DEV Community

forlazin-debug
forlazin-debug

Posted on

How to Calculate the Payback of Leads from Telegram

It is easy to eyeball leads from Telegram chats: you wrote a couple of replies, and someone seems to have come in. A month later you can't tell whether the time was worth it. Payback is simple to calculate here, but you have to collect the data from day one. You can't reconstruct it from memory later.

For small businesses and independent specialists, lead generation in chats comes down to specific messages from people who are looking for someone to do a job. This first part covers what to record, where to look for requests, and how to tell a real request from ordinary chatter. Replies, a worked calculation, and common mistakes come in the second part.

What to record from day one: five numbers

Lead generation, in simple terms, is finding people who already want to order a service. To know whether it pays off in chats, you need five numbers for a given period. How many suitable requests you saw. How many of them you replied to. How many people responded. How many of those became clients. How much revenue they brought in. Reach, the number of chats, and the number of messages you sent are secondary. You can reply to twenty messages and get no payment, or reply to three and close two orders.

The formula: subtract your costs from the revenue of clients who came from chats, then divide the result by the costs. Costs include tool subscriptions, discounts for new clients, and your time. Count your time in hours at your own rate. If you spend an hour a day reading chats and writing replies, that is a real cost. Without it, the payback figure will look better than it is. Even if you have plenty of spare hours, write them down: later you can decide whether they would be better spent elsewhere.

Set up a table where one row equals one request. The columns:

  • the date and the chat where you saw the request;
  • the gist of the request in one phrase;
  • whether you replied, and at what time;
  • whether the person responded;
  • the payment and its amount.

Mark the client as "came from a chat" in their record right away, on the day of the first message. A month later you won't remember where a given customer came from. Repeat orders and referrals without that mark will end up under "word of mouth" or "unknown", and the channel will look worse than it is.

Where to look for requests: chat types and what to expect

Among the requests the service found, about 45% were written in marketplace seller communities, about 35% in topic-based and city chats, and about 10% in chats for apartment buildings, residential complexes, and neighborhoods. Another 5% or so came from business communities, and about 5% from freelance and contractor-search chats (according to the service's data for the last 30 days). This is a slice of a limited set of niches, with the most data on services for marketplace sellers. So for your field the proportions will be different: it depends on the niche. A renovation contractor or a cleaner will probably find orders in neighborhood and city chats, a psychologist or coach in topic-based ones, and a contractor serving sellers in seller communities.

Requests are widely scattered: according to the service's data for the last 30 days, they turned up in roughly 450 chats. For lead generation for business, that means one thing: covering everything is impossible, and you don't need to. Start with 3 or 4 chats in your niche, not a hundred. The fewer chats you have, the easier it is to see which one brings clients and which one only eats your time while you scroll. After a month, drop the weak chats and add one or two new ones in their place.

Calculate payback separately for each type of chat, or better, for each chat. In one, there are many requests but they are random: people ask for a price and vanish. In another, there are few, but almost every one ends in a payment. Four numbers per chat are enough for comparison:

  • how many suitable requests appeared;
  • how many you replied to;
  • how many payments you received;
  • the average check of those payments.

If a chat produces no payments in a month despite a reasonable number of replies, that is a signal to reconsider either the chat or the wording of your replies.

How to tell a real request from chatter

A real request contains a task and an intention to order something. According to the service's data for the last 30 days, about 40% of the messages found were hot, meaning from people ready to order now. The rest are people just looking around: gathering information, comparing options, or simply sharing a problem. Hot messages usually look like this:

I need someone to steam-clean a sofa this week. Can anyone recommend a person, and roughly how much does it cost?

It has everything you need: a clear task, a deadline, and a request to name a provider. You can reply to this person right away. A message like "does anyone else have a sofa this dirty, what do I do" doesn't count as a request: the person isn't searching for anything, they are complaining. Don't add such messages to your funnel, or your payback will come out too low and you may abandon a channel that works.

Mark two types of messages separately. Requests for recommendations make up about 25% of the messages found: several providers reply to them at once, so it matters who writes first and most specifically. Messages with a deadline ("urgent", "today") make up about 5% (both figures according to the service's data for the last 30 days). They are rarer, but the person decides quickly. In your table, add the tags "recommendation" and "urgent", and after a month compare which type pays back better for you. For some specialists urgent orders work better, for others an ordinary request for a recommendation does, and you can't know in advance.

When and how to reply so you have enough leads to calculate with

Reply speed directly affects the numbers in your table. In a chat, a person's message quickly scrolls up, other providers reply under it, and the first reply with specifics gets more responses. Among the requests the service found, about 45% were written during the day, from 12:00 to 18:00 Moscow time, about 25% in the morning and about 25% in the evening, and about 10% at night, with the peak at around 12:00 (according to the service's data for the last 30 days, and it depends on the niche). So check your chats more often around midday and in the afternoon than in the morning or late in the evening. If you are a specialist who works with clients personally, set reminders for those hours.

A good reply in a chat is short and has three parts:

  • one sentence showing you understood the task in your own words;
  • one example of similar work, without a long portfolio;
  • a question or a proposed next step: clarify a detail, schedule a call, ask for a photo or a link.

Be careful with contacts and website links, and follow the chat's rules: some chats ban people for posting them. It is often safer to write "I can tell you more, I'll message you privately" and wait for a reaction.

Record your reply time and the outcome in the table. After 2 weeks you will see whether fast replies lead to more conversations than delayed ones, and which phrasings get more responses. For example, a reply ending with a question may work better than one with a ready price. Without these records you will be guessing. With them, you can drop what doesn't work.

Example: a month's calculation with real numbers

Take a typical situation. A marketplace seller writes in a seller community chat:

I need someone to figure out my product listings on the marketplace, they barely get any views. Message me if you can help.

A good reply: first clarify what exactly isn't working, because "listings" can mean photos, texts, or the category. Then show one similar case where you found the cause, and offer to look at the listing if they send a link. No promises and no price in the first message: the goal is to start a conversation.

Now the calculation. Suppose that over a month you saw 12 suitable requests, replied to 10, started a conversation in 4 cases, and 2 clients paid. Conversion from reply to client: 2 out of 10. These are numbers for planning, not a promise of results, and another specialist's will differ. Revenue is the sum of the payments from those two clients. Costs are made up of tool subscriptions, discounts for new clients, and time: for example, 10 hours a month spent reading chats and messaging, multiplied by your hourly rate. Payback equals revenue minus costs, divided by costs. A result above zero means the channel returned what you put in. If the result is below zero, look at which step is losing people: the reply, the conversation, or the price.

What not to do when calculating, and what to compare against

Don't compare channels by cost per contact. Cold calls and ads produce many contacts, but most of those people weren't looking for anything. Leads from chats usually come from people who wrote themselves that they need a service, so there are fewer contacts, but the share of payments can be higher. Compare the cost of a paying client: divide all the channel's costs by the number of clients who paid. That lets you compare chats, ads, and calls on the same scale.

Don't draw conclusions from a single week. One or two clients may not show up because of the season, holidays, or because the right requests simply didn't appear. Take at least a month, preferably two, and keep the raw numbers, not just the final ratio. If you had very few clients in a month, that doesn't mean the channel is bad: you may simply not have had enough data to judge. Extend the observation and change one parameter, such as when you check chats or how you word your replies, to see what actually matters.

Don't forget repeat orders and referrals. If a client from a chat came back a couple of months later or brought a friend, mark it in the table and add the revenue to the same channel. The typical mistake cuts both ways:

  • the channel's costs are counted in full, but only the first order's revenue is;
  • revenue from repeat clients is logged as "general sales", so the channel looks unprofitable;
  • or the reverse: all revenue is counted, but the hours spent on replies are not counted at all.

Check your table against these three points before you decide whether to keep looking for clients in chats.

The service is called Leadolovets, with a five-day free trial, no card required.

Top comments (0)