MarketFlow

Marketing

A Practical Guide to Building Better Marketing Campaigns

Most campaigns are built in the wrong order: the message first, the audience second, and the measure of success somewhere after it has already sent. Start at the end Write down the number that will tell you whether this worked before you write a word of the copy. Replies, orders, bookings. If you cannot name it, the campaign is an announcement, which is a legitimate thing to send and a different thing to judge. Then the audience The audience decides the message far more than the message decides the audience. A campaign to everybody has to be vague enough to be true for everybody, which is why broad sends underperform so reliably. Then the template Reusable templates are not just a time saving. They are what makes results comparable - when the structure is held constant, the difference in performance is attributable to the thing you actually changed. Review within the week What did it produce against the number you named first? Which segment carried the result, and which one diluted it? What would you change, in one sentence? Is that change worth making before the next send, or is it noise? A campaign nobody reviewed is a campaign nobody learned from, and the second one will cost exactly as much as the first.

MarketFlow2 min read

Most campaigns are built in the wrong order: the message first, the audience second, and the measure of success somewhere after it has already sent.

Start at the end

Write down the number that will tell you whether this worked before you write a word of the copy. Replies, orders, bookings. If you cannot name it, the campaign is an announcement, which is a legitimate thing to send and a different thing to judge.

Then the audience

The audience decides the message far more than the message decides the audience. A campaign to everybody has to be vague enough to be true for everybody, which is why broad sends underperform so reliably.

Then the template

Reusable templates are not just a time saving. They are what makes results comparable - when the structure is held constant, the difference in performance is attributable to the thing you actually changed.

Review within the week

  • What did it produce against the number you named first?
  • Which segment carried the result, and which one diluted it?
  • What would you change, in one sentence?
  • Is that change worth making before the next send, or is it noise?

A campaign nobody reviewed is a campaign nobody learned from, and the second one will cost exactly as much as the first.

Share this article

Keep reading

Related articles

More practical insights on Marketing and the topics around it, from the MarketFlow blog.

View all articles

WhatsApp Automation

How to Turn WhatsApp Conversations Into Automated Customer Journeys

Most teams start on WhatsApp the same way: one shared phone, a handful of saved replies, and somebody who remembers to follow up. It works until it does not. The moment volume rises, the follow-ups are the first thing to go - not because anyone decided to stop, but because nobody owns the reminder. A customer journey fixes that by writing the follow-up down once, as a workflow, instead of relying on a person to remember it every time. Start with the trigger, not the message A journey begins with something happening: an inbound message, a form submission, a lead moving stage, an order placed. Pick the trigger first and the rest of the journey writes itself - you are no longer asking what to send, you are asking what should happen next. Inbound message from a new number - send a greeting, capture the contact, tag the enquiry. Lead reaches Qualified - wait a day, then send the pricing template. Order placed - confirm it, then check in a week later. No reply after three days - one follow-up, then stop. Let the wait do the work The step teams skip most often is the wait. A sequence with no delays fires everything within a minute of the trigger and reads as a machine. Delays are what make an automated journey feel like a team that is paying attention: a message the next morning lands very differently to one that arrives four seconds after somebody says hello. Keep the handover open Automation should never be a wall. Every journey needs a point where a human takes the conversation back, and the inbox is where that happens - an agent picks up the thread with the full history already attached, and the workflow stops rather than talking over them. Build one journey, watch it for a fortnight, and only then build the second. The teams that get the most out of automation are rarely the ones with the most workflows.

2 min read
Read article

Customer Management

Why Customer Segmentation Matters for Modern Marketing

Segmentation has a reputation as an advanced feature. It is closer to the opposite: it is the thing that makes everything else simpler, because a smaller audience is easier to write for and much easier to learn from. Segment on behaviour, not demographics What somebody did is a far better predictor than who they are. Ordered twice in ninety days, opened the last three campaigns, asked about pricing and never replied - each of those is a group with an obvious next message. Start with three Customers who bought recently - the group most likely to buy again. Leads who engaged and went quiet - the group most worth recovering. Contacts who have never engaged - the group to stop paying to reach. A segment is a hypothesis The value is not the list, it is what the list teaches you. If a segment never performs differently to the whole audience, it is not a segment - and deleting it is as useful as creating it was.

1 min read
Read article

Commerce

Turning Customer Conversations Into Measurable Revenue

Conversational selling has an old reporting problem: the conversation lives in a messaging tool and the order lives in a store, so the link between them is somebody's recollection. Attach the order to the contact Once an order belongs to a contact rather than to an anonymous checkout, every question downstream becomes answerable. Which campaign preceded it. Which conversation it came out of. Whether this customer has bought before, and what happened last time. What becomes measurable Revenue by channel, rather than opens by channel. The conversion rate of a conversation, not just of a campaign. Repeat rate per segment, which is the number that decides what to spend. The real cost of a slow reply, in orders rather than in minutes. Be honest about attribution No model is exact, and a customer who saw a post, read an email and then messaged you is not one channel's win. Attribution is useful as a direction, not as a verdict - the moment it becomes a scoreboard, teams start optimising for the model rather than for the customer.

1 min read
Read article

Analytics

What to Track When Measuring Marketing Performance

Marketing analytics fails in two directions. Either nothing is tracked, or everything is - and a dashboard with forty numbers on it is the same as no dashboard, because nobody can tell which one moved for a reason. The short list Reply rate, per channel - the clearest signal that a message landed. Conversion rate from conversation to order. Revenue per campaign, not per send. Repeat purchase rate, which is the slowest number to move and the most worth moving. Time to first response, because it predicts most of the above. What to ignore Impressions, follower counts and open rates are diagnostics, not outcomes. They are useful for explaining why a number moved and dangerous as the number itself - open rate in particular has been unreliable since inbox providers began pre-fetching images. Pick a cadence and hold it Weekly for campaign activity, monthly for revenue and repeat rate. Checking a slow metric daily produces noise, and reacting to that noise is how teams end up rewriting a strategy that was working.

1 min read
Read article

WhatsApp Automation

Writing WhatsApp Templates That Get Approved the First Time

A rejected template is rarely rejected for its wording. It is rejected because it is the wrong category, because a variable is doing something a variable should not, or because it reads as an advert sent to somebody who did not ask for one. Get the category right first Utility, marketing and authentication are not labels you pick to suit the approval odds. They describe what the message does, and a marketing message submitted as a utility one is the single most common rejection. Write the variable, not around it Give every placeholder a real example when you submit it. Never start or end a message with a variable. Do not use a variable to smuggle in the part of the message that would have failed review. Keep the fixed text meaningful on its own - if the sentence only makes sense once filled in, it will not pass. Say who you are The recipient may not have your number saved. A template that opens by naming the business it comes from is both better received and more likely to be approved, and it costs one line.

1 min read
Read article

WhatsApp Automation

Running a Shared Inbox Without Two People Answering the Same Chat

The failure mode of a shared inbox is not messages going unanswered. It is two people answering the same one differently, four minutes apart, in front of the customer. Ownership beats etiquette Every open conversation should have exactly one name on it. Teams that rely on a convention - whoever saw it first - discover that the convention breaks at precisely the times it matters, which is when the inbox is busy. Three conventions worth writing down Assign before you reply, not after. Close a conversation when it is done, so the open count means something. Hand over in the thread, not in a side channel - the next person needs the context, not the summary. Let the automation take the easy half Acknowledgements, opening hours, order status. Anything a workflow can answer correctly every time should not be arriving in a human queue at all, and removing it is usually what makes the rest of the inbox manageable.

1 min read
Read article

Ready to start?

Turn every chat into a sale

Bring your chats, campaigns, automation, products and sales together in one place

Get Started FreeSee All Features