional, etc.)- Of course, in a majority of cases, the nature overlapped. We consider the most powerful, highlighted, or weighted nature.

CTAA call-to-action to join, support, sign, call or donate.
Negative campaignAn attack or negative remarks on the opposition party or candidate.
InformativeSharing information about the elections, party, policies, etc.
TransactionalSending links or asking for information. Or confirming registration or another action.
Mail-in ballot mobilizationNudging the audience to sign for or send their mail-in ballots.
Voter identificationUnderstanding the supporter level of a person.
Voter persuasionMaking a case for and urging a person to support the candidate.
CampaigningShowing the impact, policy benefits, or making a case for a candidate without explicitly asking for support.
NeutralAgent introduction or a tone/support-neutral message.
GOTVNudging people to go out and vote on election day.

Nearly a quarter of the messages (23.3%) were aimed at campaigning and starting a conversation on the same lines. Next in line, at 19.6% were a call to action for the supporters followed closely by “informative” texts at 18%. Surprisingly, voter persuasion (5%) and GOTV messages (5.9%)– two typically advocated use cases for P2P texting- together accounted for only 10.9% of the total messages.

The tone of the message

We identified four distinct tones: Friendly, Formal, Neutral, and Affective (emotionally appealing).

  • A friendly tone marks itself either with words and language (e.g., just texting to remind you or I want to talk to you about…) or with the use of punctuations like exclamation after {first_name} or the use of emojis. 
  • A formal tone is directed towards a person but uses professional language. 
  • A neutral tone is a more business-like approach and is typically used for transactional purposes. 
  • An emotionally appealing tone is self-explanatory.

Over 38% of the messages used a neutral tone– neither too friendly nor too formal. We’ve plotted the average response rates against each tone, which we will see shortly. It is interesting to note how hundreds of campaigns take a more diplomatic approach to their initial texts. We would have recommended a more friendly approach– the least used tone.

21.23% of the campaigns started with a formal tone, while less than 16% began with a friendly tone. This choice rests entirely with the campaign since the tone, message, and image that the candidate portrays must also reflect in your text campaigns. 

Here are the average response rates for the message tones:


Both formal and friendly tones scored an average of ~11% response rate— with the difference only in decimal points. Affective and neutral tones got 6.15% and 7.55% of response rates, respectively. We suggest you run A/B tests on your contacts, using different tones for the same message (distribute the messages over a few days’ span to avoid irking contacts). Calculate the number and types of responses you get for each and go with the tone/s that give you the best results.

Use of questions, emojis, and media

We have reserved a section exclusively for questions asked in the concluding line, but in this section, we look at questions anywhere* in the text (including the conclusion). Remember, these are all non-rhetorical questions. 

We also look at the use of emojis and media in the initial P2P text (spoiler alert: rarely are they used).

Let’s start with the questions.

Nearly 48% of the messages included at least one question in the text. The distribution is as follows: 


Our tip: Non-rhetorical questions are a good prompt for a response. Unless you ask a person a question or prompt them to reply, they rarely would. How would you otherwise take the conversation forward? Of course, if your intention is for people to click on a link or call their representative, you would rather have them take that action than reply to you. However, if you aim to engage them in conversation, we suggest you add questions and prompts for the same (but add one question per message for the best result).

Let’s look at the average response rate Vs. the number of questions asked in the initial text:


Only 0.45% of the texts used media (graphics and images), and just as many used emojis. However, these were all different campaigns, i.e., the campaign using emojis did not also use the media. Two emojis were used per text, and they were: 🤭, 🚨 and 😆. 

Our tip: Using communication devices like emojis can add color to your text and attract attention to that part of the text. Of course, some cellular phones are not emoji-friendly. However, if your key demographic is below 50 years of age, you can rest assured that they will most likely use a smartphone.

Message conclusion


A conclusion highlighted with spacing from the main body, the difference in the color or length of line (e.g., an underline for a link or a sign-off with a name) is sure to attract attention. Additionally, if a person is looking for who sent the message, they are likely to scan the message’s bottom deliberately. 

The message conclusion is thus an essential part of P2P texting. It must stand out from the rest of the text and include your most important element(s). These could be your candidate’s name, a donation link, or a response prompt.

So, how did the political campaigns we studied end their texts? 

What we analyzed:

  • Number of messages ending with a question and their nature.
  • Nature of sign-off.

Here’s what we found:

Messages ending with a question

We’ve seen before that almost 48% of the texts included at least one question. 38.81% of the total had a question at the end (i.e., ~9% had a question in the message body but not in conclusion).

We identified five categories of questions that the texts ended with

  1. Support confirmation: Questions like “can we count on your vote” or “Do you support ___?”
  2. Response prompt: Messages ending with prompts like “Reply Yes” or “Y/N?” or “Are you interested in ___?”
  3. Need help?: Some texts ended with asking the audience if they needed help or if the campaign agents could solve their doubts. We have categorized them under “Need help?”
  4. Contact request: One of the best practices before calling your contacts is to confirm their availability. Several campaigns did just that with their P2P texting campaign, and we categorized such endings under “contact request.”
  5. P2P request: A handful of the campaigns asked contacts to bring their friends and family to support their candidate. These concluding questions go under the category “P2P request”.

Let’s look at the share each question category took:


Support confirmations and response prompts constituted 93% of the total concluding questions asked (47.1% and 45.9%, respectively). Both are great ways to identify supporters and get them involved in a conversation. 

A simple but direct question is more likely to get responses than one which requires a lot of thinking and typing on the contact’s side. Once the campaign has looped them in a conversation, they can go on to ask more open-ended or detailed questions.

Next in line was a “need help?” question at 5.3% of the share, followed by a contact request (1.2%) and a P2P request (0.6%).

Here’s how the average response rates scored for each type of question:


Our tip: Try to provide options for responses rather than replying as an option to kickstart your P2P texting campaign. For instance, a support confirmation asking how likely the contact is to vote for candidate A on a scale of 5 might still get a “1– least likely”. But a negative response for “Can you bring friends?” will probably result in no response. Instead, adding options like “Reply Y/N” or “How many friends can you bring? A- 0, B- 1-3…” can give a higher likelihood of replies.

Nature of sign-off

How did messages sign off when they didn’t end with a question? According to our study, with a

  • Link
  • CTA
  • Agent name
  • Candidate name/identification
  • Rhetorical question
  • Greeting
  • “Paid by” information
  • Hashtag(s)
  • Opt-out option
  • Or with no distinct sign-off (categorized as “None”).

Here’s the breakdown:


Most messages that did not end with a question ended with a link (44.78%). 19.03% ended with no sign-off at all, while 13.81% ended with a CTA asking for a vote or support. We have broken down the CTA conclusion (below). 0.37% of the texts ended with a rhetorical question (the smallest share), while 0.75% of messages ended with an agent name or an opt-out option.  

Here’s the breakdown of CTA sign-offs:


43% of the concluding CTAs were an appeal to vote for their candidate. 29.7% of conclusions asked their contacts to call their representative or officers to address or raise an issue. 

The smallest share was taken by donation or reply CTAs (at 2.7% each).

Our tip: Segment lists with their history of support and go for different CTAs accordingly. Such political microtargeting will ensure you get the most out of your campaigns. For instance, if a supporter had donated to your campaign previously, they are likely to vote for you anyway. So, target them for fundraising appeals while reserving the support/reply CTAs for less engaged contacts.

To know more about political microtargeting, read Beginner’s Guide to Micro-targeting for Political Campaigns.


CallHub strives for excellence, whether in the texting and calling tools we provide or the best practices we suggest to our readers. As part of that endeavor, we wanted to study the trends of campaigns and offer unique, accurate insights to political campaigns, nonprofits, and businesses. 

This article, in particular, looked at ~450 campaigns that initiated over 900,000 messages. Of course, no one can conclude any trend as the absolute truth. But through rigorous research and study, we can project outcomes to help your P2P texting campaigns perform better.
If you have specific topics about calling or texting that we can research, mail us at [email protected] with the subject line: CallHub Research Ideas.

Feature Image Source: Canva.


Who should win the 2024 elections – Trump or Biden?

"I'm rooting for Trump. Biden's last term showed how he can send America..."

Join the discussion