Launching a review programme

Most review programmes fail quietly. The software is installed, a sequence is configured, and six months later the catalogue has forty reviews spread across eight hundred products. Nothing broke; the programme was simply built in an order that could not produce volume.

This is the order that does.

Decide what “working” means first

Write down a number before you start. Two useful ones:

• Coverage — the share of your products carrying at least, say, five reviews. This is what actually moves conversion, because a shopper only sees one product page.
• Response rate — reviews received divided by requests sent. Somewhere between 2% and 8% is normal. Below 2% and something is wrong with deliverability or timing, not with your customers.

Total review count is the vanity metric. A thousand reviews on your ten bestsellers and nothing on the other four hundred products is a programme that looks successful and is not.

1. Import your history before you ask anyone for anything

If you have reviews anywhere — another platform, your ecommerce platform’s native reviews, a spreadsheet — bring them first.

An empty product page is worse than no reviews section at all: the blank space reads as “nobody bought this”. Importing gives every product page something on day one and gives your new requests somewhere to accumulate.

See Importing & exporting reviews. Two things decide whether it works: matching on the product identifier your catalogue actually uses, and mapping the original review dates so five years of history does not all land on today.

2. Fix deliverability before you fix copy

Authenticate a sending subdomain. This single step does more for response rate than every subject line you will ever write, and no amount of copywriting compensates for landing in spam.

Add the DNS records, verify, and set your from-address to that subdomain. See Sending domain & deliverability.

While you are there, set a reply-to that someone reads. Review requests get replies, and they are usually support issues you want to see.

3. Send one email, not four

Start with a single request at 7–14 days after the order ships, and one reminder five to seven days later.

Resist building the four-step sequence you have in mind. Each additional email costs deliverability and produces less than the one before it, and you cannot tell whether step three is worth having until you know what steps one and two do on their own.

The timing question is really “has the customer used the thing yet?” — which depends on what you sell. Consumables and apparel: a week. Something with a setup or a learning curve: three or four.

4. Ask for photos, and pay more for them

Turn photo requests on in the initial step. A product page with five photo reviews is more persuasive than one with thirty text reviews, because photos are hard to fake and show the product in a real room.

If you are going to offer an incentive, tier it — little or nothing for text, meaningfully more for a photo, more again for video. A flat incentive buys you a wall of one-line text reviews, which is the least useful thing you can pay for. See Review incentives.

5. Turn on auto-approval for verified purchases, and only that

Moderating every review by hand does not scale past a few hundred a month, and the queue is where programmes die — not because anyone decided to stop, but because nobody had time on Thursday.

Auto-approve verified purchases. Keep a human in front of anything anonymous or unmatched.

Do not auto-approve by rating. A product page with nothing below four stars is a page nobody believes, and a shopper who notices will discount everything else on it. See Handling negative reviews.

6. Put the evidence where it is seen

Reviews that exist and are not visible are worth nothing.

• A star badge on product cards in collection and search listings. This is the highest-leverage placement in most stores — it is the only evidence a shopper sees before choosing which product to open.
• A star badge under the product title, linked to the review section.
• The review list inline below the description, not inside a tab.

See Showing evidence on product pages.

7. Reply to things

Budget an hour a week. Reply to every review below three stars and to any review that asks a question.

This is the cheapest conversion work available to you. A calm, specific reply under a two-star review does more good than the review does harm, because it demonstrates to every future reader that a person is paying attention.

The first ninety days

• Week 1 — import history, authenticate the domain, install widgets, one sequence step.
• Week 2–3 — first requests go out. Watch response rate, not review count.
• Week 4 — add the reminder step. Turn on verified-purchase auto-approval.
• Month 2 — start replying weekly. Check coverage by product, not in aggregate.
• Month 3 — decide about incentives, based on which products are still bare.

When it is not working

Nothing is arriving at all. Orders are the trigger. Check that orders are syncing and that your qualifying status matches what your platform actually sets.

Requests send, nobody responds. Almost always deliverability. Check the domain is verified, then try the minimal template style.

Reviews arrive only for your bestsellers. Normal, and the thing coverage is for. This is when incentives on the long tail earn their keep.

Everything is five stars. Either you are filtering, or your request only reaches happy customers. Both are worth fixing — an all-five-star page converts worse than a 4.6.