By Wole Olukoya, WTB AI Marketing Agency
A Lagos retailer opens WhatsApp on Monday morning to dozens of messages from an ad. The dashboard reports conversations. The sales ledger reports something less exciting. Which number should the owner trust?
Both numbers describe different stages. WhatsApp Business’s own ad reporting includes “conversations started”; that is not a count of paid orders. A useful campaign review therefore follows the customer from the ad to the chat and, where possible, to a confirmed sale. The following five checks help an owner see where that journey breaks.
1. Where did the conversation come from?
Record the campaign, creative and placement associated with each enquiry, using the information available in the ad account and inbox. Compare placements by the quality of enquiries and eventual sales, not simply the lowest cost per conversation. Do not assume Facebook is always worse than Instagram, or that one placement is always accidental-click territory. The answer depends on the offer, audience and evidence from that business. If a placement produces many greetings but few relevant questions, test a narrower placement mix against the current one while keeping the offer and budget comparable.
A simple source column in a shared sheet is enough to begin. Where the source cannot be reliably identified, mark it “unknown” rather than guessing.
2. Did the ad answer the obvious buying questions?
If an ad promises “premium shoes” but conceals the starting price, location, delivery area or lead time, the chat may become an expensive FAQ desk. State the price or honest price range when practical. For services, explain what is included, the starting cost if known, and who the offer is for. Show the product or outcome clearly. This will not eliminate questions; questions are often a good sign. It helps customers decide whether to start a relevant conversation before they tap.
Be careful with exaggerated “from ₦X” claims that apply only to an unavailable option. Clear terms build more trust than a low number that changes as soon as someone asks.
3. What did the customer actually ask?
Classify first messages into a few useful groups: specific product or service enquiry, availability or delivery question, general price check, support request, and unclear or unrelated. A pre-filled “Hi, I’m interested” may be convenient, but it does not by itself show buying intent. The stronger signals are in the next exchange: a requested size, delivery location, event date, quantity or booking slot.
This is not a licence to dismiss people who are still researching. It is a way to make the next reply useful. Someone asking about delivery needs a different answer from someone ready to pay. Write down what a qualified enquiry means for your own offer before comparing campaigns.
4. Did the business reply and keep the conversation moving?
Log when a customer wrote, when the first useful answer arrived, whether a promised quote or catalogue was sent, and who owns the next step. An instant greeting is not a useful response if the customer asked about stock and nobody checked stock. A human or automated first reply can acknowledge the request, collect the essential details and give an honest time for a proper answer. Complex questions, complaints and payment concerns should have a clear human handoff.
The aim is not to message people endlessly. Agree on a reasonable follow-up, respect a customer’s refusal, and keep customer details accessible only to staff who need them. If the team is closed at night, say so and specify when someone will respond. A clear next step is better than silence or a bot pretending to know what it does not.
5. Did an enquiry become a confirmed result?
Choose the outcome that matters: paid order, booked appointment that took place, or another verifiable business result. Record it against the enquiry without treating every chat as a sale. For each campaign, review four separate numbers: spend, conversations started, qualified enquiries, and confirmed results. Then calculate cost per qualified enquiry and cost per confirmed result. A campaign with fewer chats can be the better one if it brings customers who can actually buy.
For example, imagine two ads each receive the same budget. Ad A starts 60 chats and produces two paid orders. Ad B starts 25 chats and produces five. The figures are hypothetical, but they show why “more messages” is not a complete success report. Also note cancellations, refunds and repeat customers where they matter. If sales happen outside WhatsApp, join the records carefully rather than claiming attribution you cannot establish.
How to use the scorecard this week
Take the last seven days of ad-originated chats. Review a manageable sample if volume is high, and note the sample size. Add one row per enquiry with source, whether the ad disclosed key terms, first-message category, time to useful reply, next step and confirmed outcome. Use codes rather than copying private chats into a shared document. At the end of the week, look for the largest drop between stages. If most enquiries are unrelated, revisit the audience, placement and message. If qualified buyers go quiet while waiting for a quote, fix the handoff. If people receive quotes but do not buy, inspect the offer, delivery terms and objections before increasing spend.
Do not change the creative, audience, placement and sales process all at once; you will not know which change helped. Test one meaningful adjustment, compare like with like, and allow enough data for the result to be useful. Small businesses may not have enough orders in a week for a confident verdict, so treat an early pattern as a question to investigate, not a law of marketing.
WhatsApp ads can open a door. The business still has to understand who walked in, answer well, and measure what happened after the chat began.
Disclosure: The author works in marketing. This article is an independent, unpaid contribution and does not recommend a particular product.











