TTC's audience comments at roughly 15 to 40 times the normal Instagram rate. Those comments currently go nowhere. This is a plan to intercept them, open a conversation, and end up with a customer database that Instagram cannot take away.
The whole thing in about a minute.
A normal Instagram account gets 1 to 3 comments for every 100 likes. TTC gets 45. Comment-to-DM automation only works if people comment, and most accounts fail that test before they start. TTC passes it by a wide margin. Nothing about their content has to change for this to work.
Someone comments, the comment sits there, and that is the end of the relationship. Every one of those is a conversation that could have been opened and an email address that could have been captured.
A 500,000-follower competition site running the same system saw 99.8% of delivered messages opened and 94.4% clicked. Email runs at roughly 25% opened and 2 to 3% clicked. Even after platform blocking, the same list produced about fifteen times the clicks email would have. See Proof it works.
That same site's data shows it plainly. Messages sent to people who had just commented delivered at 84.6%. Messages blasted to the whole list delivered at 49.2%, because the platform blocked everything outside the window. This one rule shapes the entire design.
Most brands have no reason for a customer to get back in touch, so the window shuts and stays shut. A comp site has four reasons every week: the draw opens, it closes, it runs, a winner is announced. The whole plan is built on that rhythm rather than on a list.
One of their Reels got 21,700 views and 104 comments. Another got 8,000 views and 115 comments. The smaller one captured more people. Chasing viral reach is the obvious instinct and it is the wrong one.
The automation itself should be off-the-shelf software at about £40 a month. The part worth building properly is the connection that pushes every captured name, email and phone number into TTC's own database. That is the asset. Everything else is rented.
The sample
| Content | Views | Likes | Comments | Comments as % of views | Comments per 100 likes |
|---|---|---|---|---|---|
| Reel 1 | 11,600 | 265 | 105 | 0.91% | 40 |
| Reel 2 | 8,000 | 163 | 68 | 0.85% | 42 |
| Reel 3 | 21,700 | 275 | 104 | 0.48% | 38 |
| Reel 4 | 8,000 | 170 | 115 | 1.44% | 68 |
| Reels combined | 49,300 | 873 | 392 | 0.80% | 45 |
| Photo post 1 | not supplied | 234 | 19 | n/a | 8 |
| Photo post 2 | not supplied | 253 | 67 | n/a | 27 |
Reel 4 is highlighted because it is the outlier. See "the ask, not the reach" below.
Assuming roughly one Reel a day, with photo posts less frequent:
The per-post figures are real, supplied from the account. The follower count is real. The monthly figures are an extrapolation from a four-Reel sample and an assumed daily posting rate, so treat them as an order of magnitude rather than a forecast. Cadence and a 30-day average should be pulled from Instagram Insights before anything is committed to.
The single most important number about TTC is 45 comments per 100 likes. A typical Instagram account runs at 1 to 3. TTC is running at fifteen to forty times that.
The reason is obvious once you see it. Competition content asks people to comment: answer the question, name your colour, tag who you would take with you. The audience has already been trained into the exact behaviour this system depends on.
That matters because comment-to-DM has one hard precondition, and it is not budget or clever copy. It is whether people comment. Most accounts have to change their content to make this work. TTC does not.
| Views | Comments | Capture rate | |
|---|---|---|---|
| Reel 3 | 21,700 | 104 | 0.48% |
| Reel 4 | 8,000 | 115 | 1.44% |
Reel 4 had 2.7 times less reach and captured more people, at three times the rate. The extra 13,700 views on the bigger Reel were almost all passive scrolling. Going viral does not convert proportionally into conversations.
Do not build the strategy around chasing reach. Build it around the ask. Working out what Reel 4 did differently, and then doing it deliberately, is worth more than any amount of extra distribution. It also means the pilot is cheap: no viral hit is required to prove the system works.
28,000 followers, but an average Reel reaches 12,300 of them and overall engagement sits around 1.1%. That is a lot of people not paying attention.
For a competition site there is a straightforward explanation. Followers acquired through giveaways are low-intent by design. Someone follows to enter one draw, does not win, goes quiet, and never gets round to unfollowing. TTC has been accumulating that population for as long as it has been running.
Those people cannot be reached by email or SMS, because TTC never captured their details. Instagram is the only place they still exist. That is what makes this channel worth the effort rather than just being another inbox.
Benchmark account
Palladium Competitions, a 500,000-follower UK competition site, on Facebook Messenger. Data window 24 June to 8 July 2025.
| Flow | Sent | Delivered | Opened | Clicked |
|---|---|---|---|---|
| Broadcast to the whole list Ticket discount offer | 3,405 | 49.2% | 99.8% | 94.4% |
| Triggered by a comment Nightly freebie post | 298 | 84.6% | 100% | 99.6% |
99.8% opened and 94.4% clicked. Email typically runs around 25% opened and 2 to 3% clicked.
Worked through on the 3,405-message broadcast: roughly 1,675 were delivered, and roughly 1,580 of those clicked through to the site. The same 3,405 names contacted by email would have produced somewhere around 100 clicks.
Even with more than half the messages blocked before they arrived, the channel produced roughly fifteen times the clicks that email would have from the same list.
The same account carried a platform warning stating that 100% of recent messages sent after the 24-hour window were blocked to prevent a policy violation. That is why the broadcast delivered at only 49.2%: half the list was out of window and the platform refused to send.
Now compare the second flow. Same account, same fortnight, same platform, delivering at 84.6%. The only difference is that those people had just commented, so their window was open.
Messaging people whose window is open delivers at around 85%. Blasting a stored list delivers at around 49% and gets you a policy warning.
Which is why this plan is built around the draw cycle rather than around a contact list. The cycle keeps giving people a reason to get in touch, which keeps windows open, which keeps delivery high. A plan built on list size instead would decay month after month and nobody would understand why.
That last figure is worth sitting with. Contacts go quiet at a real rate, which is another reason the recurring cycle matters more than the raw list.
The nightly free draw. Their comment-triggered flow was attached to a "tonight's freebie" post. A free draw every night gives people a reason to comment every night, which keeps the messaging window open continuously rather than weekly. It is the cheapest possible engine for this system and TTC do not currently run one.
Themed discount codes rather than discount blasts. Their offer message was framed as "you have been chosen", tied to the theme of the competition, carrying two bundle prices with named codes. It reads as something that happened to you rather than as a promotion sent to everybody.
This is Facebook Messenger, not Instagram. The mechanics are broadly comparable and the 24-hour rule works the same way, but the delivery figures should be treated as a comparable rather than a prediction.
And that account has 500,000 followers against TTC's 28,000. The rates transfer. The volumes do not. At roughly 1% of followers converted into contacts, the equivalent ceiling for TTC is in the hundreds to low thousands, not the tens of thousands. TTC's far better comment ratio should push that higher, but nobody should be promised a five-figure list.
2,940 comments a month is not 2,940 people. On a competition account the same engaged core comments on nearly everything, so a few hundred names will generate most of that volume.
The contact list will build very fast for four to six weeks and then flatten once the active core has been harvested. After that, growth depends on reaching the dormant audience and beyond.
Say this at the start and the slowdown is a predicted milestone. Leave it out and month two is a difficult meeting. Reporting should track unique contacts and delivery rate, never comment volume.
| Trigger | Works? | Notes |
|---|---|---|
| Comment on a post or Reel | Reliable | Keyword or any comment. The workhorse. |
| Comment during a Live | Yes | Fires while the draw is running |
| Story reply | Yes | |
| Story emoji reaction | Yes | Lowest-friction reopener available |
| Story mention (they tag you) | Yes | |
| Shares your post to their story | Yes | High intent, almost nobody uses it |
| DM keyword | Yes | Any inbound message containing a word |
| Conversation Starters | Yes | Tappable menu for people who have never messaged |
| Default reply (no keyword match) | Yes | The FAQ catch-all |
| New follower | Unreliable | See below |
| Likes on posts, Reels or stories | No | A like creates no permission and no reply path |
| Saves, profile views, story views | No |
"Automatically DM every new follower" gets pitched constantly. It exists, but it is a Meta beta with an account-level eligibility lottery and no published criteria. There is no follow webhook in Meta's publicly subscribable fields, which is why it is held together with tape. Even when it works it fires once per follower ever, capped at one per user per week, and ManyChat's own support forum is full of people reporting it silently failing.
Use it if the account happens to have it. Never make it a pillar. The reliable substitute is a pinned welcome post with a comment keyword: new followers land on the profile, see the pin, comment, and enter the flow.
Worth understanding, because it is the most common question. Meta only lets you message someone who has made contact. A comment gives you a public reply path. A message or story reply gives you a direct one. A like gives you neither permission nor anything to reply to.
The useful exception is that a story emoji reaction does fire the trigger, because underneath it actually is a direct message and lands in the inbox. A like on the story itself does not. That distinction matters, because "react with a flame" converts far better than "reply to this story" and does exactly the same job.
Prove it before building anything
Take their next draw-open Reel. Add a keyword call to action. The software replies publicly and sends the entry link by DM. One opt-in button. Capture the email address.
Measure five things: comment volume before and after adding the call to action, delivery rate, open rate, opt-in click rate, and click-through to the site. That replaces every assumption in this document with real numbers, and it costs almost nothing.
Run it on a normal Reel, not a big one. Their own data says the 8,000-view Reel out-captured the 21,700-view one.
Separately in the same fortnight: work out what Reel 4 did differently. A 1.44% capture rate against a 0.48% to 0.91% baseline is a threefold difference, and it is the highest-leverage unknown here. Whatever that Reel did, doing it on purpose beats any automation.
The actual product
Build the five flows in Part six around the rhythm the business already has. The cycle is what makes this work: it hands you a legitimate reason for the customer to restart the conversation every week, which is what keeps delivery near 85% instead of near 49%.
Two additions in this phase. The nightly free draw, copied from the benchmark account, because it turns a weekly reason to make contact into a daily one. And the Instagram Broadcast Channel, which has no 24-hour window and no send limit, making it the only tool that reaches the dormant audience who are not seeing the Reels. Recruiting into it should be an objective of every flow, not an afterthought.
Where the value compounds
Everything above sits on rented ground. The contact list belongs to the software vendor, which in turn depends on Meta. This phase moves the value into TTC's own systems: email and phone into their own database, tied to which draw, which keyword and which surface each person came from.
It is also the point at which genuinely personalised messaging becomes possible, because for the first time there is customer data worth personalising against.
An average Reel reaches 44% of the follower base. Automation harvests whatever the content reaches; it cannot improve the reach itself. Those are two different problems and this plan only solves one of them. Better said plainly than letting the automation get blamed for a distribution issue.
Trigger: comment on a Reel
Prize-reveal Reel with a keyword call to action. Public reply plus a DM carrying the entry link. Opt-in button, then capture the email address. This is the volume flow and the one the pilot tests.
Trigger: story emoji reaction
A story saying "react with a flame for the last-chance link". The reaction is user-initiated, so it legitimately reopens the messaging window, and it is the lowest-friction action available on the platform.
This is the highest-intent moment in the cycle and at present it is only being hit by email, if at all.
Trigger: comment during a Live
TTC already run their draws live on Instagram, so the moment exists and is unused. Comments during the draw trigger a DM confirming entry and teasing the next competition.
It is also the best moment to recruit people into the Broadcast Channel, because the whole audience is present at once.
Trigger: Conversation Starters and default reply
Covers the five questions that arrive every cycle: where are my tickets, when is the draw, did I win, how do I claim, how does the skill question work.
This gives time back immediately and it is the easiest part to justify. It is also the one most likely to be undervalued in the pitch, because it saves cost rather than making sales.
Trigger: share to story
When a winner shares their win to their story, that fires an automatic DM thanking them and asking for a short video testimonial.
For a competition site, winner content is the entire trust engine. Nobody automates the collection of it, and the moment someone has just won is the moment they are most likely to say yes.
People who comment but do not buy get a same-day sequence: a previous winner, ticket count and odds transparency, a closing reminder. It has to be same-day rather than a drip over weeks, because the messaging window does not allow a drip unless an email address was captured. Which is the argument for capturing the email in flow 1.
| Option | Verdict | Why |
|---|---|---|
| Off-the-shelf software (ManyChat) | Phases 1 and 2 | Roughly £15 to £65 a month, and it absorbs every Meta platform change for you |
| Custom data connection | Build this | Small piece of work, and the only part that creates a lasting asset |
| Fully custom integration | No | App review, ongoing maintenance, and you inherit every platform change Meta ships |
| AI-driven conversational agent | Phase 3 at the earliest | Needs customer data to personalise against and volume to justify it. Neither exists yet. |
At 28,000 followers and roughly 700 commenters a week, a custom Meta integration cannot pay for itself. Off-the-shelf software does the flows perfectly well and someone else maintains it.
What is worth building properly is the connection that pushes every captured contact into TTC's own database: email, phone, which draw, which keyword, which surface. It is a small job, it is cheap, and it is the difference between building an asset and renting a chatbot.
Any freelancer can set up flows. Almost none of them set up the data layer, because it is the part that lives outside the tool. If the engagement ends, TTC keeps a database either way. That is the argument for doing it properly rather than cheaply.
Does not exist here. This was a Messenger-only advertising format and Meta has been retiring it. There is no Instagram equivalent. Any plan quoting a fraction of a penny per message is describing a channel that is not available.
Not available in the UK. Comment-to-DM is in beta and restricted to Vietnam, Thailand and Indonesia. That is TikTok's own API restriction rather than any one vendor's, so no tool routes around it. TikTok stays a traffic channel. Worth revisiting in six months.
Not possible. A like creates no messaging permission. This will not change.
Too unreliable. Meta beta, eligibility lottery, silent failures. Use a pinned welcome post instead.
Delivers at about 49% and triggers a policy warning. The benchmark account's own data shows it. Broadcasts have a place, but the system has to be built on the cycle, not on the list.
Not credible at this scale. Plans of this type are usually written for accounts with hundreds of thousands of followers and then recycled. The numbers in this document are TTC's own, and the benchmark is labelled as a benchmark.
Run the phase 1 pilot on the next draw. One Reel, one keyword, one opt-in button, five numbers measured. It costs almost nothing, it needs no commitment beyond a fortnight, and it turns every estimate in this document into a fact.